JP7614598B2 - Vector field information generating device, state judgment support system, state prediction system, vector field information generating method and program - Google Patents

Description

The present invention relates to a vector field information generating device, a state determination support system , a state prediction system , a vector field information generating method, and a program.

Techniques have been proposed for determining the presence or absence of a disease based on the expression level of microRNA.
For example, the disease prevalence assessment device described in Patent Literature 1 acquires sample data including the expression levels of each biomarker including microRNA. The disease prevalence assessment device also includes a trained model for determining the presence or absence of each of a plurality of diseases. The disease prevalence assessment device then uses the sample data and the trained model to determine the presence or absence of each of the plurality of diseases.

International Publication No. 2018/079840

It would be preferable to be able to assess a subject's condition in more detail, rather than just determining whether or not they have a disease, such as determining their susceptibility to a certain disease.

The present invention provides a vector field information generating device, a condition determination support system, a condition prediction system , a vector field information generating method , and a program that can assist in determining not only whether or not a subject is suffering from a disease, but also in determining the subject's condition in more detail.

According to a first aspect of the present invention, a vector field information generating device comprises a biomarker amount data acquisition unit that acquires biomarker amount data indicating the amount of biomarker for each type of biomarker; a feature extraction unit that extracts two-dimensional or higher dimensional features as features of the biomarker amount data; a vector field information generation unit that acquires a plurality of different change amount vectors at the same point in feature space based on historical information of the features of a plurality of people , and generates vector field information that probabilistically indicates the change in the feature for each of the features; and a status information addition unit that generates status information-added vector field information in which status information indicating the status of a portion of the feature space in the vector field information is added to the portion , and status information-added vector field information in which status information for each of a plurality of types of cancer is added to one of the vector field information .
The vector field information generating unit may generate a vector field indicating a rate of change of the feature amount for each of the feature amounts.
The condition information adding unit may generate the condition-information-added vector field information to which condition information indicating a pre-illness state is added.

The feature extraction unit may extract one-dimensional features for each expected state, and the vector field information generation unit may generate the vector field information for each feature.

The feature extraction unit may extract the feature quantities of two or more dimensions, and the state information addition unit may generate the state information-added vector field information in which state information of multiple states is added to one vector field information.

According to a second aspect of the present invention, a state determination support system comprises any of the vector field information generation devices described above and a state determination support device , wherein the state determination support device comprises a second biomarker amount data acquisition unit that acquires the biomarker amount data of a subject, a second feature extraction unit that extracts features of the biomarker amount data of the subject using a feature extraction method similar to that of the feature extraction unit of the vector field information generation device, and a state determination support processing unit that indicates the positions of the features of the biomarker amount data of the subject in the feature space in the vector field information.

According to a third aspect of the present invention, a state prediction system comprises any of the vector field information generation devices described above and a state prediction device, wherein the state prediction device comprises a second biomarker amount data acquisition unit that acquires the biomarker amount data of a subject, a second feature extraction unit that extracts features of the biomarker amount data of the subject by a feature extraction method similar to that used by the feature extraction unit of the vector field information generation device, and a state prediction unit that reads a rate of change of a feature corresponding to the feature of the biomarker amount data of the subject from the state information-added vector field information, adds the rate of change to the feature to calculate a predicted value of the feature after a unit period has elapsed, and repeats the steps of reading the rate of change corresponding to the feature and calculating the predicted value of the feature after a unit period has elapsed by adding the rate of change to the feature, until a prediction target time is reached, thereby predicting the state of the subject.

According to a fourth aspect of the present invention, a vector field information generating method includes the steps of acquiring biomarker amount data indicating the amount of biomarker for each type of biomarker, extracting two-dimensional or higher dimensional features as features of the biomarker amount data, acquiring a plurality of different change amount vectors at the same point in feature space based on historical information of the features of a plurality of people , and generating vector field information probabilistically indicating the change amount of the feature for each of the features, and generating vector field information with status information in which status information indicating the status of a portion of the feature space in the vector field information is added to the portion , and generating vector field information with status information in which the status information of each of a plurality of types of cancer is added to one of the vector field information .

According to a fifth aspect of the present invention, the program causes a computer to execute the steps of acquiring biomarker amount data indicating the amount of biomarker for each type of biomarker, extracting two- or more dimensional features as features of the biomarker amount data, acquiring a plurality of different change vectors at the same point in feature space based on historical information of the features of a plurality of people , and generating vector field information that probabilistically indicates the change in the feature for each of the features, and generating vector field information with status information in which status information indicating the status of a portion of the feature space in the vector field information is added to the portion , and generating vector field information with status information in which status information for each of a plurality of types of cancer is added to one of the vector field information .

The present invention can help determine not only whether or not a subject is suffering from a disease, but also the subject's condition in more detail.

1 is a configuration diagram showing an example of a device configuration of a state determination support system according to an embodiment; FIG. 1 is a configuration diagram illustrating an example of a device configuration of a state prediction system according to an embodiment. FIG. 1 is a schematic block diagram illustrating an example of a functional configuration of a vector field information generating device according to an embodiment. FIG. 11 is a diagram illustrating an example of the configuration of GCM processing when GCM is applied to the vector field information generation device according to the embodiment. 1 is a schematic block diagram illustrating an example of a functional configuration of a state determination support device according to an embodiment. 1A to 1C are diagrams illustrating an example of display of one-dimensional state information-added vector field information by the state determination support device according to the embodiment. 1A to 1C are diagrams illustrating an example of display of two-dimensional state information-added vector field information by the state determination support device according to the embodiment. 1A to 1C are diagrams illustrating an example of display of three-dimensional state information-added vector field information by the state determination support device according to the embodiment. 11 is a diagram illustrating an example of a change vector of a feature amount calculated by a vector field information generation unit according to the embodiment. FIG. 11A to 11C are diagrams illustrating an example of interpolation of a change vector of a feature amount by a vector field information generation unit according to the embodiment. 13 is a flowchart illustrating an example of a processing procedure in which a vector field information generation unit according to the embodiment interpolates a change vector into vector field information. FIG. 11 is a diagram illustrating an example of plotting feature amounts in a feature amount space according to the embodiment. FIG. 11 is a diagram illustrating an example of display of vector field information according to the embodiment. FIG. 11 is a diagram illustrating an example of a transition of a feature amount according to the embodiment. 11 is a flowchart illustrating an example of a processing procedure in which the vector field information generating device according to the embodiment generates and updates history information of feature amounts of expression amount data of microRNA. 11 is a flowchart illustrating an example of a processing procedure in which the vector field information generating device according to the embodiment generates state information-added vector field information. 13 is a flowchart illustrating an example of a processing procedure in which the state determination support device according to the embodiment plots features of a subject in a feature space of state-information-attached vector field information. 1 is a schematic block diagram illustrating an example of a functional configuration of a state prediction device according to an embodiment. 1A to 1C are diagrams illustrating an example of display of one-dimensional state information-added vector field information by the state prediction device according to the embodiment. 1 is a diagram showing an example of display of two-dimensional state information-added vector field information by a state prediction device according to an embodiment; 1 is a diagram showing an example of display of three-dimensional state information-added vector field information by a state prediction device according to an embodiment; 11 is a flowchart illustrating an example of a processing procedure in which a condition prediction device according to the embodiment predicts a condition of a subject. FIG. 1 is a schematic block diagram illustrating a configuration of a computer according to at least one embodiment.

The following describes embodiments of the present invention, but the following embodiments do not limit the scope of the invention. Furthermore, not all of the combinations of features described in the embodiments are necessarily essential to the solution of the invention.
1 is a block diagram showing an example of a device configuration of a state determination support system according to an embodiment. In the configuration shown in FIG. 1, a state determination support system 100 includes a vector field information generating device 200 and a state determination support device 300.

The condition determination support system 100 generates and presents information for supporting the determination of the condition of a subject of condition determination based on biomarkers. The subject of condition determination may also be simply referred to as a subject.
In the following, an example will be described in which the state to be judged (the state for which the state judgment support system 100 supports the judgment) is a health state such as a healthy state (healthy state), a diseased state, or a pre-disease state. However, the state for which the state judgment support system 100 supports the judgment is not limited to a healthy state. The state judgment support system 100 can support the judgment of various states that are correlated with features extracted from the measurement data of biomarkers.

In the following, a case where microRNA (miRNA) is used as a biomarker will be described as an example. However, the biomarker used by the condition determination support system 100 is not limited to a specific one, and various biomarkers that are correlated with the condition of the subject to be determined can be used.
The expression amount of microRNA is used as an index value indicating the state of the subject of microRNA collection. Therefore, the feature amount of the expression amount of microRNA is also the feature amount of the state of the subject of microRNA collection. The subject of microRNA collection is also simply called the subject. The feature amount of the expression amount of microRNA of a certain person (subject or subject) is also called the feature amount of that person.

The vector field information generating device 200 generates state information-attached vector field information based on measurement data of the expression levels of microRNA.
The vector field information here is information indicating the relationship between the feature amount related to the expression level of microRNA and the change in the feature amount. Specifically, the vector field information is information indicating a vector field indicating the distribution of the feature amount change vector (vector indicating the amount of change) in the feature amount space (space of values that the feature amount can take). The feature amount change vector is also simply referred to as the change amount vector.

The state information-added vector field information referred to here is information in which state information indicating a predetermined health state is further added to the vector field information. Specifically, in the state information-added vector field information, state information indicating the health state of a portion of the feature space in the vector field information (a part of the feature space) is added.
The portion of the feature space here may be a point (a certain coordinate). For example, a certain point in the feature space may be added with state information such as "an example of a feature value one year before the onset of XX cancer."
Alternatively, the portion of the feature space here may be a partial region of the feature space. In this case, the health state indicated by the state information may be, but is not limited to, a healthy state, a diseased state, or a pre-disease state. Adding state information to a portion of the feature space of the state-information-added vector field information is also referred to as state labeling.

The method of expressing the state information is not limited to a specific method, and may be, for example, a word, a color, a graphic name, or a combination of these.
For example, when status information is added to points in the feature space, the status information may be indicated by the color of the point (plot), such that points showing examples of features for cancer A are shown in blue and points showing examples of features for cancer B are shown in green. In this case, for example, blue points are plotted together in a part of the feature space, so that it can be shown that the vicinity of the group of blue points is highly likely to be cancer A.
In addition, status information may be shown by the shape of the plot in addition to or instead of the color of the points (plots), such as a point showing an example of a feature for cancer A being shown as a circle (◯) and a point showing an example of a feature for cancer B being shown as a square (□).

The change vector indicates at least the direction of change. If the feature space is one-dimensional, the change vector indicates whether a feature is increasing or decreasing. If the feature space is two or more dimensions, the change vector indicates the rate of change of each feature (the feature for each coordinate axis in the feature space).

According to the condition determination support system 100, the tendency of change in the subject's health condition can be determined by referring to the direction of the change vector. For example, features extracted from the measurement data of the expression level of the subject's microRNA are plotted in a feature space in the state-information-attached vector field information generated by the vector field information generating device 200. Using the state-information-attached vector field information in which the features are plotted, it can be determined that the subject's condition is, for example, relatively likely to become a state of the area reached by tracing the change vector from the plotted point (a state indicated by the state information of that area).
A certain region may be reached by tracing one change vector from the plot of the subject's features, or a certain region may be reached by tracing two or more change vectors from the plot of the subject's features.

For example, consider a case where regions for each type of cancer, such as cancer A state, cancer B state, etc., are set in the feature space in the state-information-attached vector field information. For example, when the region for cancer A state is reached by tracing the change vector from the plot of the subject's features, it can be determined that the subject is in a state where he or she is relatively susceptible to developing cancer A. Based on this determination result, the subject can take measures to prevent developing cancer A.
Alternatively, when the plot of the subject's features falls within a certain region, the subject's state can be determined to be the state of that region.

Alternatively, if status information is attached to a point in the feature space, the subject's status may be determined to be close to the state indicated by the status information attached to a point in the vicinity of the plot of the subject's features.
The vector field information generating device 200 is configured using a computer such as a personal computer (PC) or a workstation.

The state determination support device 300 indicates the position of the subject's features in the feature space in the state information-added vector field information. Specifically, the state determination support device 300 acquires measurement data of the subject's microRNA expression levels and extracts (calculates) features (feature amounts of the microRNA expression levels). The state determination support device 300 then plots the features extracted from the measurement data of the subject's microRNA expression levels in the feature space in the state information-added vector field information generated by the vector field information generating device 200.

In this way, the condition determination support device 300 supports the determination of the subject's condition. As described above, by tracing the change amount vector from the plot of the subject's feature amount, it is possible to estimate the tendency of change in the subject's condition.
The state determination support device 300 is configured using a computer such as a personal computer (PC) or a workstation.

The vector field information generating device 200 and the state determination support device 300 may be configured as a single device. In particular, when the vector field information generating device 200 frequently generates or updates vector field information with state information, the vector field information generating device 200 and the state determination support device 300 can be integrated to share data such as history information of the expression amount of microRNA, and also share functions such as a function for extracting features of the expression amount of microRNA, thereby making it possible to make the state determination support system 100 more compact.
On the other hand, when the vector field information with state information has been determined and does not need to be updated, the state determination support device 300 can be configured as a device separate from the vector field information generation device 200, thereby making the state determination support device 300 more compact.

State prediction may be performed as state determination using state-information-added vector field information.
Fig. 2 is a configuration diagram showing an example of the device configuration of a state prediction device system according to an embodiment. In the configuration shown in Fig. 2, a state prediction system 100b includes a vector field information generation device 200 and a state prediction device 300b.

The state prediction device 300b performs state prediction as assistance for state determination. In other respects, the state prediction system 100b is similar to the state determination assistance system 100.
The vector field information may be information indicating the amount of change in a feature per unit period (unit time) for each feature (value). The vector field in this case is also referred to as a velocity field, and the vector field information in this case is also referred to as velocity field information. The vector field information with state information in this case is also referred to as velocity field information with state information. The amount of change in a feature per unit period (e.g., one year) is also referred to as the rate of change of the feature. The rate of change of a feature is indicated by a vector in the coordinate system of the feature space. The vector of the rate of change of a feature is also referred to as a rate of change vector or a velocity vector.

Since the vector field is configured as a velocity field, the point reached by tracing one change vector from the plot of the subject's features indicates the predicted value of the subject's features after a unit period has passed. Furthermore, the point reached by tracing the feature vector indicates the predicted value of the subject's features after a further unit period has passed. The predicted state of the subject is then indicated by the state information attached to the area that contains the predicted value.

For example, consider a case where regions for each type of cancer, such as Cancer A state, Cancer B state, etc., are set in the feature space. For example, if the predicted value of the feature is included in the region for Cancer A state, a prediction is shown that the subject will develop Cancer A at the time of prediction (after the elapse of a period equal to the unit period multiplied by the number of change vectors traced). Based on this prediction, the subject can take measures to prevent themselves from developing Cancer A in the future.

Figure 3 is a schematic block diagram showing an example of the functional configuration of the vector field information generating device 200. In the configuration shown in Figure 3, the vector field information generating device 200 includes a first communication unit 210, a first display unit 220, a first operation input unit 230, a first memory unit 270, and a first control unit 280. The first control unit 280 includes a first expression amount data acquisition unit 281, a first feature amount extraction unit 282, a class classification unit 283, a machine learning control unit 284, a history information registration unit 285, a vector field information generation unit 286, and a state information addition unit 287.

The first communication unit 210 communicates with other devices. For example, the first communication unit 210 receives measurement data of the expression level of microRNA from a microRNA analysis system. The first communication unit 210 also communicates with the condition determination support device 300 to transmit vector field information with condition information.
The first display unit 220 has a display screen, such as a liquid crystal panel or an LED (Light Emitting Diode) panel, and displays various images. For example, the first display unit 220 displays the vector field information with state information generated by the vector field information generating device 200. The first display unit 220 also displays an operation screen that allows the user of the vector field information generating device 200 to give various instructions, such as an instruction to obtain measurement data of the expression level of microRNA and an instruction to generate vector field information with state information.
The first operation input unit 230 includes input devices such as a keyboard and a mouse, and receives user operations. For example, the first operation input unit 230 receives user operations such as an instruction to obtain measurement data of the expression level of microRNA, an instruction to generate vector field information with state information, and the like.

The first storage unit 270 stores various data. For example, the first storage unit 270 stores history information of the measurement data of the expression level of microRNA for each subject. The first storage unit 270 also stores history information of the feature amount extracted from the measurement data of the expression level of microRNA for each subject. Alternatively, the first storage unit 270 may store only one of the history information of the measurement data of the expression level of microRNA or the history information of the feature amount.
Moreover, the first storage unit 270 stores vector field information with state information.
The first storage unit 270 is configured using a storage device provided in the vector field information generation device 200.

The first control unit 280 performs various processes by controlling each unit of the vector field information generation device 200. The function of the first control unit 280 is executed, for example, by a CPU (Central Processing Unit) included in the vector field information generation device 200 reading out a program from the first storage unit 270 and executing it.
The first expression level data acquisition unit 281 compiles the measurement data of the expression level of the microRNA into expression level data. The expression level data here is data indicating the expression level for each type of microRNA. For example, it is said that there are more than about 2500 types of human microRNA, and when the expression levels of 2500 types of microRNA are analyzed, the expression level data is represented by data of a 2500-dimensional vector.

The expression level data is an example of biomarker amount data, which indicates the amount of each type of biomarker.
The first expression amount data acquiring section 281 corresponds to an example of a first biomarker amount data acquiring section and an example of a biomarker amount data acquiring section.
The first storage unit 270 may store the history information of the measurement data of the expression levels of microRNA in the format of history information of expression level data.

The first feature amount extraction unit 282 extracts features of the expression amount data. The first feature amount extraction unit 282 is an example of a feature amount extraction unit.
The first feature extraction unit 282 may extract one-dimensional features for each condition (e.g., each disease name) of the subject. Alternatively, the first feature extraction unit 282 may extract two or more dimensional features. The dimension of the feature extracted by the first feature extraction unit 282 is the dimension of the vector field information with condition information.

The method for extracting features from expression level data is not limited to a specific method. For example, known dimensionality reduction methods such as Principal Component Analysis (PCA), neural networks such as AE (Autoencoder) or VAE (Variational Autoencoder), or Self Organizing Map (SOM) may be used.

In principal component analysis, for example, the direction in which the variance from the center of gravity of the entire multidimensional data is greatest is calculated as the first principal component, and the direction perpendicular to the first principal component in which the variance is greatest is calculated as the second principal component, thereby compressing multidimensional data into two dimensions.
When the first feature amount extraction unit 282 uses principal component analysis, the expression amount data can be compressed to N dimensions by calculating the first component to the Nth component (N is a positive integer) of the expression amount data.

AE is a model that consists of an encoder and decoder that are hierarchical neural networks of an identity mapping that outputs data from the output layer that is the same as the input data, and it is believed that the features of the input data appear in the intermediate layer (latent variables).
When the first feature amount extraction unit 282 uses AE, the expression amount data can be compressed to N dimensions by making the intermediate layer N-dimensional.

VAE is a model that introduces a probability distribution into the intermediate layer (the latent variable part) of AE. In VAE, dimensionality is reduced by approximating input data with a Gaussian distribution.
Even when the first feature extraction unit 282 uses VAE, the expression amount data can be compressed to N dimensions by making the latent variables N-dimensional.

SOM maps high-dimensional data to a low-dimensional space such as one or two dimensions. The destination space is called the SOM map. By creating a random map in which data that is closer in the current space (the source high-dimensional data) is closer in the SOM map (the destination low-dimensional space), clustering (grouping) can be performed without a teacher.
When the first feature amount extraction unit 282 uses SOM, the expression amount data can be compressed to N dimensions by making the SOM map N-dimensional.
As a feature extraction method used by the first feature extraction unit 282, various dimensionality compression methods can be used in which, when the expression amount data of the source (input to feature extraction) is continuous, the features of the destination (output of feature extraction) are also continuous.

Figure 4 is a diagram showing an example of the configuration of GCM processing when GCM is applied to the vector field information generating device 200. In the example of Figure 4, the first control unit 280 that executes the processing of GCM400 includes an encoder 411, a first class decoder 412-1 to an Nth class decoder 412-N, a first multiplier 413-1 to an Nth multiplier 413-N, a first average calculation unit 414-1 to an Nth average calculation unit 414-N, and an Argmax calculation unit 415. Here, N is a positive integer indicating the number of classes in the classification.

The first class decoder 412-1 through the Nth class decoder 412-N are collectively referred to as decoder 412. The first multiplier 413-1 through the Nth multiplier 413-N are collectively referred to as multiplier 413. The first average calculation unit 414-1 through the Nth average calculation unit 414-N are collectively referred to as average calculation unit 414.

The encoder 411 receives expression amount data and extracts features of the input data. The encoder 411 is an example of the first feature amount extraction unit 282.
A decoder 412 is provided for each class, and reconstructs the feature amount calculated by the encoder 411 into data of the same dimension as the input data. This data indicates the weight of how likely each part of the input data (expression amount data) is to the class of interest.

A multiplier 413 is provided for each class and multiplies the input data by the data calculated by the decoder 412. This results in data that is weighted for each part of the input data according to its contribution to class classification.

An average calculation unit 414 is provided for each class, and calculates, for each class, the average of the data calculated by the multiplier 413. The average calculated by the average calculation unit 414 is used as an evaluation value (class score) in class classification.
The Argmax calculation unit 415 compares the class scores calculated for each class by the average calculation unit 414 and determines the class with the largest class score. In this way, the Argmax calculation unit 415 classifies the input data into classes.
The combination of the first average calculation unit 414-1 to the N-th average calculation unit 414-N and the Argmax calculation unit 415 corresponds to the classifying unit 283 examples.

The machine learning control unit 284 controls the learning of the first control unit 280. For example, the first feature extraction unit 282 and the class classification unit 283 may be configured using a computational model such as a neural network. The machine learning control unit 284 may then cause the first feature extraction unit 282 and the class classification unit 283 to learn and determine parameter values of the computational model.

4, the encoder 411 and the decoder 412 may be configured using a computational model such as a neural network. The machine learning control unit 284 may then cause the encoder 411 and the decoder 412 to learn and determine parameter values of the computational model.
Known techniques can be used for the training of the first feature extraction unit 282 and the class classification unit 283.

The history information registration section 285 stores, for each subject, in the first storage section 270, history information on the expression amount data, the feature amount of the expression amount data, and the class classification results of the expression amount data.
The vector field information generating unit 286 generates vector field information indicating the relationship between the feature amounts and changes in the feature amounts, based on the history information of the feature amounts.

As described above, the vector field information is information indicating a vector field that indicates the distribution of the change vector of the feature in the feature space. The vector field information generation unit 286 calculates the change (change vector) of the feature for the same subject from the feature history information. The vector field information generation unit 286 then links the calculated change to the feature before the change and plots it in the feature space of the vector field information. Specifically, the vector field information generation unit 286 links the change vector indicating the calculated change to the coordinates of the feature before the change in the feature space (plotting of the feature before the change in the feature space).

When the first feature extraction unit 282 extracts one-dimensional features for each state of the subject, the vector field information generation unit 286 generates one-dimensional vector field information for each feature. Alternatively, when the first feature extraction unit 282 extracts two or more dimensional features, the vector field information generation unit 286 generates vector field information of the same dimension as the feature.

The state information adding unit 287 generates vector field information with state information. As described above, the vector field information with state information is information in which state information is added to the feature space portion of the vector field information generated by the vector field information generating unit 286. When the first feature extraction unit 282 extracts features of two or more dimensions, the state information adding unit 287 may generate vector field information with state information in which state information of multiple states is added to one vector field information.

Figure 5 is a schematic block diagram showing an example of the functional configuration of the state judgment support device 300. In the configuration shown in Figure 5, the state judgment support device 300 includes a second communication unit 310, a second display unit 320, a second operation input unit 330, a second memory unit 370, and a second control unit 380. The second control unit 380 includes a second expression amount data acquisition unit 381, a second feature amount extraction unit 382, and a state judgment support processing unit 383.

The second communication unit 310 communicates with other devices. For example, the second communication unit 310 receives measurement data on the expression level of the subject's microRNA from a microRNA analysis system. The second communication unit 310 also communicates with the vector field information generating device 200 to receive vector field information with state information.

In addition, when the first memory unit 270 of the first operation input unit 230 stores historical information of the expression amount data of the subject, the second communication unit 310 may receive the historical information of the expression amount data of the subject from the first operation input unit 230. Using this historical information, the state judgment support device 300 may plot not only the current feature amount of the subject but also past feature amounts on the state information-added vector field information.

The second display unit 320 includes a display screen such as a liquid crystal panel or an LED (Light Emitting Diode) panel, and displays various images. For example, the second display unit 320 displays the state information-added vector field information including a plot of the subject's features.
The second operation input unit 330 includes input devices such as a keyboard and a mouse, and receives a user operation. For example, the second operation input unit 330 receives a user operation to instruct the subject's features to be plotted and displayed in the state information-added vector field information.

The second storage unit 370 stores various data. For example, the second storage unit 370 stores the vector field information with state information obtained from the vector field information generating device 200.
The second storage unit 370 may also be configured to store history information of the measurement data of the expression level of the microRNA of the subject. Using this history information, the condition determination support device 300 may plot not only the current feature amount of the subject but also the past feature amount on the vector field information with condition information. When the condition determination support device 300 supports the condition determination of each of multiple subjects, the second storage unit 370 may store history information of the measurement data of the expression level of the microRNA for each subject.

The second control unit 380 performs various processes by controlling each unit of the state determination support device 300. The functions of the second control unit 380 are performed, for example, by a CPU included in the state determination support device 300 reading out a program from the second storage unit 370 and executing the program.
The second expression level data acquisition unit 381 compiles the measurement data of the expression levels of the subject's microRNA into expression level data.
The second expression amount data acquiring section 381 corresponds to an example of a second biomarker amount data acquiring section and an example of a biomarker amount data acquiring section.

The second feature extraction unit 382 extracts features of the expression amount data of the subject by the same feature extraction method as the first feature extraction unit 282. This is to extract features to be plotted on the vector field information with state information generated by the vector field information generating device 200.
The second feature amount extraction unit 382 is an example of a feature amount extraction unit.
The state determination support processing unit 383 plots the feature amount of the subject in the state information-attached vector field information. As described above, the state determination support processing unit 383 may plot the past feature amount of the subject in addition to the current feature amount of the subject in the state information-attached vector field information.

Fig. 6 is a diagram showing an example of one-dimensional vector field information with state information displayed by the state judgment support device 300. In the example of Fig. 6, the feature amount space is shown in a one-dimensional coordinate space of variable _x1 . In this one-dimensional coordinate space, a "healthy" region (region of a healthy state) and a "xx cancer" region (region of a xx cancer state) are shown.

The second display unit 320 may also display the pre-disease region for cancer in addition to the healthy state region and the cancer state region. Pre-disease (state) here refers to a state in which the patient is not afflicted with a particular disease, but has some subjective symptoms or shows abnormal values when tested, and is at high risk of contracting the disease. For example, fatty liver is an afflicted state for the disease fatty liver, but is a pre-disease state for the disease liver cancer.

When the classifying unit 283 of the vector field information generating device 200 classifies the features, in addition to the classes of healthy states and diseased states (e.g., the class of XX cancer), a class of pre-disease states (e.g., a pre-disease class for XX cancer) can be set, making it possible to set a pre-disease region in the vector field information with condition information. The second display unit 320 may also be configured to clearly indicate the pre-disease region.

In the example of Figure 6, the subject's features are plotted in a vector field. Because the subject's features are plotted in the area between "healthy" and "XX cancer," it can be determined that the subject is not suffering from XX cancer, but cannot be said to be in a healthy state either. Furthermore, because tracing the change vector from the plot of the subject's features reaches the "XX cancer" area, it can be determined that the subject is in a state where he or she is relatively susceptible to developing XX cancer.

The subject can refer to the results of this assessment and take preventive measures to avoid contracting cancer.
When there are multiple states to be determined, such as when the state determination support device 300 supports state determination for multiple types of cancer, the second display unit 320 displays, for each state, one-dimensional vector field information with state information, as illustrated in Figure 6, and a plot of the transition of the subject's features.

Fig. 7 is a diagram showing an example of display of two-dimensional vector field information with state information by the state determination support device 300. In the example of Fig. 7, the feature amount space is shown in a two-dimensional orthogonal coordinate space of variables _x1 and _x2 . In this two-dimensional coordinate space, a "healthy" region and regions of each type of cancer from "cancer A" to "cancer D" are shown.

In the example of FIG. 7, the second display unit 320 may also display pre-disease areas in addition to the areas between the "healthy" area and each cancer area. In that case, the second display unit 320 may display the pre-disease area for each cancer type. Alternatively, the second display unit 320 may display an area that combines the pre-disease areas for each cancer type as a pre-disease area.

In the example of FIG. 7, the subject's features are also plotted in a vector field. Since the subject's features are plotted in the area between "healthy" and each cancer, it can be determined that the subject is not suffering from these cancers, but is not in a healthy state either. Furthermore, since tracing the change vector from the plot of the subject's features reaches the area of "cancer C," it can be determined that the subject is relatively susceptible to cancer C.
The subject can refer to the result of this assessment and take preventive measures to avoid contracting C cancer.

Fig. 8 is a diagram showing an example of three-dimensional vector field information with state information displayed by the state judgment support device 300. In the example of Fig. 8, the feature amount space is shown in a three-dimensional orthogonal coordinate space of variables _x1 , _x2, and _x3 . In this three-dimensional coordinate space, a "healthy" region and regions of each type of cancer from "cancer A" to "cancer D" are shown.

In the example of FIG. 8, the second display unit 320 may also display the pre-disease region. In that case, the second display unit 320 may display the pre-disease region for each cancer type. Alternatively, the second display unit 320 may display an area that combines the pre-disease regions for each cancer type as the pre-disease region.

In the example of Figure 8, the subject's features are also plotted in a vector field. Since the subject's features are plotted in the area between "healthy" and each cancer, it can be determined that the subject is not suffering from these cancers, but is not in a healthy state either. Furthermore, since tracing the change vector from the plot of the subject's features reaches the area of "cancer D," it can be determined that the subject is relatively susceptible to cancer D.
The subject can refer to the result of this assessment and take preventive measures to avoid contracting D cancer.

FIG. 9 is a diagram showing an example of a change amount vector of a feature amount calculated by the vector field information generating unit 286. As shown in FIG.
Fig. 9 shows an example of a feature change vector calculated by the vector field information generating unit 286 for the feature space of the example of Fig. 7. As described above, the vector field information generating unit 286 calculates the change in the feature of the same subject as a change vector from the feature history information. Then, the vector field information generating unit 286 links the calculated change to the feature before the change and plots it in the feature space of the vector field information.

By aggregating data from multiple people, it is possible to obtain multiple different actual measurement vectors (change vectors based on actual measurements) at the same point in the feature space. In this case, the vector field information generating unit 286 may uniquely calculate the change in the feature by, for example, finding the average vector of the multiple actual measurement vectors obtained. Alternatively, the vector field information generating unit 286 may calculate the change in the feature probabilistically, or may set it according to conditions, such as by age group.

When the vector field information generating unit 286 probabilistically calculates the amount of change in the feature quantity, the state assessment support processing unit 383 may determine the amount of change in each case probabilistically using random numbers, or may calculate the average or expected value. Alternatively, the second display unit 320 may display all of the multiple possibilities.

FIG. 10 is a diagram showing an example of the interpolation of the amount of change in the feature amount by the vector field information generating unit 286. In FIG.
Fig. 10 shows an example of the case where the amount of change is interpolated from the example of Fig. 9. The vector field information generating unit 286 detects three points _p1 , _p2 , and _p3 closest to the point P to be interpolated and determines them as reference points (points referenced for interpolation). The reference points are preferably actual measurement points, but may be interpolated points.

The actual measurement point here is a point to which a change vector due to an actual measurement value is associated. The change vector due to an actual measurement value here is a change (change vector) in a feature amount calculated by the vector field information generating unit 286 from history information of the feature amount.
The interpolation point here is a point at which the vector field information generating unit 286 interpolates the change amount vector (plot of the feature amount in the feature amount space).

FIG. 11 is a flowchart showing an example of a processing procedure in which the vector field information generating unit 286 interpolates a change amount vector in vector field information.
In the process of FIG. 11, the vector field information generator 286 determines a reference point (step S1).
Here, a case will be described where the feature space is two-dimensional, as in the example of Fig. 10. The coordinates of a point P to be interpolated are denoted as ( _xp , _yp ), and the change vector _vp at the point P is denoted as ( _vpx , _vpy ).

The vector field information generator 286 detects N points (N is an integer N≧2) closest to the point P to be interpolated and sets them as reference points. That is, the vector field information generator 286 determines, as reference points, the N points closest to the point P to be interpolated among the points to which the change amount vectors are linked.
N reference points are denoted as _p1 , _p2 , ..., _pN , and the coordinates of these reference points are denoted as _p1 = ( _xp1 , _yp1 ), _p2 = ( _xp2 , _yp2 ), ..., _pN = ( _xpN , _ypN ). Also, the change vectors at the reference points _p1 , _p2 , ..., _pN are denoted as _vp1 = ( _vxp1 , _vyp1 ), _vp2 = ( _vxp2 , _vyp2 ), ..., _vpN = ( _vxpN , _vypN ), respectively.

Next, the vector field information generator 286 calculates the distance from the point P to be interpolated to each of the reference points p ₁ , p ₂ , . . . , p _N (step S2).
The distances from the point P to be interpolated to the reference points p ₁ , p ₂ , . . . , p _N are denoted as d ₁ , d ₂ , . . . , d _N, respectively.

Next, the vector field information generating unit 286 calculates a change amount vector at the point to be interpolated, based on the change amount vector at the reference point and the distance from the point to be interpolated to the reference point (step S3).
For example, the vector field information generator 286 calculates _vpx based on equation (1).

Moreover, the vector field information generator 286 calculates v _py based on equation (2).

In the examples of formulas (1) and (2), the vector field information generator 286 calculates a weighted average of the change vector at the reference point with a weight inversely proportional to the distance from the point to be interpolated to the reference point. As a result, the vector field information generator 286 calculates the change vector at the point to be interpolated by weighting the reference point closer to the point to be interpolated.
In addition, the examples of formulas (1) and (2) show an example in which the feature space is two-dimensional (the number of features is two). Even if the feature space is not two-dimensional, the vector field information generating unit 286 may calculate a weighted average for each coordinate axis in the same manner as in the examples of formulas (1) and (2).

Next, the vector field information generator 286 links the calculated change amount vector _vp to the coordinates ( _xp , _yp ) of the point P to be interpolated in the feature amount space of the vector field information (step S4).
After step S4, the vector field information generator 286 ends the process of FIG.

In this way, it is possible to interpolate the change vector of any point in the feature space of the vector field information based on the change vectors at N reference points p ₁ , p ₂ , ..., p _N. By determining the change at any point in the feature space, a vector field can be obtained.
However, the method of interpolating the change amount is not limited to the above-mentioned method, and various other methods can be used.
Furthermore, the change amount vectors at the four corners of the rectangular feature space shown in FIG. 10 may be set to zero vectors as singular points.

FIG. 12 is a diagram showing an example of plotting feature amounts in a feature amount space.
In the example of Fig. 12, the feature values are plotted roughly together for each state of the subject (here, each type of cancer). From this, it is expected that it will be possible to set regions for each state of the subject as in the example of Fig. 9, and also to set a vector field for each of the change vectors of the feature values toward one of the states.
Even when the second display unit 320 displays vector field information with status information as in the example of Fig. 7 or Fig. 9, a plot of the feature amount may be displayed showing the status of the subject, as described with reference to Fig. 12. By referring to this plot, it can be confirmed whether the area for each status is appropriately set.

FIG. 13 is a diagram showing an example of displaying vector field information.
In FIG. 13 , the feature values of a subject suffering from colorectal cancer for the period from one year to less than two years before the cancer diagnosis, the period less than one year before the cancer diagnosis, after the cancer diagnosis, and after treatment, as well as the feature values of another subject who is not a patient (a healthy subject) are plotted in feature space.

13, the change vector is calculated using the history information of the same subject (data at different times for the same subject) and displayed in the feature space. For points where the change vector could not be obtained from the actual measurement data, the change vector is interpolated using the change vector obtained from the actual measurement data.
This resulted in the vector field illustrated in Fig. 13. In the example of Fig. 13, the feature amount of the healthy state ("healthy subject") and the feature amount of the colorectal cancer state ("post-cancer diagnosis") are plotted roughly together, and a region for each state can be set.

FIG. 14 is a diagram illustrating an example of transition of a feature amount.
In Fig. 14, in the example of the feature amounts in Fig. 13, a healthy state region is set so as to roughly surround the plot of the feature amounts for the healthy state ("healthy subject"), and a colon cancer state region is set so as to roughly surround the plot of the feature amounts for the colon cancer state ("post-cancer diagnosis").
In addition, lines connect the feature quantities of the same subject, and arrows show examples of changes in state quantities when going from a healthy state to a colorectal cancer state. When recovering from a colorectal cancer state, dotted arrows show examples of changes in state quantities.

The feature values of subjects with colorectal cancer gradually approach the colorectal cancer range as time progresses, such as the period between one year and two years before being diagnosed with cancer, and the period less than one year before being diagnosed with cancer. By plotting the feature values, it is also possible to estimate the approximate stage at which a subject is in the range from a healthy state to a state with colorectal cancer.

When the second display unit 320 plots and displays the subject's features in the vector field information with condition information, it may also display an example of a path from a healthy state to a diseased state, as illustrated by the arrow in FIG. 14. By determining which position on the illustrated path the subject's features correspond to, it is possible to use this as a reference for estimating at which stage the subject's condition is, from a healthy state to a diseased state.

Furthermore, sample plots may be displayed, such as the state quantity plot for the period from one year to less than two years before a cancer diagnosis, and the state quantity plot for the period less than one year before a cancer diagnosis, in FIG. 14. By comparing the position of the subject's feature quantity plot with the position of the sample plot, it is possible to estimate at what stage the subject's condition is, from a healthy state to a diseased state.

Figure 15 is a flowchart showing an example of a processing procedure in which the vector field information generating device 200 generates and updates historical information on the features of the expression amount data of the microRNA. The vector field information generating device 200 repeatedly performs the process of Figure 15 according to instructions from, for example, a user operation, and accumulates historical information on the features of the measurement data of the expression amount of the microRNA.

15, the first expression level data acquisition unit 281 acquires expression level data (step S11). Specifically, the first expression level data acquisition unit 281 organizes the measurement results of the expression levels of the microRNA into the format of expression level data.
Next, the first feature amount extraction unit 282 calculates (extracts) features from the expression amount data (step S12).

The classification unit 283 classifies the feature calculated by the first feature extraction unit 282 (step S13). This classification corresponds to classifying the health state of the subject.
Then, the history information registration unit 285 stores the feature amount (value) and class as the subject's history in the first storage unit 270 (step S14). If there is history information for the subject, the history information registration unit 285 adds data to the history information. On the other hand, if there is no history information for the subject, the history information registration unit 285 generates new history information for the subject and stores it in the first storage unit 270.
The history information registration unit 285 may store expression amount data in addition to the feature amount in the first storage unit 270 .
After step S14, the vector field information generation device 200 ends the process in FIG.

16 is a flowchart showing an example of a processing procedure for generating state-information-added vector field information by the vector field information generation device 200. The vector field information generation device 200 performs the processing of FIG. 16 after the amount of feature history information accumulated in the processing of FIG. 15 exceeds a predetermined standard.
In the process of FIG. 16, the vector field information generation unit 286 calculates the amount of change based on the history information of the feature amount stored in the first storage unit 270, and generates state-information-added vector field information (step S21).

Then, the state information addition unit 287 sets areas for each class in the vector field information with state information according to the class classification by the class classification unit 283 (for example, based on a class classification model), and stores the vector field information with state information after the classes have been set in the first memory unit 270 (step S22).
After step S22, the vector field information generation device 200 ends the process in FIG.
Note that the vector field information generating device 200 may not perform feature extraction (step S13) and class classification (step S14) at the processing stage of FIG. 15, but may perform feature extraction and class classification during the processing of FIG. 16.

FIG. 17 is a flowchart showing an example of a processing procedure in which the state determination support device 300 plots the feature amounts of the subject in the feature amount space of the state-information-attached vector field information.
17, the second expression level data acquisition unit 381 acquires expression level data of the subject (step S31). Specifically, the second expression level data acquisition unit 381 organizes the measurement results of the expression levels of the microRNA of the subject into the format of expression level data.

Next, the first feature extraction unit 282 calculates (extracts) features from the expression amount data of the subject (step S32).
Next, the state assessment support processing unit 383 plots the feature amount of the subject in vector field information with state information (step S33).
After step S33, state determination support device 300 ends the process of FIG.

As described above, the first expression amount data acquisition unit 281 acquires expression amount data indicating the amount of microRNA for each type of microRNA. The first feature extraction unit 282 extracts features from the expression amount data. The vector field information generation unit 286 generates vector field information based on feature history information. The vector field information is information indicating the relationship between features and changes in the features. The state information addition unit 287 generates vector field information with state information. The vector field information with state information is information in which state information indicating the state of a portion of the feature space in the vector field information is added to the portion.

By using the vector field information with status information generated by the vector field information generating device 200, the feature (current value of the feature) extracted from the measurement data of the expression level of the microRNA of the subject can be plotted in the vector field information with status information and used to determine the state of the subject. For example, when status information is added to an area in the feature space, the state of the subject can be determined as the state indicated by the area in which the feature of the subject is plotted. In addition, the state indicated by the area reached by tracing the feature vector from the plot of the feature of the subject can be determined as the state to which the subject is relatively likely to be.
Alternatively, when state information is added to a point in the feature space, the state of the subject can be determined to be close to the state indicated by the state information added to a point in the vicinity of the point where the subject's feature is plotted. Also, the state indicated by the state information added to a point in the vicinity of the point reached by tracing the feature vector from the plot of the subject's feature can be determined to be the state to which the subject is relatively likely to be.
In this way, the vector field information generating device 200 can assist in determining the subject's condition.

The first feature amount extracting unit 282 extracts a one-dimensional feature amount for each state of the subject. The vector field information generating unit 286 generates vector field information for each feature amount.
According to the vector field information generating device 200, the subject's condition can be displayed in the vector field information with condition information for each condition (for example, each type of illness) as in the example of Fig. 6. Since the subject's condition can be displayed for each individual condition, the subject's condition can be displayed in an easy-to-read manner.
As mentioned above, it can also indicate a pre-illness state.

Furthermore, the first feature amount extraction unit 282 extracts two or more dimensional feature amounts. The state information addition unit 287 generates state information-added vector field information in which state information of a plurality of states is added to one piece of vector field information.
The vector field information generating device 200 can provide vector field information with state information common to a plurality of states. This allows a person referring to the vector field information with state information, such as a subject, to understand a plurality of states of the subject by referring to one vector field information with state information.

The second expression data acquisition unit 381 acquires expression data indicating the amount of microRNA for each type of microRNA for the subject. The second feature extraction unit 382 extracts features from the expression data. The state determination support processing unit 383 indicates the position of the features of the expression data of the subject in the feature space in the state information-added vector field information.
According to the condition determination support device 300, the condition of the subject can be presented in the feature space of the expression amount data. For example, when condition information is added to an area in the feature space, the condition of the subject can be determined as the condition indicated by the area in which the subject's features are plotted. In addition, the condition indicated by the area reached by tracing the feature vector from the plot of the subject's features can be determined as the condition to which the subject is relatively likely to be.
Alternatively, when state information is added to a point in the feature space, the state of the subject can be determined to be close to the state indicated by the state information added to a point in the vicinity of the point where the subject's feature is plotted. Also, the state indicated by the state information added to a point in the vicinity of the point reached by tracing the feature vector from the plot of the subject's feature can be determined to be the state to which the subject is relatively likely to be.
In this manner, the condition determination assistance device 300 can assist in determining the subject's condition.

As described with reference to Fig. 2, state prediction may be performed as state determination using state-information-added vector field information. This point will be further described.
Fig. 18 is a schematic block diagram showing an example of the functional configuration of a state prediction device 300b of the state prediction system 100b (Fig. 2). In the configuration shown in Fig. 18, the state prediction device 300b includes a second communication unit 310, a second display unit 320, a second operation input unit 330, a second storage unit 370, and a second control unit 380. The second control unit 380 includes a second expression amount data acquisition unit 381, a second feature amount extraction unit 382, and a state prediction unit 383b.

In the state prediction device 300b, the second control unit 380 includes a state prediction unit 383b instead of the state determination support processing unit 383 included in the second control unit 380 of the state determination support device 300 (FIG. 5).
In the following, a case will be described in which the vector field information generating device 200 generates the above-mentioned state-information-added velocity field information as the state-information-added vector field information. As described above, in the state-information-added velocity field information, a change velocity vector is used as a change amount vector.

The change rate vector is a vector that indicates the amount of change in a feature per unit period. For example, in step S21 of FIG. 16, the vector field information generating unit 286 calculates the change rate vector as the amount of change in a feature (amount of change vector). The vector field information generating unit 286 calculates the amount of change in a feature for the same subject from the feature history information, and divides it by the period of change (the time interval for measuring the expression level of microRNA) to calculate the amount of change (change rate) per unit period.

The state prediction unit 383b predicts the state of the subject based on the feature amount of the expression amount data of the subject and the state information-added velocity field information.
Specifically, the state prediction unit 383b plots the feature of the expression amount data of the subject on the velocity field information with state information, and reads the change rate according to the feature (value). The state prediction unit 383b then adds the change rate to the feature to calculate a predicted value of the feature after a unit period has elapsed. The state prediction unit 383b repeats this process until the prediction target time is reached, and calculates a predicted value of the feature at the prediction target time. The state prediction unit 383b then predicts the state of the subject by reading state information added to a portion (here, a partial region) of the feature space in the velocity field information that includes the predicted value of the feature at the prediction target time.
In other respects, the state prediction device 300 b is similar to the state determination support device 300 .

Fig. 19 is a diagram showing an example of display of one-dimensional state-information-added vector field information by the state prediction device 300b. As in Fig. 6, in the example of Fig. 19, the feature amount space is shown in a one-dimensional coordinate space of variable _x1 . In this one-dimensional coordinate space, a "healthy" region (a region of a healthy state) and a "xx cancer" region (a region of a xx cancer state) are shown.
The second display unit 320 may be configured to clearly indicate the area of the pre-cancer state in addition to the area of the healthy state and the area of the XX cancer state.

In the example of Fig. 19, the feature amount (values) of the subject are displayed for each year from 2020 to 2024. As described above, the state prediction unit 383b adds the change rate read from the feature amount in the velocity field information to the feature amount (values) of the subject, and repeats the process of calculating the predicted value of the feature amount after a unit period (one year in the example of Fig. 19) has elapsed, up to the year 2024. The second display unit 320 displays the feature amount calculated for each year in the _x1 coordinate space according to the control of the second control unit 380.

In the example of Figure 19, the feature values for 2020 are plotted in the "healthy" area, indicating that the subject is in a healthy state. For 2021 and 2022, the feature values are plotted in the area between "healthy" and "XX cancer," indicating that the subject is not suffering from XX cancer, but is not in a healthy state.

For the years 2023 and 2024, the feature values are plotted in the "XX cancer" area, indicating that the subject will be diagnosed with XX cancer.
Furthermore, for the year 2023, the cancer may be predicted to be in an early stage based on the feature being plotted closer to the "healthy" region in the "XX cancer" region. For the year 2024, the cancer may be predicted to be in an advanced state based on the feature being plotted on the opposite side of the "healthy" region in the "XX cancer" region.

The subjects can refer to the prediction results and take preventive measures to avoid contracting the specific cancer.
When there are multiple states to be predicted, such as when the state prediction device 300b makes predictions for multiple types of cancer, the second display unit 320 displays, for each state, one-dimensional velocity field information with state information, as exemplified in FIG. 19 , and a plot of the transition of the subject's features.

Fig. 20 is a diagram showing an example of display of two-dimensional state-information-added vector field information by the state prediction device 300b. As in the case of Fig. 7, in the example of Fig. 20, the feature amount space is shown in a two-dimensional orthogonal coordinate space of variables _x1 and _x2 . In this two-dimensional coordinate space, a "healthy" region and regions of each type of cancer from "cancer A" to "cancer D" are shown.

In the example of FIG. 20, the second display unit 320 may also display the pre-disease region. In that case, the second display unit 320 may display the pre-disease region for each cancer type. Alternatively, the second display unit 320 may display an area that combines the pre-disease regions for each cancer type as the pre-disease region.

Also, as in the case of FIG. 19, in the example of FIG. 20, the feature amount (values) of the subject are displayed for each year from 2020 to 2024. In the example of FIG. 20, the state prediction unit 383b adds the rate of change read from the feature amount in the velocity field information to the feature amount (values) of the subject, and repeats the process of calculating the predicted value of the feature amount after a unit period has elapsed up to the year 2024. The second display unit 320 displays the feature amount calculated for each year in a two-dimensional coordinate space in accordance with the control of the second control unit 380.

In the example of Figure 20, the feature values for 2020 are plotted in the "healthy" area, indicating that the subject is in a healthy state. For 2021 and 2022, the feature values are plotted in the area between "healthy" and each cancer, indicating that the subject is not affected by these cancers, but is not in a healthy state either.

For the years 2023 and 2024, the feature values are plotted in the "cancer C" area, indicating that the subject will be diagnosed with cancer C.
Furthermore, for the year 2023, the cancer may be predicted to be in an early stage based on the feature being plotted on the side of the "cancer C" region closer to the "healthy" region. For the year 2024, the cancer may be predicted to be in an advanced state based on the feature being plotted on the side of the "cancer C" region opposite the "healthy" region.
The subject can refer to the prediction results and take preventive measures to avoid contracting C cancer.

Fig. 21 is a diagram showing an example of display of three-dimensional state-information-added vector field information by the state prediction device 300b. As in the case of Fig. 8, in the example of Fig. 21, the feature amount space is shown in a three-dimensional orthogonal coordinate space of variables _x1 , _x2 , and _x3 . In this three-dimensional coordinate space, a "healthy" region and regions of each type of cancer from "cancer A" to "cancer D" are shown.

In the example of FIG. 21, the second display unit 320 may also display the pre-disease region. In that case, the second display unit 320 may display the pre-disease region for each cancer type. Alternatively, the second display unit 320 may display an area that combines the pre-disease regions for each cancer type as the pre-disease region.

Also, as in the case of FIG. 20, in the example of FIG. 21, the feature amount (values) of the subject are displayed for each year from 2020 to 2024. In the example of FIG. 21, the state prediction unit 383b adds the rate of change read from the feature amount in the velocity field information to the feature amount (values) of the subject, and repeats the process of calculating the predicted value of the feature amount after a unit period has elapsed up to the year 2024. The second display unit 320 displays the feature amount calculated for each year in a two-dimensional coordinate space in accordance with the control of the second control unit 380.

In the example of Figure 21, the feature values for 2020 are plotted in the "healthy" area, indicating that the subject is in a healthy state. For 2021, 2022, and 2023, the feature values are plotted in the area between "healthy" and each cancer, indicating that the subject is not affected by these cancers, but is not in a healthy state either.

For the year 2024, the feature value is plotted in the "cancer D" region, indicating that the subject will be diagnosed with cancer D.
Furthermore, it may be possible to predict that cancer is in an early stage based on the feature values for 2024 being plotted on the side of the "D Cancer" region closer to the "Healthy" region.
The subject can refer to the prediction results and take preventive measures to avoid contracting D cancer.

FIG. 22 is a flowchart illustrating an example of a processing procedure in which the condition prediction device 300b predicts the condition of a subject.
Steps S31 and S32 in Fig. 22 are similar to steps S31 and S32 in Fig. 17. Step S33 in Fig. 22 is similar to step S33 in Fig. 17 except that the state prediction unit 383b performs processing instead of the state determination support processing unit 383.

After step S33, the state prediction unit 383b predicts the state of the subject (step S34). Specifically, as described above with reference to Figs. 19 to 21, the state prediction unit 383b also plots the predicted value of the state of the subject in the feature amount space of the state-information-added velocity field information. Then, the state prediction unit 383b predicts the state labeled in the area in which the feature amount of the subject is plotted in the feature amount space of the state-information-added velocity field information as the future state of the subject. After step S34, the state prediction device 300b ends the process of Fig. 22.

As described above, the second expression data acquisition unit 381 acquires expression data indicating the amount of microRNA for each type of microRNA for the subject. The second feature extraction unit 382 extracts features of the expression data. The condition prediction unit 383b predicts the condition of the subject based on the features of the expression data and the vector field information with condition information.
The condition prediction device 300b can predict the condition of a subject. The subject can refer to the prediction result and take measures, such as taking preventive measures to avoid contracting a disease.

FIG. 23 is a schematic block diagram showing the configuration of a computer according to at least one embodiment. In the configuration shown in FIG. 18, the computer 700 includes a CPU (Central Processing Unit) 710, a main memory device 720, an auxiliary memory device 730, and an interface 740.

Any one or more of the above-mentioned vector field information generating device 200, state judgment assistance device 300, and state prediction device 300b may be implemented in the computer 700. In this case, the operation of each of the above-mentioned processing units is stored in the auxiliary storage device 730 in the form of a program. The CPU 710 reads the program from the auxiliary storage device 730, expands it in the main storage device 720, and executes the above-mentioned processing according to the program. The CPU 710 also secures memory areas in the main storage device 720 corresponding to each of the above-mentioned memory units according to the program.

When the vector field information generating device 200 is implemented in a computer 700, the operations of the first control unit 280 and each of its units are stored in the form of a program in the auxiliary storage device 730. The CPU 710 reads the program from the auxiliary storage device 730, loads it into the main storage device 720, and executes the above-mentioned processing in accordance with the program.
Furthermore, the CPU 710 reserves a memory area corresponding to the first memory unit 270 in the main memory device 720 in accordance with the program.

Communication by the first communication unit 210 is performed by the interface 740 having a communication function and performing communication under the control of the CPU 710. The function of the first display unit 220 is performed by the interface 740 having a display device and displaying an image under the control of the CPU 710. The function of the first operation input unit 230 is performed by the interface 740 having an input device, accepting user operations, and outputting a signal indicating the accepted user operation to the CPU 710.

When the state judgment assistance device 300 is implemented in the computer 700, the operations of the second control unit 380 and each of its units are stored in the form of a program in the auxiliary storage device 730. The CPU 710 reads the program from the auxiliary storage device 730, loads it into the main storage device 720, and executes the above-mentioned processing in accordance with the program.
Furthermore, the CPU 710 reserves a memory area corresponding to the second memory unit 370 in the main memory device 720 in accordance with the program.

Communication by the second communication unit 310 is performed by the interface 740 having a communication function and performing communication under the control of the CPU 710. The function of the second display unit 320 is performed by the interface 740 having a display device and displaying an image under the control of the CPU 710. The function of the second operation input unit 330 is performed by the interface 740 having an input device, accepting user operations, and outputting a signal indicating the accepted user operation to the CPU 710.

When the state prediction device 300b is implemented in the computer 700, the operations of the second control unit 380 and each of its units are stored in the form of a program in the auxiliary storage device 730. The CPU 710 reads the program from the auxiliary storage device 730, loads it into the main storage device 720, and executes the above-mentioned processing in accordance with the program.
Furthermore, the CPU 710 reserves a memory area corresponding to the second memory unit 370 in the main memory device 720 in accordance with the program.

Communication by the second communication unit 310 is performed by the interface 740 having a communication function and performing communication under the control of the CPU 710. The function of the second display unit 320 is performed by the interface 740 having a display device and displaying an image under the control of the CPU 710. The function of the second operation input unit 330 is performed by the interface 740 having an input device, accepting a user operation, and outputting a signal indicating the accepted user operation to the CPU 710.
The auxiliary storage device 730 is, for example, a non-volatile (non-transitory) recording medium such as a CDC (Compact Disc) or a DVD (digital versatile disc).

Note that a program for realizing all or part of the functions of the first control unit 280 and the second control unit 380 may be recorded in a computer-readable recording medium, and the program recorded in the recording medium may be read into a computer system and executed to process each unit. Note that the term "computer system" here includes hardware such as an OS (Operating System) and peripheral devices.
Furthermore, the term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs (Read Only Memory), and CD-ROMs (Compact Disc Read Only Memory), as well as storage devices such as hard disks built into computer systems. The above-mentioned program may be one for implementing part of the above-mentioned functions, or may be one that can implement the above-mentioned functions in combination with a program already recorded in the computer system.

The above describes an embodiment of the present invention in detail with reference to the drawings, but the specific configuration is not limited to this embodiment, and design changes and the like that do not deviate from the gist of the present invention are also included.

100 State judgment support system 100b State prediction system 200 Vector field information generation device 210 First communication unit 220 First display unit 230 First operation input unit 270 First memory unit 280 First control unit 281 First expression amount data acquisition unit 282 First feature amount extraction unit 283 Classification unit 284 Machine learning control unit 285 History information registration unit 286 Vector field information generation unit 287 State information addition unit 300 State judgment support device 300b State prediction device 310 Second communication unit 320 Second display unit 330 Second operation input unit 370 Second memory unit 380 Second control unit 381 Second expression amount data acquisition unit 382 Second feature amount extraction unit 383 State judgment support processing unit 383b State prediction unit

Claims (7)

a biomarker amount data acquiring unit for acquiring biomarker amount data indicating the amount of each type of biomarker;
a feature extraction unit for extracting two-dimensional or higher-dimensional features as features of the biomarker amount data;
a vector field information generation unit that acquires a plurality of different change amount vectors at the same point in a feature amount space based on history information of the feature amounts of a plurality of people , and generates vector field information that probabilistically indicates the change amount of the feature amount for each of the feature amounts;
a state information adding unit that generates state information-added vector field information in which state information indicating a state of a portion of the feature space in the vector field information is added to the portion, and generates state information-added vector field information in which state information of each of a plurality of types of cancer is added to one of the vector field information ;
A vector field information generating device comprising: The vector field information generation unit generates a vector field indicating a rate of change of the feature amount for each of the feature amounts.
The vector field information generating device according to claim 1 . The state information adding unit generates the state information-added vector field information to which state information indicating a non-illness state is added.
3. The vector field information generating device according to claim 1 or 2. A system comprising: the vector field information generating device according to any one of claims 1 to 3 ; and a state determination support device;
The state determination support device includes:
a second biomarker amount data acquisition unit that acquires the biomarker amount data of the subject;
a second feature extraction unit that extracts features of the biomarker amount data of the subject by a feature extraction method similar to that of the feature extraction unit of the vector field information generating device;
a state determination support processing unit that indicates a position of a feature of the biomarker amount data of the subject in a feature space in the vector field information;
A condition determination support system. A system comprising: the vector field information generating device according to any one of claims 1 to 3 ; and a state prediction device;
The state prediction device includes:
a second biomarker amount data acquisition unit that acquires the biomarker amount data of the subject;
a second feature extraction unit that extracts features of the biomarker amount data of the subject by a feature extraction method similar to that of the feature extraction unit of the vector field information generating device;
a state prediction unit which predicts the state of the subject by repeating the steps of reading the change rate according to the feature amount of the biomarker amount data of the subject from the state-information-attached vector field information, calculating a predicted value of the feature amount after a unit period has elapsed by adding the change rate to the feature amount, and reading the change rate according to the feature amount and calculating a predicted value of the feature amount after a unit period has elapsed by adding the change rate to the feature amount, until a prediction target time is reached ;
A state prediction system. acquiring biomarker amount data indicating the amount of the biomarker for each type of biomarker;
extracting two or more dimensional features as features of the biomarker amount data;
obtaining a plurality of different change amount vectors at the same point in the feature amount space based on history information of the feature amounts of a plurality of people, and generating vector field information that probabilistically indicates the change amount of the feature amount for each of the feature amounts;
generating state-information-added vector field information in which state information indicating a state of a portion of the feature space in the vector field information is added to the portion, and generating state-information-added vector field information in which state information of each of a plurality of types of cancer is added to one of the vector field information ;
A vector field information generating method comprising: On the computer,
acquiring biomarker amount data indicating the amount of the biomarker for each type of biomarker;
extracting two or more dimensional features as features of the biomarker amount data;
obtaining a plurality of different change amount vectors at the same point in the feature amount space based on history information of the feature amounts of a plurality of people, and generating vector field information that probabilistically indicates the change amount of the feature amount for each of the feature amounts;
generating state-information-added vector field information in which state information indicating a state of a portion of the feature space in the vector field information is added to the portion, and generating state-information-added vector field information in which state information of each of a plurality of types of cancer is added to one of the vector field information ;
A program for executing.

Images