JP7509960B2 - Mild cognitive impairment assessment system - Google Patents

Description

The present disclosure relates to a mild cognitive impairment assessment system. The present disclosure also relates to a sensor device, a cerebral glucose metabolic rate estimation system, a hippocampal volume estimation system, a gray matter volume estimation system, an amyloid beta accumulation estimation system, and an MMSE score estimation system that can be used in the mild cognitive impairment assessment system. Furthermore, the present disclosure also relates to a mild cognitive impairment prediction system, a mild cognitive impairment prevention system, and a mild cognitive impairment simulation system. Furthermore, the present disclosure also relates to a computer program for executing the mild cognitive impairment assessment system, the mild cognitive impairment prediction system, and the mild cognitive impairment prevention system.

Treatment for dementia varies depending on the cause and progression of the disease. For this reason, it is necessary to accurately diagnose the cause and progression of dementia. In general, the cause and progression of dementia are diagnosed comprehensively based on information such as cerebral glucose metabolic rate, hippocampal volume, gray matter volume, amount of amyloid beta accumulation, and MMSE (Mini Mental State Examination) score. However, this information requires a long measurement time and is expensive to measure. For this reason, methods that can easily and accurately determine the cause and progression of dementia are being investigated.

For example, Patent Document 1 discloses a system that generates sleep data from biological data of a subject while they are sleeping, and compares the sleep data with assessment data for the three types of dementia, Alzheimer's disease, Lewy body dementia, and frontotemporal dementia, to assess the risk of the three types of dementia.

Patent No. 6411111

Mild cognitive impairment (MCI), a precursor to dementia, can be prevented from developing or improved by taking appropriate measures. For this reason, it would be desirable to detect MCI early. It would also be even more desirable to predict the risk of developing MCI in the future and prevent the onset of MCI.

The technology disclosed herein has been developed in view of the above problems. One of the objectives of the present disclosure is to provide a novel mild cognitive impairment assessment system that can easily determine whether or not a subject has MCI without directly measuring information on the subject, such as cerebral glucose metabolic rate, hippocampal volume, gray matter volume, amyloid beta accumulation, and MMSE score. Another objective of the present disclosure is to provide a mild cognitive impairment assessment prediction system that can easily predict the occurrence of MCI, a mild cognitive impairment assessment prevention system that can prevent the occurrence of MCI, and a mild cognitive impairment simulation system that can simulate the probability of occurrence of MCI. Another objective of the present disclosure is to provide a sensor device for acquiring data useful in determining the occurrence of MCI, and an analogy system that can easily infer information (cerebral glucose metabolic rate, hippocampal volume, gray matter volume, amyloid beta accumulation, MMSE score) that can be used for the determination of mild cognitive impairment and the diagnosis of dementia. Furthermore, one of the objectives of the present disclosure is to provide computer programs for executing a mild cognitive impairment assessment system, a mild cognitive impairment prediction system, and a mild cognitive impairment prevention system.

In order to achieve the above object, the present disclosure provides the following techniques.
A mild cognitive impairment assessment system included in one embodiment of the technology disclosed herein includes a data input unit that inputs attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
A judgment model storage unit stores a first judgment model for determining whether a subject corresponds to a healthy subject or a person with mild cognitive impairment, the first judgment model being created using attribute data of a third party including at least one of a healthy subject and a person with mild cognitive impairment and factor information influencing the judgment of mild cognitive impairment obtained from the third party; and
The device has an MCI judgment information calculation unit that calculates judgment information capable of determining whether the subject is a person with mild cognitive impairment based on the subject's attribute data, the sensor data, and the first judgment model.

A cerebral glucose metabolic rate estimation system included in one embodiment of the technology disclosed herein includes a data input unit that inputs attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
a cerebral glucose metabolic rate estimation model storage unit that stores a cerebral glucose metabolic rate estimation model obtained by a model created by an analysis process including at least one of regression analysis and machine learning, with attribute data of a third party and sensor data acquired from a sensor device worn by the third party as explanatory variables, and with the cerebral glucose metabolic rate acquired from the third party as a target variable;
The apparatus further includes a cerebral glucose metabolic rate estimation unit that estimates the cerebral glucose metabolic rate of the subject based on the attribute data of the subject, the sensor data, and the cerebral glucose metabolic rate estimation model.

A hippocampal volume estimation system included in one embodiment of the technology disclosed herein includes a data input unit that inputs attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
a hippocampal volume estimation model storage unit that stores a hippocampal volume estimation model obtained by a model created by an analysis process including at least one of regression analysis and machine learning, with attribute data of a third party and sensor data acquired from a sensor device worn by the third party as explanatory variables, and with a hippocampal volume acquired from the third party as a target variable;
The device further comprises a hippocampal volume estimation unit that estimates the hippocampal volume of the subject based on the attribute data of the subject, the sensor data, and the hippocampal volume estimation model.

In addition, a gray matter volume estimation system included in one embodiment of the technology disclosed herein includes a data input unit that inputs attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
a gray matter volume estimation model storage unit that stores a gray matter volume estimation model obtained by a model created by an analysis process including at least one of regression analysis and machine learning, with attribute data of a third party and sensor data acquired from a sensor device worn by the third party as explanatory variables, and with the gray matter volume acquired from the third party as a response variable;
The device further includes a gray matter volume estimation unit that estimates the gray matter volume of the subject based on the attribute data of the subject, the sensor data, and the gray matter volume estimation model.

In addition, a system for estimating amyloid β accumulation included in one embodiment of the technology disclosed herein includes a data input unit that inputs attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
an amyloid β accumulation amount estimation model storage unit that stores an amyloid β accumulation amount estimation model obtained by a model created by an analysis process including at least one of regression analysis and machine learning, with attribute data of a third party and sensor data acquired from a sensor device worn by the third party as explanatory variables, and with the amyloid β accumulation amount acquired from the third party as a response variable;
The device further includes an amyloid β accumulation amount estimation unit that estimates the amyloid β accumulation amount of the subject based on the attribute data of the subject, the sensor data, and the amyloid β accumulation amount estimation model.

In addition, an MMSE score estimation system included in one embodiment of the technology disclosed herein includes a data input unit that inputs attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
an MMSE score estimation model storage unit that stores an MMSE score estimation model obtained by a model created by an analysis process including at least one of regression analysis and machine learning, with attribute data of a third party and sensor data acquired from a sensor device worn by the third party as explanatory variables, and an MMSE (Mini Mental State Examination) score acquired from the third party as a response variable;
The apparatus further includes an MMSE score estimation unit that estimates an MMSE score of the subject based on the attribute data of the subject, the sensor data, and the MMSE score estimation model.

In addition, a mild cognitive impairment prediction system included in one embodiment of the technology disclosed herein includes a data input unit that inputs attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
a prediction model storage unit that stores a first prediction model for predicting whether a subject corresponds to either a healthy subject or a person with mild cognitive impairment, the first prediction model being created using time-series attribute data of a third party, including healthy subjects and persons with mild cognitive impairment, and prediction factor information influencing the prediction of mild cognitive impairment in a time series obtained from the third party; and
The method includes an MCI prediction information calculation unit that calculates prediction information for predicting whether the subject is a person with mild cognitive impairment at one or more points in a time series based on the attribute data of the subject, the activity data of the subject, and the first prediction model.

In addition, a mild cognitive impairment prevention system included in one embodiment of the technology disclosed herein includes a data input unit that inputs attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
a prediction model storage unit that stores a first prediction model for predicting whether a subject corresponds to either a healthy subject or a subject with mild cognitive impairment, the first prediction model being created using attribute data of a third party, including a healthy subject and a subject with mild cognitive impairment, and prediction factor information influencing the prediction of mild cognitive impairment over time obtained from the third party; and
An MCI prediction information calculation unit that calculates prediction information for predicting whether the subject corresponds to a mild cognitive impairment at one or more points in a time series based on the attribute data of the subject, the sensor data of the subject, and the first prediction model;
and a goal calculation unit that determines activity data representing the subject's activity converted from the subject's sensor data that can reduce the probability that the subject is predicted to be a patient with mild cognitive impairment at one or more points in the time series.

In addition, a mild cognitive impairment simulation system included in one embodiment of the technology disclosed herein includes a data input unit that inputs attribute data of a subject and activity data representing an activity of the subject;
a prediction model storage unit that stores a first prediction model for predicting whether a subject corresponds to either a healthy subject or a subject with mild cognitive impairment, the first prediction model being created using attribute data of a third party, including a healthy subject and a subject with mild cognitive impairment, and prediction factor information influencing the prediction of mild cognitive impairment over time obtained from the third party; and
The method further includes an MCI simulated judgment information calculation unit that calculates simulated judgment information for judging whether the subject is a person with mild cognitive impairment at one or more points in a time series based on the attribute data of the subject, the activity data of the subject, and the first prediction model.

Furthermore, a sensor device included in one embodiment of the technology disclosed herein is a sensor device capable of providing sensor data to a mild cognitive impairment assessment system, which is one embodiment of the technology disclosed herein, and includes a wearing unit that can be worn on a part of a subject's body, one or more sensor elements, a sensor data storage unit that can store sensor data measured by the sensor elements, and a data providing unit that can provide the sensor data to the mild cognitive impairment assessment system.
The above object may be achieved by a computer constituting a system included in each embodiment of the technology disclosed herein and a computer program (software program) executed on the computer.

In addition, a computer program included in one embodiment of the technology disclosed herein is a program for executing the program on an information processing device,
A process of inputting attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
The system executes a process of calculating judgment information capable of determining whether a subject is a person with mild cognitive impairment based on attribute data of a third party including at least one of healthy individuals and individuals with mild cognitive impairment, and a first judgment model for determining whether a subject is a person with mild cognitive impairment, the first judgment model being created using attribute data of a third party including at least one of healthy individuals and individuals with mild cognitive impairment, and information on factors influencing the determination of mild cognitive impairment obtained from the third party, and a process of calculating judgment information capable of determining whether the subject is a person with mild cognitive impairment based on the attribute data of the subject and the sensor data.

In addition, a computer program included in one embodiment of the technology disclosed herein is a program for executing the program on an information processing device,
A process of inputting attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
The information processing device executes a process of: a first prediction model for predicting whether a subject corresponds to either a healthy person or a person with mild cognitive impairment, the first prediction model being created using attribute data of a third party including at least one of a healthy person and a person with mild cognitive impairment, and predictive factor information influencing the time-series prediction of mild cognitive impairment obtained from the third party; and calculating predictive information for predicting whether the subject corresponds to a person with mild cognitive impairment at one or more points in the time series based on the attribute data of the subject and the sensor data of the subject.

In addition, a computer program included in one embodiment of the technology disclosed herein is a program for executing the program on an information processing device,
A process of inputting attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
A first prediction model for predicting whether a subject corresponds to either a healthy subject or a person with mild cognitive impairment, the first prediction model being created using attribute data of a third party, including healthy subjects and persons with mild cognitive impairment, and prediction factor information influencing prediction of mild cognitive impairment over time obtained from the third party; and a process for calculating prediction information for predicting whether the subject corresponds to a person with mild cognitive impairment at one or more points in time based on the attribute data of the subject and the sensor data;
and determining activity data representing the subject's activities converted from the subject's sensor data that is capable of reducing the probability that the subject is predicted to be a patient with mild cognitive impairment at one or more points in the time series.

According to the technology of the present disclosure, it is possible to provide a mild cognitive impairment assessment system that can easily determine whether or not a subject has MCI without directly measuring information on the subject, such as cerebral glucose metabolic rate, hippocampal volume, gray matter volume, amyloid beta accumulation, and MMSE score. According to the technology of the present disclosure, it is possible to provide a mild cognitive impairment assessment prediction system that can predict the occurrence of MCI, a mild cognitive impairment assessment prevention system that contributes to preventing the occurrence of MCI, and a mild cognitive impairment simulation system that can simulate the probability of occurrence of MCI. Furthermore, according to the technology of the present disclosure, it is possible to provide a sensor device for acquiring data useful in determining the occurrence of MCI, and an analogy system that can easily infer information (cerebral glucose metabolic rate, hippocampal volume, gray matter volume, amyloid beta accumulation, MMSE score) that can be used for the determination of mild cognitive impairment and the diagnosis of dementia. According to the technology of the present disclosure, it is possible to provide a sensor device that can provide data to the mild cognitive impairment assessment system. Furthermore, one of the objectives of the technology disclosed herein is to provide a computer program for executing a mild cognitive impairment assessment system, a mild cognitive impairment prediction system, and a mild cognitive impairment prevention system.

FIG. 1 is a block diagram of a mild cognitive impairment determination system according to one embodiment of the present disclosure. FIG. 2 is a flow diagram of a mild cognitive impairment assessment method executed by the mild cognitive impairment assessment system shown in FIG. FIG. 3 is an example of an ROC curve obtained by the first judgment model. FIG. 4 is a graph showing an example of the relationship between the actual measured MMSE score value and the estimated value calculated by the MMSE score estimation model. FIG. 5 is a graph showing an example of the relationship between the actual measured cerebral glucose metabolic rate value and the estimated value calculated by the cerebral glucose metabolic rate estimation model. FIG. 6 is a graph showing an example of the relationship between the actual measured value of the hippocampal volume and the estimated value calculated by the hippocampal volume estimation model. FIG. 7 is a graph showing an example of the relationship between the actual measured value of the amyloid β accumulation amount and the estimated value calculated by the amyloid β accumulation amount estimation model. FIG. 8 is a block diagram of a mild cognitive impairment determination system according to another embodiment of the present disclosure. FIG. 9 is a block diagram of a mild cognitive impairment prediction system according to one embodiment of the present disclosure. FIG. 10 is a flow diagram of a mild cognitive impairment prediction and prevention method executed by the mild cognitive impairment prediction system shown in FIG. 9 . FIG. 11 is a block diagram of a mild cognitive impairment prediction system according to another embodiment of the present disclosure. FIG. 12 is a block diagram of a mild cognitive impairment prevention system according to one embodiment of the present disclosure. FIG. 13 is a block diagram illustrating a configuration of an information processing device capable of implementing the technology according to the present disclosure.

Embodiments of the technology disclosed herein are described in detail below with reference to the drawings.

<Mild cognitive impairment assessment system>
Fig. 1 is a block diagram of a mild cognitive impairment determination system according to an embodiment of the present disclosure. Fig. 2 is a flow diagram of a mild cognitive impairment determination method executed by the mild cognitive impairment determination system shown in Fig. 1.
As shown in FIG. 1 , the mild cognitive impairment assessment system 1 includes a data input unit 10 for inputting data related to a subject, and an MCI assessment unit 20 .

(Data Entry Section)
The data input unit 10 includes an attribute data input unit 11 for inputting attribute data of the subject, and a sensor data input unit 12 for inputting sensor data acquired from the sensor device 100 worn by the subject. The attribute data of the subject is data that specifies the properties and characteristics of an individual subject. The attribute data of the subject may be, for example, data related to the living environment and lifestyle of the subject that is not measured by the sensor device, or data that is difficult to measure by the sensor device. The attribute data of the subject may include, for example, one or more pieces of data specific to an individual, such as age, sex, educational history, medical history, smoking status, drinking status, high blood pressure, diabetes, and high cholesterol. For example, data acquired by a questionnaire, questionnaire, diagnosis, or the like previously administered to the subject may be input as the attribute data of the subject to the data input unit 10. Note that the data input unit 10 is not limited to the above, and may provide an interface that allows users, administrators, subjects, and the like of the present system to input attribute data. The sensor data is measurement data measured using a sensor element of the sensor device 100. The sensor data may be an average value of measurement data measured during a specific period.

The sensor device 100 is, for example, a wearable sensor, and has a mounting part that can be attached to a subject. The sensor device 100 includes one or more sensor elements. Examples of the sensor elements that can be used include a skin temperature sensor, a pulse (or heart rate) sensor, an acceleration sensor, a microphone (sound sensor), an ambient light sensor, and an ultraviolet sensor. These sensor elements may be used alone or in combination of two or more. The sensor device 100 also includes a sensor data storage unit (e.g., a flash memory) that can store sensor data measured by the sensor element, and a data providing unit that can provide the sensor data to the sensor data input unit 12 of the mild cognitive impairment assessment system 1. The data providing unit may provide data by, for example, data communication via a wireless connection, data communication via a wired connection, or the like. The communication method (protocol) is not particularly limited, and an existing method may be appropriately adopted.

(MCI Judgment Unit)
The MCI assessment unit 20 judges whether the current subject has symptoms of MCI, i.e., whether the current subject corresponds to a person with mild cognitive impairment, based on the data input to the data input unit 10. The MCI assessment unit 20 has an activity data conversion unit 21, a data integration unit 22, a data cleansing unit 23, a judgment model storage unit 24, a factor information calculation unit 25, a judgment information calculation unit 26, and an output unit 27.

The activity data conversion unit 21 converts the sensor data into activity data. The activity data conversion unit 21 generates one or more pieces of activity data by combining the measurement data. The activity data may be data that represents some activity related to the subject, such as physical activity or environmental changes. The activity data may include one or more pieces of information, such as the number of steps taken, conversation time, heart rate (pulse rate), sleep time, UV exposure time, and skin temperature.

The data integration unit 22 integrates data from the attribute data and activity data to be used by the factor information calculation unit 25 and the judgment information calculation unit 26 to create a dataset.

The data cleansing unit 23 detects errors, such as duplications or clerical errors, contained in the data set, and deletes or corrects the detected errors. For example, by cleansing this data, the mild cognitive impairment assessment system 1 improves the quality of the data set, enabling more accurate assessments.

The judgment model storage unit 24 is a storage unit (storage device) that stores the first judgment model. The judgment model storage unit 24 can further store the second judgment model. The judgment model storage unit 24 can also store a factor information estimation model. The first judgment model is a model for calculating judgment information for judging whether the subject corresponds to a person with mild cognitive impairment. The second judgment model is a model for calculating judgment information for judging whether the subject corresponds to a person with mild cognitive impairment of Alzheimer's type (AD type). The factor information estimation model is a model for calculating a factor information estimate value that affects the judgment of mild cognitive impairment (MCI). The factor information estimation model may include, for example, one or more of an MMSE (Mini Mental State Examination) score estimation model, a cerebral glucose metabolic rate estimation model, a hippocampal volume estimation model, a gray matter volume estimation model, and an amyloid β accumulation amount estimation model. The method of determining (creating and learning) these models will be described later.

The factor information calculation unit 25 calculates an estimate of factor information affecting the determination of MCI, using the data set cleansed by the data cleansing unit 23 and the factor information estimation model. The estimate of factor information is, for example, an estimate of the subject's MMSE score, an estimate of cerebral glucose metabolic rate, an estimate of hippocampal volume, an estimate of gray matter volume, and an estimate of the amount of amyloid β accumulation.
The factor information calculation section 25 may calculate an estimate of the MMSE score, for example, by inputting a data set of subjects integrated for MMSE score estimation into an MMSE score estimation model.
The factor information calculation unit 25 may calculate an estimate of the cerebral glucose metabolic rate by, for example, inputting a data set of the subject integrated for estimating the cerebral glucose metabolic rate into a cerebral glucose metabolic rate estimation model.
The factor information calculation section 25 may calculate an estimate of the hippocampal volume by, for example, inputting the subject's data set integrated for hippocampal volume estimation into a hippocampal volume estimation model.
The factor information calculation section 25 may calculate an estimate of the gray matter volume by, for example, inputting a subject's data set integrated for the gray matter volume into a gray matter volume estimation model.
The factor information calculation unit 25 may calculate an estimated value of the accumulated amyloid β amount by, for example, inputting a subject's data set integrated for estimating the accumulated amyloid β amount into an accumulated amyloid β amount estimation model.

The judgment information calculation unit 26 has an MCI judgment information calculation unit that calculates judgment information capable of determining whether the subject corresponds to mild cognitive impairment using the subject's attribute data, the factor information estimate, and the first judgment model. The MCI judgment information calculation unit can calculate the probability that the subject has mild cognitive impairment as the judgment information, for example, by inputting the attribute data and the factor information estimate into the first judgment model. If the calculated probability is equal to or greater than the cutoff value, the subject is judged to have mild cognitive impairment.

Furthermore, the judgment information calculation unit 26 may have an AD-type MCI judgment information calculation unit that, when the subject is judged to have mild cognitive impairment, calculates judgment information for judging whether the subject corresponds to AD-type mild cognitive impairment using the subject's attribute data, the factor information estimate, and the second judgment model. The AD-type MCI judgment information calculation unit can calculate the probability that the subject has AD-type mild cognitive impairment as judgment information, for example, by inputting the attribute data and the factor information estimate into the second judgment model. When the calculated probability is equal to or greater than the cutoff value, the subject is judged to have AD-type mild cognitive impairment.

The output unit 27 outputs the judgment information calculated by the judgment information calculation unit 26 or the judgment result judged from the judgment information, i.e., whether the subject is a healthy individual, a person with mild cognitive impairment due to AD, or a person with mild cognitive impairment due to non-AD. The output unit 27 can be, for example, a display that displays the judgment result as an image, a printer that prints it on paper, or a communication device that transmits it as electronic information.

Next, a method for determining (creating and learning) the first judgment model, the second judgment model, and the factor information estimation model will be described.
In order to determine the above model, first, data of a third party is prepared. The third party includes one or more of a healthy person, an AD-type mild cognitive impairment person, and a non-AD-type mild cognitive impairment person. Typically, the third party may include one or more of each of a healthy person, an AD-type mild cognitive impairment person, and a non-AD-type mild cognitive impairment person.
Then, attribute data of the third party, activity data obtained by converting sensor data obtained from a sensor device worn by the third party, and factor information obtained from the third party (MMSE score, cerebral glucose metabolic rate, hippocampal volume, gray matter volume, amyloid β accumulation amount) are obtained. The MMSE score can be obtained, for example, by MMSE (short-term mental status examination). The cerebral glucose metabolic rate can be obtained by FDG-PET examination, the hippocampal volume and gray matter volume by MRI examination, and the amyloid β accumulation amount by amyloid-PET examination. The attribute data of the third party may be obtained, for example, from a questionnaire, a questionnaire, a preliminary diagnosis, or the like, administered in advance to the third party.

The first decision model can be determined as follows.
The third-party data will be divided into healthy individuals and others, i.e., individuals with mild cognitive impairment (including those with AD-type mild cognitive impairment and those with non-AD-type mild cognitive impairment).
Next, a dataset is created in which the categories of healthy individuals and those with mild cognitive impairment are used as dependent variables, and the subject's attribute data and subject's factor information are used as explanatory variables, with the dependent variables and explanatory variables being row vectors.

Next, an analysis process including at least one of regression analysis and machine learning is performed on the created dataset to create a model. The analysis process is preferably machine learning using regression analysis. There are no particular limitations on the regression analysis method, and binomial logistic regression analysis, support vector machine regression analysis, and random forest regression analysis can be used. For the created model, an ROC curve is created, and the cutoff value, sensitivity, specificity, and accuracy rate for determining that the patient has mild cognitive impairment are calculated. Then, a model is selected so that the sensitivity and specificity of the ROC curve are high and the accuracy rate is high (false positive rate and false negative rate are low), and this is set as the first judgment model.

Figure 3 shows an example of an ROC curve obtained by the first judgment model. The ROC curve shown in Figure 3 has a sensitivity of 91.7%, a specificity of 82.5%, and a cutoff value of 0.5311981.

The second judgment model can be determined as follows.
The third-party data is divided into those with mild cognitive impairment due to AD and those without mild cognitive impairment due to AD.
Then, the second model is determined in the same manner as the first judgment model, except that the categories of AD-type mild cognitive impairment and non-AD-type mild cognitive impairment are used as objective variables.

The factor information estimation model can be determined as follows.
A principal factor analysis is performed using the third-party attribute data and activity data as explanatory variables to select the main explanatory variables. The number of explanatory variables selected may typically be 10 or more. During this selection, some variables may be standardized.
Next, a data set is created in which the third party's factor information is the objective variable, and the objective variable and the selected explanatory variable are row vectors. Here, by appropriately setting, for example, the MMSE score, cerebral glucose metabolic rate, hippocampal volume, gray matter volume, and amyloid β accumulation amount as the objective factor information, a factor information estimation model including one or more of the MMSE score estimation model, cerebral glucose metabolic rate estimation model, hippocampal volume estimation model, gray matter volume estimation model, and amyloid β accumulation amount estimation model can be created.

Next, an analysis process including at least one of regression analysis and machine learning is performed on the created dataset to create a model. A factor information estimation model is selected from the created models using the coefficient of determination and the root mean square error of the created model as judgment criteria. When creating the model, a neural network may be used to make adjustments such as increasing the number of hidden layers until an optimal model is obtained.

Figures 4 to 6 show an example of the relationship between the estimated value of factor information calculated using the factor information estimation model determined by the above method and the actual measured value of factor information.

4 is a graph showing the relationship between the actual measured MMSE scores and the estimated values calculated by the MMSE score estimation model. The regression line between the actual measured and estimated MMSE scores shown in FIG. 4 has a coefficient of determination ( ^R2 ) of 0.98 and a root mean square error (RMSE) of 0.24.

Fig. 5 is a graph showing the relationship between the measured cerebral glucose metabolic rate and the estimated value calculated by the cerebral glucose metabolic rate estimation model. The regression line between the measured and estimated cerebral glucose metabolic rates shown in Fig. 5 has a coefficient of determination ( ^R2 ) of 0.97 and a root mean square error (RMSE) of 0.02.

Fig. 6 is a graph showing the relationship between the measured hippocampal volume and the estimated value calculated by the hippocampal volume estimation model. The regression line between the measured and estimated hippocampal volumes shown in Fig. 6 has a coefficient of determination ( ^R2 ) of 0.99 and a root mean square error (RMSE) of 125.71.

7 is a graph showing the relationship between the actual measured values of the amount of accumulated amyloid β and the estimated values calculated by the amyloid β accumulation estimation model. The regression line between the actual measured values and the estimated values of the amount of accumulated amyloid β shown in FIG. 7 has a coefficient of determination (R ² ) of 0.884 and a root mean square error (RMSE) of 0.026.

(MCI Assessment Method)
The MCI assessment method that can be executed by the mild cognitive impairment assessment system 1 includes a sensor data acquisition step S10, an attribute data input step S11, a sensor data input step S12, an activity data conversion step S21, a data integration step S22, a data cleansing step S23, a factor information calculation step S25, a judgment information calculation step S26, and an output step S27, as shown in FIG. 2. These steps (steps) may be further subdivided into detailed steps, or some steps may be integrated. In addition, the mild cognitive impairment assessment system 1 may change the order of execution of at least some of these steps to the extent that the result is not affected, or may execute at least some of the steps in parallel.

In the sensor data acquisition step S10, sensor data of the subject is acquired using the sensor device 100 worn by the subject.
In an attribute data input step S11, attribute data of the subject is input to the attribute data input unit 11. In a sensor data input step S12, sensor data acquired from the sensor device 100 is input to the sensor data input unit 12. The attribute data input to the attribute data input unit 11 is sent to the data integration unit 22 and the determination information calculation unit 26. The attribute data input to the sensor data input unit 12 is sent to the data integration unit 22 and the determination information calculation unit 26.

In the activity data conversion step S21, the sensor data is converted into activity data by the activity data conversion unit 21. The activity data is sent to the data integration unit 22. In the data integration step S22, the data integration unit 22 integrates the activity data and the attribute data to create a dataset. The dataset is sent to the data cleansing unit 23.

In the data cleansing step S23, the data cleansing unit 23 cleanses the data set by deleting or correcting errors such as duplications and clerical errors. The cleansed data set is then sent to the factor information calculation unit 25. In the factor information calculation step S25, the factor information calculation unit 25 calculates a factor information estimate value using the cleansed data set and a factor information estimation model. The calculated factor information estimate value is sent to the judgment information calculation unit 26.

In the judgment information calculation step S26, the judgment information calculation unit 26 calculates judgment information for judging whether the subject corresponds to a person with mild cognitive impairment using the attribute data, the factor information estimate, and the first judgment model sent from the judgment model storage unit 24. In addition, if the subject is judged to be mildly cognitively impaired in the judgment information calculation step S26, the judgment information calculation unit 26 may calculate judgment information for judging whether the subject corresponds to a person with mild cognitive impairment due to AD using the attribute data, the factor information estimate, and the second judgment model sent from the judgment model storage unit 24. The judgment information is sent to the output unit 27. In the output step S27, the judgment information or the judgment result judged from the judgment information (whether the subject is a healthy person, a person with mild cognitive impairment due to AD, or a person with mild cognitive impairment due to non-AD) is output by the output unit 27.

The mild cognitive impairment assessment system 1 of this embodiment configured as described above does not need to directly measure the factor information of the subject, for example. Therefore, it is possible to easily determine whether or not the subject has MCI. In addition, the mild cognitive impairment assessment system 1 of this embodiment determines whether or not the subject has MCI using a first assessment model created using attribute data of third parties including healthy people and people with mild cognitive impairment, and factor information that influences the assessment of mild cognitive impairment obtained from the third party. Therefore, it is possible to measure with high accuracy whether or not the subject has MCI. Therefore, it is possible to suppress the onset of dementia or improve it by taking appropriate measures based on the obtained assessment result.

In addition, in the mild cognitive impairment assessment system 1 of this embodiment, a factor information estimate is calculated based on the subject's attribute data, activity data, and a factor information estimation model, and whether or not the subject has MCI is determined based on this factor information estimate, the subject's attribute data, and the first assessment model. Therefore, it is possible to measure with higher accuracy whether or not the subject has MCI.

Furthermore, in the mild cognitive impairment assessment system 1 of this embodiment, it is possible to determine whether or not a subject has AD-type MCI using a second assessment model created using attribute data of a third party, including persons with mild cognitive impairment due to AD and healthy persons with non-AD mild cognitive impairment, and factor information influencing the assessment of mild cognitive impairment obtained from the third party. In other words, subjects can be divided into healthy persons, persons with mild cognitive impairment due to AD, and persons with non-AD mild cognitive impairment.

The mild cognitive impairment assessment system 1 of this embodiment can also be used as a cerebral glucose metabolic rate estimation system, a hippocampal volume estimation system, a gray matter volume estimation system, an amyloid β accumulation amount estimation system, and an MMSE score estimation system. These systems can be used for purposes other than diagnosing MCI. When these systems are used for purposes other than diagnosing MCI, the third party that provides the attribute data, activity data, and actual measured values such as cerebral glucose metabolic rate, hippocampal volume, gray matter, amyloid β accumulation amount, and MMSE score can be set according to the purpose, and does not have to include people with mild cognitive impairment.

The mild cognitive impairment assessment system according to the present disclosure is not limited to the mild cognitive impairment assessment system 1 of the present embodiment. For example, the mild cognitive impairment assessment system 1 of the present embodiment calculates a factor information estimate value based on a factor information estimation model. In contrast, the mild cognitive impairment assessment system according to the present disclosure may, for example, determine whether or not the subject has MCI or whether or not the subject has AD-type MCI based on the subject's attribute data, the subject's activity data, and the first judgment model without calculating a factor information estimate value. In addition, the mild cognitive impairment assessment system according to the present disclosure may, for example, determine whether or not the subject has MCI or whether or not the subject has AD-type MCI based on the subject's attribute data, data representing factor information measured and determined from the subject using a diagnostic device, a measuring device, or the like, the subject's activity data, and the first judgment model. Furthermore, it may be determined whether or not the subject has MCI or whether or not the subject has AD-type MCI based on the first judgment model without converting the sensor data into activity data.

In addition, the mild cognitive impairment assessment system 1 of this embodiment uses a cerebral glucose metabolic rate estimate, a hippocampal volume estimate, a gray matter volume estimate, an amyloid β accumulation amount estimate, and an MMSE score estimate as factor information estimates. The mild cognitive impairment assessment system according to the present disclosure does not need to use all of these factor information estimates, and may use, for example, any one or more of the factor information estimates. In addition, the mild cognitive impairment assessment system according to the present disclosure may use estimates other than the cerebral glucose metabolic rate estimate, the hippocampal volume estimate, the gray matter volume estimate, the amyloid β accumulation amount estimate, and the MMSE score estimate as factor information estimates. Furthermore, the mild cognitive impairment assessment system according to the present disclosure may use actual measured values for some items of the cerebral glucose metabolic rate, hippocampal volume, gray matter volume, amyloid β accumulation amount, and MMSE score. Specifically, the MMSE score may be an actual measured value obtained from the subject, and the cerebral glucose metabolic rate, hippocampal volume, gray matter volume, and amyloid β accumulation amount may be estimated values.

Furthermore, as one embodiment, the mild cognitive impairment assessment system according to the present disclosure may be realized as a mild cognitive impairment assessment system 1000 illustrated in FIG. 8, for example.
The mild cognitive impairment assessment system 1000 may have a data input unit 1001, a assessment model storage unit 1002, and an MCI assessment information calculation unit 1003.

The data input unit (data input means) 1001 inputs the subject's attribute data and the sensor data acquired from the sensor device worn by the subject. The data input unit 1001 may be configured in the same manner as the data input unit 10 illustrated in FIG. 1, for example.

The judgment model storage unit 1002 stores a first judgment model for judging whether a subject corresponds to either a healthy individual or a person with mild cognitive impairment. The first judgment model may be created, for example, using attribute data of a third party including at least one of a healthy individual and a person with mild cognitive impairment, and factor information influencing the judgment of mild cognitive impairment obtained from the third party. The first judgment model may be created by the method described above.

The MCI determination information calculation unit 1003 calculates determination information that can determine whether the subject corresponds to a person with mild cognitive impairment based on the subject's attribute data, the activity data representing the subject's activity converted from the sensor data, and the first determination model. The MCI determination information calculation unit 1003 may calculate the determination information, for example, by applying the subject's attribute data and factor information that affects the determination of mild cognitive impairment (MCI) for the subject to the first determination model. In addition, as described above, the mild cognitive impairment determination system 1000 may calculate the determination information by estimating factor information for the subject from the subject's attribute data and the activity data representing the subject's activity converted from the sensor data, and applying the estimated value and the subject's attribute data to the first determination model. The determination information calculated by the MCI determination information calculation unit 1003 may be, for example, information representing the probability that the subject is mildly cognitively impaired.

The mild cognitive impairment assessment system 1000 configured as described above can easily determine whether or not a subject has MCI, for example, based on the subject's attribute data and activity data, similar to the mild cognitive impairment assessment system 1.

<Mild cognitive impairment prediction system>
Fig. 9 is a block diagram of a mild cognitive impairment prediction system according to an embodiment of the present disclosure. Fig. 10 is a flow diagram of a mild cognitive impairment prediction and prevention method executed by the mild cognitive impairment prediction system of Fig. 9.
9, the mild cognitive impairment prediction system 2 includes a data input unit 10, an MCI prediction unit 30, and a target calculation unit 40 that determines the activity data that can reduce the probability that the subject is predicted to be a mild cognitively impaired person. Specifically, the target calculation unit 40 may calculate a target value of the activity data that can reduce the probability that the subject is predicted to be a mild cognitively impaired person.

(Data Entry Section)
The data input unit 10 includes an attribute data input unit 11 for inputting attribute data of a subject, and a sensor data input unit 12 for inputting sensor data acquired from a sensor device 100 worn by the subject. The data input unit 10 and the sensor device 100 are the same as the data input unit 10 and the sensor device 100 of the mild cognitive impairment assessment system 1, and therefore the same symbols are used and detailed descriptions are omitted.

(MCI prediction unit)
The MCI prediction unit 30 predicts, for example, whether the subject will develop MCI in the future, that is, whether the subject will be a mild cognitive impairment at one or more time points in the time series, based on the data input to the data input unit 10. The one or more time points in the time series may be, for example, one or more time points in the future time series, such as one year, two years, or three years from a certain time point (for example, the present). The MCI prediction unit 30 has an activity data conversion unit 31, a data integration unit 32, a data cleansing unit 33, a prediction model storage unit 34, a prediction factor information calculation unit 35, a prediction information calculation unit 36, and an output unit 37. The activity data conversion unit 31, the data integration unit 32, the data cleansing unit 33, and the output unit 37 have the same functions as the activity data conversion unit 21, the data integration unit 22, the data cleansing unit 23, and the output unit 27 of the mild cognitive impairment assessment system 1, respectively, and therefore will not be described.

The prediction model storage unit 34 is a storage unit (storage device) that stores the first prediction model. The prediction model storage unit 34 may further store a second prediction model. The prediction model storage unit 34 may further store a prediction factor information estimation model. The first prediction model is a model for calculating prediction factor information for predicting whether a future subject will be a patient with mild cognitive impairment. The second judgment model is a model for calculating prediction factor information for predicting whether a future subject will be a patient with mild cognitive impairment of Alzheimer's type (AD type). The prediction factor information estimation model is a model for calculating a prediction factor information estimate value that affects the prediction of MCI in a time series. The prediction factor information estimation model may include, for example, one or more of a predicted MMSE score estimation model, a predicted cerebral glucose metabolic rate estimation model, a predicted hippocampal volume estimation model, a predicted gray matter volume estimation model, and a predicted amyloid β accumulation amount estimation model. A method for determining (creating and learning) these models will be described later.

The prediction factor information calculation unit 35 uses the data set cleansed by the data cleansing unit 33 and the prediction factor information estimation model to calculate a prediction factor information estimate capable of predicting the occurrence of MCI in the subject in the future. The prediction factor information estimate may be, for example, an estimate of the predicted MMSE score of the subject in the future (at one or more points in time in the time series), an estimate of the predicted cerebral glucose metabolic rate, an estimate of the predicted hippocampal volume, an estimate of the predicted gray matter volume, or an estimate of the predicted amount of amyloid beta accumulation. For example, the prediction factor information calculation unit 35 may calculate the estimate of the predicted MMSE score by inputting the subject's data set integrated for estimating the predicted MMSE score into the predicted MMSE score estimation model. The prediction factor information calculation unit 35 may calculate the estimate of the predicted cerebral glucose metabolic rate by inputting the subject's data set integrated for estimating the predicted cerebral glucose metabolic rate into the predicted cerebral glucose metabolic rate estimation model. The estimated value of the hippocampal volume may be calculated by the prediction factor information calculation unit 35 inputting the subject's data set integrated for estimating the predicted hippocampal volume into the predicted hippocampal volume estimation model. The estimated value of the predicted gray matter volume may be calculated by the prediction factor information calculation unit 35 inputting the subject's data set integrated for estimating the predicted gray matter volume into the predicted gray matter volume estimation model. The estimated value of the predicted amyloid β accumulation amount may be calculated by the prediction factor information calculation unit 35 inputting the subject's data set integrated for estimating the predicted amyloid β accumulation amount into the predicted amyloid β accumulation amount estimation model.

The prediction information calculation unit 36 has an MCI prediction information calculation unit that uses the subject's attribute data, the prediction factor information estimate, and the first prediction model to calculate judgment information for determining whether a future subject will have mild cognitive impairment. The MCI prediction information calculation unit calculates the probability that the subject will have mild cognitive impairment at one or more points in the time series by inputting the dataset and the prediction factor information estimate into the first prediction model. If the calculated probability is equal to or greater than the cutoff value, the subject is predicted to have mild cognitive impairment at one or more points in the time series.

Furthermore, the prediction information calculation unit 36 has an AD-type MCI prediction information calculation unit that, when it is determined that the future subject has mild cognitive impairment, calculates judgment information for determining whether the future subject corresponds to AD-type mild cognitive impairment using the subject's attribute data, the prediction factor information estimate, and the second prediction model. The AD-type MCI prediction information calculation unit calculates the probability that the subject at one or more time points in the time series has AD-type mild cognitive impairment by inputting the dataset and the prediction factor information estimate into the second judgment model. When the calculated probability is equal to or greater than the cutoff value, the subject at one or more time points in the time series is predicted to have AD-type mild cognitive impairment.

Next, a method for determining the first prediction model, the second prediction model, and the prediction factor information estimation model will be described.
In order to determine the above model, first, third-party data is prepared. The third party includes one or more of a healthy person, an AD-type mild cognitive impairment person, and a non-AD-type mild cognitive impairment person. Typically, the third party may include one or more of a healthy person, an AD-type mild cognitive impairment person, and a non-AD-type mild cognitive impairment person. Then, attribute data of the third party measured at multiple time points in the time series, activity data obtained by converting sensor data acquired from a sensor device worn by the third party, and factor information acquired from the third party (MMSE score, cerebral glucose metabolic rate, hippocampal volume, gray matter volume, amyloid β accumulation amount) are obtained. The multiple time points in the time series are, for example, multiple time points spaced apart in the time series, such as the second year, the third year, etc., with the first year being the time when data such as the cerebral glucose metabolic rate is first acquired. The chronological interval may typically be within a range of 2 months to 2 years.

The first prediction model can be determined (created and learned) as follows.
The third-party data is divided into healthy individuals and others, i.e., individuals with mild cognitive impairment (including those with mild cognitive impairment due to Alzheimer's disease and those with mild cognitive impairment without Alzheimer's disease).
Next, a dataset is created in which the categories of healthy individuals and those with mild cognitive impairment are used as objective variables, and the subject's attribute data, cerebral glucose metabolic rate, hippocampal volume, gray matter volume, amyloid beta accumulation, and MMSE score are used as explanatory variables, and the objective variable and explanatory variables are used as row vectors, to which the unit of time when the data was obtained (e.g., 1st year, 2nd year, 3rd year, etc.) is added as a vertical vector.

Next, an analysis process including at least one of regression analysis and machine learning is performed on the created dataset to create a model. The analysis process is preferably machine learning using regression analysis. There are no particular limitations on the regression analysis method, and binomial logistic regression analysis, support vector machine regression analysis, and random forest regression analysis can be used. For the created model, an ROC curve is created, and the cutoff value, sensitivity, specificity, and accuracy rate for predicting mild cognitive impairment are calculated. Then, a model is selected so that the sensitivity and specificity of the ROC curve are high and the accuracy rate is high (false positive rate and false negative rate are low), and this is designated as the first prediction model.

The second prediction model can be determined as follows.
Third parties are divided into those with AD-type mild cognitive impairment and those with non-AD-type mild cognitive impairment.
Then, the second model is determined in the same manner as the first prediction model, except that the categories of AD-type mild cognitive impairment and non-AD-type mild cognitive impairment are used as objective variables.

The predictive factor information estimation model can be determined as follows.
A principal factor analysis is performed using the third-party attribute data and activity data as explanatory variables to select the main explanatory variables. The number of explanatory variables to be selected may typically be 10 or more. In this selection, some variables may be standardized.
Next, a data set is created in which the factor information is the objective variable, and the objective variable and the selected explanatory variable are row vectors, and the time unit (e.g., 1st year, 2nd year, 3rd year, etc.) when the data was acquired is added as a vertical vector. This creates a data set including the objective variable and explanatory variables at one or more time points in the time series. Here, by appropriately setting, for example, the MMSE score, cerebral glucose metabolic rate, hippocampal volume, gray matter volume, and amyloid β accumulation amount as the target factor information, a prediction factor information estimation model including one or more of a predicted MMSE score estimation model, a predicted cerebral glucose metabolic rate estimation model, a predicted hippocampal volume estimation model, a predicted gray matter volume estimation model, and a predicted amyloid β accumulation amount estimation model can be created.

Next, an analysis process including at least one of regression analysis and machine learning is performed on the created dataset to create a model. A predictive factor information estimation model is selected from the created models using the coefficient of determination and the root mean square error of the created model as criteria for judgment. When creating the model, a neural network may be used to make adjustments such as increasing the number of hidden layers until an optimal model is obtained.

(MCI prediction method)
As shown in Fig. 10, the MCI prediction method executed by the mild cognitive impairment prediction system 2 includes a sensor data acquisition step S10, an attribute data input step S11, a sensor data input step S12, an activity data conversion step S31, a data integration step S32, a data cleansing step S33, a prediction factor information calculation step S35, a prediction factor information calculation step S36, and an output step S37. These steps (steps) may be further subdivided into detailed steps, or some steps may be integrated. In addition, the mild cognitive impairment prediction system 2 may change the order of execution of at least some of these steps to the extent that the result is not affected, or may execute at least some of the steps in parallel.

In the sensor data acquisition step S10, sensor data of the subject is acquired using the sensor device 100 worn by the subject.
In an attribute data input step S11, attribute data of the subject is input to the attribute data input unit 11. In a sensor data input step S12, sensor data acquired from the sensor device 100 is input to the sensor data input unit 12. The attribute data input to the attribute data input unit 11 is sent to the data integration unit 32 and the prediction information calculation unit 36. The attribute data input to the sensor data input unit 12 is sent to the data integration unit 32 and the prediction information calculation unit 36.

In the activity data conversion step S31, the activity data conversion unit 31 converts the sensor data into activity data. The activity data is sent to the data integration unit 32. In the data integration step S32, the data integration unit 32 integrates the activity data and the attribute data to create a data set.
The data set is sent to a data cleansing unit 33 .

In the data cleansing step S33, the data cleansing unit 23 cleanses the data set by deleting or correcting errors such as duplications and clerical errors. The cleansed data set is then sent to the prediction factor information calculation unit 35. In the prediction factor information calculation step S35, the prediction factor information calculation unit 35 calculates a prediction factor information estimate value using the cleansed data set and the prediction factor information estimation model. The calculated prediction factor information estimate value is sent to the prediction information calculation unit 36.

In the prediction information calculation step S36, the prediction information calculation unit 36 calculates prediction factor information capable of predicting whether the subject at one or more time points in the time series corresponds to a person with mild cognitive impairment, using the attribute data, the factor information estimate, and the first prediction model sent from the prediction model storage unit 34. In addition, in the prediction factor information calculation step S26, when the subject at one or more time points in the time series is predicted to be a person with mild cognitive impairment, the prediction information calculation unit 36 calculates prediction factor information capable of predicting whether the subject at one or more time points in the time series corresponds to a person with mild cognitive impairment of AD type, using the attribute data, the factor information estimate, and the second prediction model sent from the prediction model storage unit 34.
The prediction factor information is sent to the output unit 37. In the output step S37, the prediction factor information or a prediction result predicted from the prediction factor information (whether the subject at one or more time points in the time series is a healthy subject, a person with mild cognitive impairment due to AD, or a person with mild cognitive impairment due to non-AD) is output by the output unit 37.

(Target Calculation Unit)
The target calculation unit 40 determines the activity data capable of reducing the probability that the subject is predicted to be a person with mild cognitive impairment. The target calculation unit 40 may, for example, calculate a target value of the activity data capable of reducing the probability that the subject is predicted to be a person with mild cognitive impairment. Specifically, the target calculation unit 40 includes an activity data input unit 41, a data integration unit 42, a data cleansing unit 43, a simulated judgment factor information calculation unit 44, a simulated judgment information calculation unit 45, a simulated judgment information confirmation unit 46, and an output unit 47.

The activity data input unit 41 inputs activity data. The activity data is, for example, data related to information such as the number of steps, conversation time, heart rate (pulse rate), sleep time, UV exposure time, and skin temperature. The activity data input unit 41 can, for example, input or accept numerical values related to these activity data. The activity data may be, for example, data converted from the above-mentioned sensor data. The activity data may also be data appropriately input by a user, administrator, or subject of the mild cognitive impairment prediction system 2. In this case, for example, simulated activity data for a certain subject (for example, activity data in which the number of steps, conversation time, heart rate (pulse rate), sleep time, UV exposure time, skin temperature, etc. are simulated) may be input to the activity data input unit 41.

The data integration unit 42 integrates the data to be used by the simulated assessment factor information calculation unit 44 and the simulated assessment information calculation unit 45 from among the attribute data input to the attribute data input unit 11 and the activity data input to the activity data input unit 41 to create a data set.

The data cleansing unit 43 detects errors such as duplications and clerical errors in the data set, and deletes or corrects the detected errors.

The simulated judgment factor information calculation unit 44 uses the data set cleansed by the data cleansing unit 23 and the prediction factor information estimation model stored in the prediction model storage unit 34 to calculate a simulated judgment factor information estimate value for simulating the occurrence of MCI in future subjects. The simulated judgment factor information calculation unit 44 can calculate a simulated judgment factor information estimate value, for example, by inputting a data set into the prediction factor information estimation model. Such a simulated judgment factor information estimate value can also be considered as an estimate value of factor information when a specific value is set as a certain activity data or when a certain activity data is changed. As a result, the simulated judgment factor information calculation unit 44 can calculate an estimate value of factor information when, for example, activity data in which a certain activity data is pseudo-changed is input in the activity data input unit 41 (for example, when data in which the number of steps, conversation time, heart rate (pulse rate), sleep time, UV exposure time, skin temperature, etc. is changed is input).

The simulated judgment information calculation unit 45 uses the subject's attribute data, the simulated judgment factor information estimate, and the first prediction model to calculate judgment information for judging whether a future subject will be a person with mild cognitive impairment. Specifically, the simulated judgment information calculation unit 45 can calculate the probability that the subject will be a person with mild cognitive impairment at one or more points in time in the time series by inputting the dataset and the simulated judgment factor information estimate into the first prediction model. If the calculated probability is equal to or greater than the cutoff value, the subject at one or more points in time in the time series is simulated to be a person with mild cognitive impairment.

Furthermore, when the future subject is simulated to be a person with mild cognitive impairment, the simulated judgment information calculation unit 45 uses the subject's attribute data, the simulated judgment factor information estimate, and the second prediction model to calculate simulated judgment information for judging whether the future subject corresponds to a person with mild cognitive impairment of AD type. Specifically, the simulated judgment information calculation unit 45 inputs the dataset and the simulated judgment factor information estimate into the second judgment model to calculate the probability that the subject at one or more points in the time series is a person with mild cognitive impairment of AD type. When the calculated probability is equal to or greater than the cutoff value, the subject at one or more points in the time series is simulated to be a person with mild cognitive impairment of AD type.

The simulated assessment information confirmation unit 46 compares the probability that the subject has mild cognitive impairment at one or more points in the time series with the probability that the subject has mild cognitive impairment at one or more points in the time series predicted by the aforementioned MCI prediction unit 30.

The output unit 47 displays activity data capable of reducing the probability that the subject is predicted to be a person with mild cognitive impairment at one or more points in the time series. As a specific example, the output unit 47 may display a target value of activity data capable of reducing the probability that the subject is predicted to be a person with mild cognitive impairment at one or more points in the time series. In other words, when the simulated judgment information confirmation unit 46 determines that the probability that the subject is a person with mild cognitive impairment at one or more points in the time series has decreased, the activity data input by the activity data input unit 41 is output.

(Methods for preventing MCI)
As shown in Fig. 9, the MCI prevention method executed by the mild cognitive impairment prediction system 2 includes an activity data input step S41, a data integration step S42, a data cleansing step S43, a simulated judgment factor information calculation step S45, a simulated judgment information confirmation step S46, and an output step S47. These steps (steps) may be further divided into detailed steps, or some steps may be integrated. In addition, the mild cognitive impairment prediction system 2 may change the order of execution of at least some of these steps to the extent that the result is not affected, or may execute at least some of the steps in parallel.

In the activity data input step S41, any activity data is input in the activity data input unit 41. The activity data is sent to the data integration unit 42. In the data integration step S42, the data integration unit 42 integrates the activity data with the attribute data sent from the attribute data input unit 11 to create a data set. The data set is sent to the data cleansing unit 43.

In the data cleansing step S43, the data cleansing unit 43 cleanses the data set by deleting or correcting errors such as duplications and clerical errors. The cleansed data set is then sent to the simulated judgment factor information calculation unit 44. In the simulated judgment factor information calculation step S44, the simulated judgment factor information calculation unit 44 calculates a simulated judgment factor information estimate value using the cleansed data set and the prediction factor information estimation model. The calculated simulated judgment factor information estimate value is sent to the simulated judgment information calculation unit 45.

In the simulated judgment information calculation step S45, the simulated judgment information calculation unit 45 calculates the probability that the subject at one or more time points in the time series corresponds to a person with mild cognitive impairment using the attribute data sent from the attribute data input unit 11, the simulated judgment factor information estimate, and the first prediction model sent from the prediction model storage unit 34. Also, in the simulated judgment information calculation step S45, if the subject at one or more time points in the time series is simulated to be a person with mild cognitive impairment, the simulated judgment information calculation unit 45 calculates the probability that the subject at one or more time points in the time series corresponds to a person with mild cognitive impairment of AD type using the attribute data, the factor information estimate, and the second prediction model sent from the prediction model storage unit 34. The prediction factor information is sent to the simulated judgment information confirmation unit 46.

In the simulated assessment information confirmation step S46, the simulated assessment information confirmation unit 46 compares the probability that the subject has mild cognitive impairment at one or more time points in the time series calculated in the simulated assessment information calculation step S45 with the probability that the subject has mild cognitive impairment at one or more time points in the time series predicted by the MCI prediction unit 30. If the probability calculated in the simulated assessment information calculation step S45 is equal to or higher than the probability predicted by the MCI prediction unit 30 (NO), the process returns to the activity data input step S41 and continues. That is, in this case, new activity data may be input in the activity data input unit 41.

If the probability calculated in the simulated judgment information calculation process S45 is lower than the probability predicted by the aforementioned MCI prediction unit 30 (YES), the activity data input in the activity data input process S41 is sent to the output unit 47. In the output process S47, the activity data input in the activity data input process S41 is output by the output unit 47.

The mild cognitive impairment prediction system 2 of this embodiment configured as described above does not need to directly measure the subject's factor information. Therefore, it is possible to easily predict whether or not a future subject will develop MCI. Furthermore, the mild cognitive impairment prediction system 2 of this embodiment predicts whether or not a future subject will develop MCI using a first prediction model created using attribute data of third parties, including healthy individuals and persons with mild cognitive impairment, and factor information influencing the prediction of MCI obtained from the third parties. Therefore, it is possible to predict with high accuracy whether or not a future subject will develop MCI.

Furthermore, the mild cognitive impairment prediction system 2 of this embodiment has a goal calculation unit 40 that calculates the above activity data that can reduce the probability that a future subject will be predicted to be a person with mild cognitive impairment. Therefore, by using the mild cognitive impairment prediction system 2 of this embodiment, it is possible to contribute to the prevention of the occurrence of MCI.

In addition, in the mild cognitive impairment prediction system 2 of this embodiment, a factor information estimate is calculated based on the subject's attribute data, activity data, and a factor information estimation model, and whether or not the subject has MCI is determined based on this factor information estimate, the subject's attribute data, and the first determination model. Therefore, it is possible to measure with higher accuracy whether or not the subject has MCI.

The mild cognitive impairment prediction system 2 of this embodiment has the MCI prediction unit 30 and the target calculation unit 40, but the MCI prediction unit 30 and the target calculation unit 40 may be separated. For example, the data input unit and the MCI prediction unit 30 may be combined to form a mild cognitive impairment prediction system. Furthermore, the attribute data input unit 11, the prediction model storage unit 34, and the target calculation unit 40 may be combined to form a mild cognitive impairment prevention system. Furthermore, the mild cognitive impairment prevention system may be a mild cognitive impairment simulation system that calculates simulated judgment information that calculates the probability that the subject is a person with mild cognitive impairment at one or more points in time in the time series as a simulated judgment based on the activity data input to the activity data input unit. Note that the mild cognitive impairment simulation system does not need to include the simulated judgment information confirmation unit 46.

Furthermore, as one embodiment, the mild cognitive impairment prediction system according to the present disclosure may be realized as a mild cognitive impairment prediction system 1100 illustrated in FIG. 11, for example.
The mild cognitive impairment prediction system 1100 has a data input unit 1101, a prediction model storage unit 1102, and an MCI prediction information calculation unit 1103.

The data input unit 1101 inputs attribute data of the subject and sensor data acquired from a sensor device worn by the subject. The data input unit 1101 may receive the attribute data and the sensor data as input.
The prediction model storage unit 1102 stores a first prediction model for predicting whether a subject corresponds to either a healthy person or a person with mild cognitive impairment. The first prediction model may be a model created using time-series attribute data of a third party including at least one of a healthy person and a person with mild cognitive impairment, and prediction factor information that influences the prediction of time-series mild cognitive impairment obtained from the third party. The above-mentioned methods can be appropriately adopted as a method for creating the first prediction model.
The MCI prediction information calculation unit 1103 calculates prediction information capable of predicting whether the subject corresponds to mild cognitive impairment at one or more time points in a time series based on the subject's attribute data, the subject's activity data, and the first prediction model. The MCI prediction information calculation unit 1103 may calculate the prediction information, for example, by applying the subject's attribute data and prediction factor information that influences the prediction of mild cognitive impairment in the time series related to the subject to the first prediction model. In addition, the MCI prediction information calculation unit 1103 may calculate the prediction information by, for example, estimating prediction factor information from the subject's attribute data and the subject's activity data as described above, and applying the estimated value and the subject's attribute data to the first prediction model.

Furthermore, as one embodiment, the mild cognitive impairment prevention system according to the present disclosure may be realized as a mild cognitive impairment prevention system 1200 illustrated in FIG. 12, for example.
The mild cognitive impairment prevention system 1200 includes a data input unit 1201, a prediction model storage unit 1202, an MCI prediction information calculation unit 1203, and a target calculation unit 1204. In this case, the data input unit 1201, the prediction model storage unit 1202, and the MCI prediction information calculation unit 1203 may be configured similarly to the data input unit 1101, the prediction model storage unit 1102, and the MCI prediction information calculation unit 1103 illustrated in FIG.

The goal calculation unit 1204 calculates activity data that can reduce the probability that the subject is predicted to be a person with mild cognitive impairment at one or more points in time in the time series. Specifically, the goal calculation unit 1204 calculates prediction information that can predict whether the subject is a person with mild cognitive impairment at one or more points in time series based on the subject's attribute data, the subject's activity data, and the first prediction model, for example, in the same way as the mild cognitive impairment prediction system 1100. Furthermore, the goal calculation unit 1204 calculates simulated judgment information that can predict whether the subject is a person with mild cognitive impairment at one or more points in time series under conditions in which the activity data has been changed, based on the subject's attribute data, the subject's activity data that is different from the data used to calculate the prediction information, and the first prediction model. The goal calculation unit 1204 may compare the calculated judgment information with the simulated judgment information to determine whether it is possible to reduce the probability that the subject is predicted to be a person with mild cognitive impairment by changing the specific activity data. In this case, the goal calculation unit 1204 may output activity data that can reduce the probability that the person is predicted to have mild cognitive impairment, for example.

The mild cognitive impairment prediction system 2, mild cognitive impairment prediction system 1100, and mild cognitive impairment prevention system 1200 of the present embodiment configured as described above can easily predict whether a future subject will develop MCI, as described above. Furthermore, the mild cognitive impairment prediction system 2 and mild cognitive impairment prevention system 1200 of the present embodiment can calculate the above activity data that can reduce the probability that a future subject will be predicted to have mild cognitive impairment, and can therefore contribute to preventing the occurrence of MCI.

<Hardware and software configuration capable of implementing the technology according to the present disclosure>
The following describes hardware and software configurations capable of realizing the technology disclosed herein.

A program for realizing the functions of any of the components in the mild cognitive impairment assessment system and mild cognitive impairment prediction system (e.g., mild cognitive impairment assessment system 1, mild cognitive impairment assessment system 1000, mild cognitive impairment prediction system 2, mild cognitive impairment prediction system 1100, mild cognitive impairment prevention system 1200, hereinafter collectively referred to simply as "mild cognitive impairment assessment system") according to the present disclosure described above may be recorded on a computer-readable recording medium, and the program may be read into a computer system and executed. Note that "computer system" here includes hardware such as an operating system (OS) or peripheral devices. Also, "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD (Compact Disc)-ROMs, and storage devices such as hard disks built into a computer system. Furthermore, "computer-readable recording medium" includes a medium that holds a program for a certain period of time, such as volatile memory (RAM) inside a computer system that serves as a server or client when a program is transmitted over a network such as the Internet or a communication line such as a telephone line. The recording medium may be a non-transitory recording medium.

The above program may be transmitted from a computer system in which the program is stored in a storage device or the like to another computer system via a transmission medium, or by transmission waves in the transmission medium. Here, the "transmission medium" that transmits the program refers to a medium that has a function of transmitting information, such as a network such as the Internet or a communication line such as a telephone line.
The above program may be for implementing some of the above functions. Furthermore, the above program may be a so-called differential file that can implement the above functions in combination with a program already recorded in the computer system. The differential file may be called a differential program.

The mild cognitive impairment assessment system may be realized by one or more dedicated hardware devices. The mild cognitive impairment assessment system according to the present disclosure may also be realized by one or more general-purpose information processing devices, such as a computer. It may also be realized by a combination of one or more dedicated hardware devices and one or more general-purpose information processing devices. When the system according to the present disclosure is realized using multiple hardware devices, those devices may be communicatively connected by an appropriate communication means.

The dedicated hardware device may be realized, for example, by circuit elements (e.g., integrated circuits, etc.) capable of implementing the functions of each component of the mild cognitive impairment assessment system according to the present disclosure described above.

The general-purpose hardware device may be, for example, an information processing device 1300 as illustrated in FIG. 23. The information processing device 1300 illustrated in FIG. 23 may include a processor (arithmetic device) 1301, a memory 1302, a storage 1303, a communication interface 1304, and an input/output interface 1305. The components included in the information processing device 1300 may be connected to each other so as to be able to communicate with each other as appropriate via a communication bus or the like.

The processor 1301 is an arithmetic device such as a microprocessor (CPU) that reads the programs and data stored in the memory 1302 described below and executes appropriate processing. The processor 1301 may be, for example, a general-purpose arithmetic device capable of interpreting a specific instruction set.

Memory 1302 is a storage device capable of storing data and programs. Memory 1302 may be realized, for example, by a dynamic random access memory (DRAM).

The storage 1303 may be a storage device capable of storing and saving various data. The storage 1303 may be, for example, a storage device having a larger capacity than the memory 1302.
The storage 1303 may be, for example, a storage device realized by a HDD (Hard Disk Drive) or a semiconductor storage device. In the specific example illustrated in FIG. 23, the storage 1303 is included in the information processing device 1300, but this embodiment is not limited to this. For example, the storage 1303 may be realized outside the information processing device 1300 as a currently well-known storage server, cloud storage, or the like. In this case, the information processing device 1300 may access the data stored in the storage 1303, for example, via a communication interface 1304 described later. The various data, models, and the like described in each of the above embodiments may be stored in the storage 1303.

The communication interface 1304 is, for example, an interface device capable of transmitting and receiving various data, etc., via a communication network. The information processing device 1300 may transmit and receive various data, for example, between the information processing device 1300 and the sensor device 100 worn by the subject, via the communication interface 1304.

The input/output interface 1305 is, for example, an interface device capable of controlling the input of various data to the information processing device 1300 and the output of various data from the information processing device 1300. For example, the data input units (10, 1001, 1101, 1201, etc.) described in each of the above embodiments may accept input of various data via the input/output interface 1305. In addition, the output units (27, 37, etc.) described in each of the above embodiments may output various data via the input/output interface 1305.
The input/output interface 1305 may also have an interface function for implementing reading and writing of data to and from the storage medium 1306. In this case, the input/output interface 1305 may function as an interface for a drive device capable of implementing reading and writing of data to and from the storage medium 1306, for example.
The storage medium 1306 is a computer-readable medium, and may be, for example, any of currently available optical storage media, semiconductor storage media, magnetic storage media, magneto-optical storage media, and the like.

A computer program capable of implementing each component of the mild cognitive impairment prediction system described in each of the above embodiments (or each process described above) may be installed in the information processing device 1300 configured as described above. Such a computer program may be stored in the storage 1303 when the information processing device 1300 is manufactured and shipped, or may be provided via the communication interface 1304, for example.

The above-described information processing device 1300 may execute the computer program to realize at least a part of the mild cognitive impairment prediction system according to the present disclosure.
More specifically, the information processing device 1300 may execute the computer program to execute each of the above-described steps, thereby realizing at least a part of the functions of the mild cognitive impairment prediction system according to the present disclosure. In this case, the mild cognitive impairment prediction system according to the present disclosure can be realized, for example, by a computer program executed on a general-purpose computer.

The information processing device 1300 described above is not limited to a physical device, and at least a part of the information processing device 1300 may be virtualized to become a virtual machine. In this case, the components constituting the information processing device 1300 may be realized as virtual devices or resources.
At least a portion of the mild cognitive impairment prediction system according to the present disclosure may be realized, for example, as a virtual system implemented on a generally available virtualization platform.

The above describes embodiments capable of realizing the mild cognitive impairment prediction system according to the present disclosure. Note that the mild cognitive impairment prediction system according to the present disclosure is not limited to the configurations described in the above embodiments. In other words, the mild cognitive impairment prediction system according to the present disclosure may also include modified or improved examples in which the configuration of each embodiment is appropriately changed or improved within the scope that a person skilled in the art can understand based on the present disclosure. Furthermore, the mild cognitive impairment prediction system according to the present disclosure may also include a system realized by appropriately combining each of the above embodiments.

1...Mild cognitive impairment assessment system, 2...Mild cognitive impairment prediction system, 10...Data input unit, 11...Attribute data input unit, 12...Sensor data input unit, 20...MCI assessment unit, 21...Activity data conversion unit, 22...Data integration unit, 23...Data cleansing unit, 24...Assessment model storage unit, 25...Factor information calculation unit, 26...Assessment information calculation unit, 27...Output unit, 30...MCI prediction unit, 31...Activity data conversion unit, 32...Data integration unit, 33...Data cleansing unit, 34...Prediction model storage unit, 35...Prediction factor information calculation unit, 36...Prediction information calculation unit, 37...Output unit, 40...Goal calculation unit, 41...Activity data input unit, 42...Data integration unit, 43...Data cleansing unit, 44...Simulated assessment factor information calculation unit, 45... Simulated assessment information calculation unit, 46...simulated assessment information confirmation unit, 47...output unit, 1000...mild cognitive impairment assessment system, 1001...data input unit, 1002...assessment model storage unit, 1003...MCI assessment information calculation unit, 1100...mild cognitive impairment prediction system, 1101...data input unit, 1102...prediction model storage unit, 1103...MCI prediction information calculation unit, 1200...mild cognitive impairment prevention system, 1201...data input unit, 1202...prediction model storage unit, 1203...MCI prediction information calculation unit, 1204...goal calculation unit, 1300...information processing device, 1301...processor, 1302...memory, 1303...storage, 1304...communication interface, 1305...input/output interface, 1306...storage medium

Claims (13)

a data input unit for inputting attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
a cerebral glucose metabolic rate estimation model storage unit that stores a cerebral glucose metabolic rate estimation model obtained by a model created by an analysis process including at least one of regression analysis and machine learning, with attribute data of a third party and sensor data acquired from a sensor device worn by the third party as explanatory variables, and with the cerebral glucose metabolic rate acquired from the third party as a target variable;
A cerebral glucose metabolic rate estimation system comprising: a cerebral glucose metabolic rate estimation unit that estimates the cerebral glucose metabolic rate of the subject based on the attribute data of the subject, the sensor data, and the cerebral glucose metabolic rate estimation model. a data input unit for inputting attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
a hippocampal volume estimation model storage unit that stores a hippocampal volume estimation model obtained by a model created by an analysis process including at least one of regression analysis and machine learning, with attribute data of a third party and sensor data acquired from a sensor device worn by the third party as explanatory variables, and with a hippocampal volume acquired from the third party as a target variable;
A hippocampal volume estimation system comprising: a hippocampal volume estimation unit that estimates the hippocampal volume of the subject based on the attribute data of the subject, the sensor data, and the hippocampal volume estimation model. a data input unit for inputting attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
a gray matter volume estimation model storage unit that stores a gray matter volume estimation model obtained by a model created by an analysis process including at least one of regression analysis and machine learning, with attribute data of a third party and sensor data acquired from a sensor device worn by the third party as explanatory variables, and with gray matter volume acquired from the third party as a response variable;
A gray matter volume estimation system comprising: a gray matter volume estimation unit that estimates the gray matter volume of the subject based on the attribute data of the subject, the sensor data, and the gray matter volume estimation model. a data input unit for inputting attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
an amyloid β accumulation amount estimation model storage unit that stores an amyloid β accumulation amount estimation model obtained by a model created by an analysis process including at least one of regression analysis and machine learning, with attribute data of a third party and sensor data acquired from a sensor device worn by the third party as explanatory variables, and with the amyloid β accumulation amount acquired from the third party as a response variable;
An amyloid beta accumulation estimation system comprising: an amyloid beta accumulation estimation unit that estimates the amyloid beta accumulation of the subject based on the attribute data of the subject, the sensor data, and the amyloid beta accumulation estimation model. a data input unit for inputting attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
an MMSE score estimation model storage unit that stores an MMSE score estimation model obtained by a model created by an analysis process including at least one of regression analysis and machine learning, with attribute data of a third party and sensor data acquired from a sensor device worn by the third party as explanatory variables, and an MMSE (Mini Mental State Examination) score acquired from the third party as a response variable;
An MMSE score estimation system comprising: an MMSE score estimation unit that estimates the subject's MMSE score based on the subject's attribute data, the sensor data, and the MMSE score estimation model. a data input unit for inputting attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
a prediction model storage unit that stores a first prediction model for predicting whether the subject corresponds to a healthy subject or a person with mild cognitive impairment, the first prediction model being created using attribute data of a third party including at least one of a healthy subject and a person with mild cognitive impairment and prediction factor information influencing the prediction of mild cognitive impairment over time obtained from the third party; and
and an MCI prediction information calculation unit that calculates prediction information for predicting whether the subject corresponds to a mild cognitive impairment at one or more points in a time series based on the attribute data of the subject, the sensor data of the subject, and the first prediction model,
The predictive factor information influencing the prediction of mild cognitive impairment is one or more of the following information: MMSE (Mini Mental State Examination) score, cerebral glucose metabolic rate, hippocampal volume, gray matter volume, and amyloid beta accumulation;
Mild cognitive impairment prediction system. The mild cognitive impairment prediction system according to claim 6, wherein the first prediction model is a model created by an analysis process including at least one of regression analysis and machine learning, with the attribute data of the third party and prediction factor information influencing the time-series prediction of mild cognitive impairment obtained from the third party as explanatory variables, and the categories of the healthy subject and the mild cognitive impairment subject as objective variables. The prediction model storage unit further stores a prediction factor information estimation model for calculating a prediction factor information estimate value affecting the prediction of the time-series mild cognitive impairment of the subject;
A prediction factor information calculation unit that calculates a prediction factor information estimation value that influences prediction of mild cognitive impairment of the subject based on the attribute data of the subject, the sensor data, and the prediction factor information estimation model,
The mild cognitive impairment prediction system according to claim 6 or 7, wherein the MCI prediction information calculation unit calculates prediction factor information capable of predicting whether the subject is a patient with mild cognitive impairment based on the prediction factor information estimate and the first prediction model. The mild cognitive impairment prediction system according to claim 8, wherein the predictive factor information estimation model is a model created by an analysis process including at least one of regression analysis and machine learning, in which the attribute data of the third party and the sensor data acquired from a sensor device worn by the third party are explanatory variables, and predictive factor information influencing the prediction of mild cognitive impairment in a time series acquired from the third party is an objective variable. a data input unit for inputting attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
a prediction model storage unit that stores a first prediction model for predicting whether the subject corresponds to a healthy subject or a person with mild cognitive impairment, the first prediction model being created using attribute data of a third party, including a healthy subject and a person with mild cognitive impairment, and prediction factor information influencing the prediction of mild cognitive impairment over time obtained from the third party; and
An MCI prediction information calculation unit that calculates prediction information for predicting whether the subject corresponds to a mild cognitive impairment at one or more points in a time series based on the attribute data of the subject, the sensor data of the subject, and the first prediction model;
A goal calculation unit that determines activity data representing the activity of the subject converted from the sensor data of the subject that can reduce the probability that the subject is predicted to be a person with mild cognitive impairment at one or more points in the time series,
The predictive factor information influencing the prediction of mild cognitive impairment is one or more of the following information: MMSE (Mini Mental State Examination) score, cerebral glucose metabolic rate, hippocampal volume, gray matter volume, and amyloid beta accumulation;
Mild cognitive impairment prevention system. a data input unit for inputting attribute data of a subject and activity data representing an activity of the subject;
a prediction model storage unit that stores a first prediction model for predicting whether the subject corresponds to a healthy subject or a person with mild cognitive impairment, the first prediction model being created using attribute data of a third party, including a healthy subject and a person with mild cognitive impairment, and prediction factor information influencing the prediction of mild cognitive impairment over time obtained from the third party; and
A mild cognitive impairment simulation system comprising: an MCI simulated judgment information calculation unit that calculates simulated judgment information for judging whether the subject is a person with mild cognitive impairment at one or more points in a time series based on the attribute data of the subject, the activity data of the subject, and the first prediction model. In the information processing device,
A process of inputting attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
A computer program that causes an information processing device to execute the following steps: a first prediction model for predicting whether the subject corresponds to either a healthy subject or a person with mild cognitive impairment, the first prediction model being created using attribute data of a third party including at least one of a healthy subject and a person with mild cognitive impairment, and prediction factor information influencing prediction of mild cognitive impairment over time obtained from the third party; and a process of calculating prediction information for predicting whether the subject corresponds to a person with mild cognitive impairment at one or more points in time based on the attribute data of the subject and the sensor data of the subject ,
The predictive factor information influencing the prediction of mild cognitive impairment is one or more of the following information: MMSE (Mini Mental State Examination) score, cerebral glucose metabolic rate, hippocampal volume, gray matter volume, and amyloid beta accumulation;
Computer program . In the information processing device,
A process of inputting attribute data of a subject and sensor data acquired from a sensor device worn by the subject;
A first prediction model for predicting whether the subject corresponds to either a healthy subject or a person with mild cognitive impairment, the first prediction model being created using attribute data of a third party, including healthy subjects and persons with mild cognitive impairment, and prediction factor information influencing prediction of mild cognitive impairment over time obtained from the third party; and a process for calculating prediction information for predicting whether the subject corresponds to a person with mild cognitive impairment at one or more points in time based on the attribute data of the subject and the sensor data;
determining activity data representing the activity of the subject, the activity data being converted from the sensor data of the subject, capable of reducing a probability that the subject is predicted to be a person with mild cognitive impairment at one or more points in the time series ;
The predictive factor information influencing the prediction of mild cognitive impairment is one or more of the following information: MMSE (Mini Mental State Examination) score, cerebral glucose metabolic rate, hippocampal volume, gray matter volume, and amyloid beta accumulation;
Computer program .

Images