JP7644456B2 - Apparatus, system, method and program for evaluating gastrointestinal condition - Google Patents

Description

This disclosure relates to an evaluation device, evaluation system, evaluation method, and evaluation program for evaluating gastrointestinal conditions.

Technologies for evaluating the state of the intestines based on bowel sounds have been proposed in the past (for example, Patent Document 1). Specifically, Patent Document 1 discloses a method for calculating the amount of gas in the intestines based on time domain features such as the duration and frequency of bowel sounds.

JP 2021-74238 A

However, as shown in the comparative example described below, time domain features of bowel sounds cannot detect changes in gastrointestinal conditions related to hunger, post-meal status, health status, lifestyle, etc. Therefore, the conventional technology described in Patent Document 1 has the problem of being unable to accurately evaluate gastrointestinal conditions.

The objective of this disclosure is to accurately evaluate the gastrointestinal condition.

In order to solve the above problems, the present disclosure includes the following aspects.
Item 1.
An evaluation device for evaluating a gastrointestinal condition of a subject, comprising:
an extracting unit for extracting a plurality of bowel sounds from acoustic data obtained from the subject;
a feature calculation unit that calculates a feature associated with a frequency domain of each of the bowel sounds;
a distribution data generating unit that generates distribution data indicating a distribution of each of the feature amounts;
an evaluation unit for evaluating a gastrointestinal condition of the subject based on the distribution data;
An evaluation device comprising:
Item 2.
The evaluation unit is
a reference data acquisition unit that acquires reference distribution data indicating a distribution of the feature amounts of a plurality of bowel sounds included in acoustic data obtained from the subject when the gastrointestinal condition of the subject is in a predetermined state;
a similarity calculation unit that calculates a similarity between the distribution data generated by the distribution data generation unit and the reference distribution data;
Equipped with
Item 2. The evaluation device according to item 1, wherein the state is evaluated based on the predetermined state and the degree of similarity.
Item 3.
3. The evaluation device according to item 1 or 2, wherein the feature amount is a representative frequency.
Item 4.
The feature amount calculation unit is
a division unit for dividing each of the bowel sounds into a plurality of segments at regular time intervals;
A peak frequency calculation unit that calculates a peak frequency of each of the segments;
a representative frequency selection unit that selects a maximum peak frequency from the peak frequencies as the representative frequency;
Item 4. The evaluation device according to item 3, comprising:
Item 5.
An evaluation device according to any one of items 1 to 4,
a sound collection device for obtaining the acoustic data from the subject;
An evaluation system comprising:
Item 6.
1. A method for evaluating a gastrointestinal condition of a subject, comprising:
extracting a plurality of bowel sounds from acoustic data obtained from the subject;
a feature calculation step of calculating a feature associated with a frequency domain of each of the bowel sounds;
a distribution data generating step of generating distribution data indicating a distribution of each of the feature amounts;
evaluating a gastrointestinal condition of the subject based on the distribution data;
The evaluation method has the following features:
Item 7.
The evaluation step includes:
a reference data acquiring step of acquiring reference distribution data indicating a distribution of the feature amounts of a plurality of bowel sounds included in acoustic data obtained from the subject when the gastrointestinal condition of the subject is in a predetermined state;
a similarity calculation step of calculating a similarity between the distribution data generated in the distribution data generation step and the reference distribution data;
having
Item 7. The evaluation method according to item 6, further comprising evaluating the state based on the predetermined state and the degree of similarity.
Item 8.
Item 8. The evaluation method according to item 6 or 7, wherein the feature amount is a representative frequency.
Item 9.
The feature amount calculation step includes:
a segmentation step of dividing each of the bowel sounds into a plurality of segments at regular time intervals;
a peak frequency calculation step of calculating a peak frequency of each of the segments;
a representative frequency selection step of selecting a maximum peak frequency from the peak frequencies as the representative frequency;
Item 9. The evaluation method according to item 8, comprising:
Item 10.
1. An assessment program for assessing a gastrointestinal condition of a subject, comprising:
extracting a plurality of bowel sounds from acoustic data obtained from the subject;
a feature calculation step of calculating a feature associated with a frequency domain of each of the bowel sounds;
a distribution data generating step of generating distribution data indicating a distribution of each of the feature amounts;
evaluating a gastrointestinal condition of the subject based on the distribution data;
An evaluation program that causes a computer to execute the above.
Item 11.
The evaluation step includes:
a reference data acquiring step of acquiring reference distribution data indicating a distribution of the feature amounts of a plurality of bowel sounds included in acoustic data obtained from the subject when the gastrointestinal condition of the subject is in a predetermined state;
a similarity calculation step of calculating a similarity between the distribution data generated in the distribution data generation step and the reference distribution data;
having
Item 11. The evaluation program according to item 10, further comprising: evaluating the state based on the predetermined state and the degree of similarity.
Item 12.
Item 12. The evaluation program according to item 10 or 11, wherein the feature amount is a representative frequency.
Item 13.
The feature amount calculation step includes:
a segmentation step of dividing each of the bowel sounds into a plurality of segments at regular time intervals;
a peak frequency calculation step of calculating a peak frequency of each of the segments;
a representative frequency selection step of selecting a maximum peak frequency from the peak frequencies as the representative frequency;
Item 13. The evaluation program according to item 12, comprising:

This disclosure makes it possible to accurately evaluate gastrointestinal conditions.

1 is a block diagram showing a configuration of an evaluation system according to an embodiment of the present disclosure. 1 is a flowchart showing a processing procedure of an evaluation method according to an embodiment of the present disclosure. 2 is an example of a waveform of acoustic data obtained from a subject. 13 is a flowchart showing more specific processing steps of the feature amount calculation step. 1 is a graph showing the frequency spectrum of each segment of a single bowel sound. 1 is an example of distribution data. 13 is a flowchart showing more specific processing steps of the evaluation step. 1 is an example of reference distribution data. 13 is a graph showing the average value for each subject of the degree of similarity of distribution data relating to three combinations. (a) to (d) are box plots showing BS power, BS length, BS frequency, and BS occurrence interval extracted from acoustic data obtained before carbonated water ingestion (BSI), before water ingestion (BWI), after carbonated water ingestion (ASI), and after water ingestion (AWI), respectively.

Embodiments of the present disclosure will be described below with reference to the accompanying drawings. Note that the present disclosure is not limited to the following embodiments, and various modifications are possible without departing from the spirit of the present disclosure.

(System Configuration)
1 is a block diagram showing a configuration of an evaluation system 1 according to an embodiment of the present disclosure. The evaluation system 1 includes a sound collection device 2 and an evaluation device 3.

The sound collection device 2 is a device for obtaining acoustic data from the subject. When collecting acoustic data at a medical institution, the sound collection device 2 is, for example, a stethoscope, and when collecting acoustic data outside a medical institution, the sound collection device 2 is, for example, a microphone. When the sound collection device 2 is a microphone, the sound collection device 2 is attached to the subject's abdomen directly or via clothing, and obtains acoustic data including the subject's bowel sounds.

The sound collection device 2 is connected to the evaluation device 3 by wire or wirelessly, and transfers the acquired acoustic data to the evaluation device 3. Note that the acoustic data may be transferred from the sound collection device 2 to the evaluation device 3 via a recording medium or a communication device.

The evaluation device 3 is a device that evaluates the gastrointestinal condition of a subject. In this disclosure, "gastrointestinal" means "at least one of the stomach and intestines." The evaluation device 3 can be configured as a general-purpose computer, as well as a portable computer such as a smartphone or tablet terminal. The evaluation device 3 has, as its hardware configuration, a processor such as a CPU or GPU, a main memory device such as a DRAM or SRAM (not shown), and an auxiliary memory device 30 such as an HDD or SSD. The auxiliary memory device 30 stores an evaluation program P, an extraction model M, reference distribution data B, etc.

The auxiliary storage device 30 may be external to the evaluation device 3. The evaluation device 3 may also be provided on the cloud.

The evaluation device 3 includes, as functional blocks, an acoustic data acquisition unit 31, an extraction unit 32, a feature calculation unit 33, a distribution data generation unit 34, and an evaluation unit 35. Furthermore, the feature calculation unit 33 includes a division unit 331, a peak frequency calculation unit 332, and a representative frequency selection unit 333, and the evaluation unit 35 includes a reference data acquisition unit 351 and a similarity calculation unit 352. These functional blocks are realized by the processor of the evaluation device 3 reading the evaluation program P into the main storage device and executing it. The evaluation program P may be downloaded to the evaluation device 3 via a communication network such as the Internet, or the evaluation program P may be recorded on a computer-readable non-transitory recording medium such as a CD-ROM or an SD card and installed on the evaluation device 3 via the recording medium.

(Procedure for evaluation method)
The function of the evaluation device 3 will be described with reference to Fig. 2. Fig. 2 is a flowchart showing the processing procedure of the evaluation method according to this embodiment. The evaluation method has steps S1 to S5, and the above-mentioned functional blocks of the evaluation device 3 execute steps S1 to S5. That is, the evaluation program P causes the evaluation device 3 to execute steps S1 to S5.

In step S1, the acoustic data acquisition unit 31 acquires acoustic data obtained from the subject by the sound collection device 2. The length (time) of the acoustic data is not particularly limited, but is, for example, 5 minutes.

In step S2 (extraction step), the extraction unit 32 extracts multiple bowel sounds from the acoustic data acquired in step S1. The acoustic data includes noise such as breathing sounds and body movement sounds in addition to bowel sounds, but bowel sounds can be extracted using known techniques. In this embodiment, the extraction unit 32 extracts bowel sounds using the technique described in WO 2019/216320.

Specifically, the extraction unit 32 detects multiple segments from the acoustic data, extracts power normalized cepstral coefficients (PNCCs), which are frequency-related features, from each segment, and inputs the extracted PNCCs to the extraction model M. The extraction model M is a neural network model that uses machine learning to learn the relationship between bowel sounds in the acoustic data and PNCCs, and outputs a prediction score that indicates the possibility that each segment contains a bowel sound according to the input PNCC. The extraction unit 32 extracts segments with a prediction score greater than a predetermined threshold as bowel sounds. A bowel sound is a continuous sound, and its length (time) is indefinite.

Figure 3 shows an example of the waveform of acoustic data obtained from a subject. The dashed frames indicate segments, and the extraction unit 32 extracts segments that are likely to contain bowel sounds.

The frequency of normal bowel sounds is generally around 100 to 500 Hz. Therefore, it is preferable for the extraction unit 32 to apply a bandpass filter that passes a frequency band including 100 to 500 Hz (e.g., 80 to 1000 Hz) to the acoustic data acquired in step S1 before extracting the bowel sounds.

In step S3 (feature calculation step) shown in FIG. 2, the feature calculation unit 33 calculates features related to the frequency range of each bowel sound extracted in step S2. In this embodiment, the feature is a representative frequency. The representative frequency is a frequency that represents the pitch of a single continuous sound, and in this embodiment, the representative frequency is calculated as follows.

Figure 4 is a flowchart showing more specific processing steps of step S3.

In step S31 (division step), the division unit 331 divides each bowel sound into multiple segments at regular time intervals.

In step S32 (peak frequency calculation step), the peak frequency calculation unit 332 calculates the peak frequency of each segment divided in step S31. Specifically, the peak frequency calculation unit 332 performs a Fourier transform on the waveform of each segment to decompose it into different frequency components, calculate the intensity of each component, and determine the frequency with the greatest intensity as the peak frequency.

For example, if one bowel sound is divided into five segments, the frequency spectrum of segments 1 to 5 is calculated as shown in Figure 5, and the frequency corresponding to the peak of each frequency spectrum indicated by a triangular mark is the peak frequency.

In step S33 (representative frequency selection step), the representative frequency selection unit 333 selects the maximum peak frequency from the peak frequencies calculated in step S32 as the representative frequency. In the example shown in Fig. 5, the peak frequency of segment 3 among the five peak frequencies indicated by triangular marks is selected as the representative frequency F _R. Note that the representative frequency selection unit 333 may select the basic statistics of the peak frequencies from the peak frequencies calculated in step S32 as the representative frequency.

By performing the above steps S31 to S33 for all bowel sounds (YES in step S34), the feature calculation unit 33 calculates the representative frequencies of all bowel sounds. Note that the feature calculation unit 33 may calculate the frequency corresponding to the peak of the frequency spectrum of one bowel sound as the representative frequency F _R without dividing each bowel sound into segments. Also, in step S32, instead of calculating the peak frequency of each segment, the feature calculation unit 33 may calculate the spectral features of each segment (spectral flatness, spectral bandwidth, spectral centroid, etc.).

In step S4 (distribution data generation step) shown in FIG. 2, the distribution data generation unit 34 generates distribution data A indicating the distribution of each feature amount (representative frequency). In this embodiment, the distribution data generation unit 34 generates the distribution data A as a histogram.

Figure 6 is an example of distribution data A. This distribution data A has 42 frequency bins in the range of 0 to 1000 Hz. Note that the format of distribution data A is not limited to a histogram.

In step S5 (evaluation step) shown in FIG. 2, the evaluation unit 35 evaluates the gastrointestinal condition of the subject based on the distribution data A. In this embodiment, the evaluation unit 35 evaluates the gastrointestinal condition based on the similarity between the distribution data A and the reference distribution data B stored in the auxiliary storage device 30. Specifically, the evaluation of the gastrointestinal condition is performed as follows.

Figure 7 is a flowchart showing more specific processing steps of step S5.

In step S51 (reference data acquisition step), the reference data acquisition unit 351 acquires reference distribution data B from the auxiliary storage device 30. The reference distribution data B is data indicating the distribution of representative frequencies of multiple bowel sounds contained in acoustic data obtained from the subject when the subject's gastrointestinal condition is in a predetermined state.

For example, when a subject suffering from severe constipation visits a medical institution for the first time, acoustic data is acquired from the subject, and distribution data of representative frequencies is generated by the processing of steps S2 to S4 described above. This distribution data is stored in the auxiliary storage device 30 as reference distribution data B.

Figure 8 is an example of reference distribution data B. The format of reference distribution data B may be the same as that of distribution data A shown in Figure 6.

In step S52 (similarity calculation step) shown in FIG. 7, the similarity calculation unit 352 calculates the similarity between the distribution data A and the reference distribution data B. In this embodiment, the similarity calculation unit 352 calculates the similarity between the distribution data A and the reference distribution data B represented by a histogram as Pearson's correlation coefficient R. The correlation coefficient R is calculated by the following formula.

N: Number of frequency bins _Ai : Value of the i-th representative frequency of distribution data _ABi : Value of the i-th representative frequency of reference distribution data _BμA : Average value of representative frequencies of distribution data _AμB : Average value of representative frequencies of reference distribution data _BσA : Standard deviation of representative frequencies of distribution data _AσB : Standard deviation of representative frequencies of reference distribution data B

The closer the correlation coefficient R is to 1, the higher the similarity, and the closer the correlation coefficient R is to 0, the lower the similarity. Note that the index representing the similarity is not limited to the correlation coefficient R, and for example, Euclidean distance, Bhattacharya distance, cosine similarity, intersection set, etc. may also be used.

In step S53, the evaluation unit 35 evaluates the gastrointestinal condition of the subject based on the subject's condition (predetermined condition) when the acoustic data for generating the reference distribution data B was acquired and the similarity calculated in step S52. As will be described in the examples below, a lower similarity indicates a greater change in the intestinal condition. In other words, the degree of change in the subject's condition relative to the predetermined condition can be determined depending on the magnitude of the similarity, and therefore the subject's condition when the acoustic data for generating the distribution data A was acquired can be evaluated.

For example, assume that the subject's gastrointestinal condition was severely constipated at the time acoustic data for generating reference distribution data B was acquired at a medical institution. The subject then acquires acoustic data and transmits it to the evaluation device 3, which then generates distribution data A from the acoustic data and calculates the similarity between distribution data A and reference distribution data B. This makes it possible to evaluate how much the gastrointestinal condition has changed (i.e., improved) from the severely constipated state at the time the acoustic data was acquired at home. In other words, the lower the similarity, the better the gastrointestinal condition can be evaluated to be.

In addition, the subject can easily obtain acoustic data using a microphone at home, etc., so gastrointestinal conditions can be monitored on a daily basis by continuously obtaining acoustic data and generating distribution data A.

As described above, the evaluation unit 35 can evaluate the intestinal condition of the subject. In the examples described below, it is shown that the present disclosure can grasp not only changes in the amount of gas in the intestines, but also changes in the contents in the intestines. Therefore, the present disclosure can accurately evaluate the gastrointestinal condition.

Below, examples of the present disclosure are described, but the present disclosure is not limited to the following examples.

In this example, we verified whether the present disclosure can accurately evaluate changes in intestinal condition due to beverage intake or the passage of time. The subjects were 19 people (11 men, 8 women, age: 33.37±7.82 years, height: 166.71±8.37 cm, weight: 59.09±8.84 kg, BMI: 21.16±1.90) who were determined not to have irritable bowel syndrome (IBS) according to the Rome III diagnostic criteria, and each subject underwent a beverage intake test twice.

Specifically, in the first test, each subject, who had fasted for about 12 hours, was given water at about 10°C, and the body sounds of each subject were collected as acoustic data for five minutes immediately before and five minutes 10 to 15 minutes after the ingestion. The second test was conducted three days after the first test. In the second test, each subject, who had fasted for about 12 hours, was given carbonated water at about 10°C, and the body sounds of each subject were collected as acoustic data for five minutes immediately before and five minutes 10 to 15 minutes after the ingestion.

In these tests, the equipment used to collect acoustic data was an electronic stethoscope (E-scope2, Cardionics Inc., Houston, TX, USA) and a multi-track recorder (R16 Zoom Co., Ltd., Tokyo, Japan). When collecting acoustic data, the subject was placed in a supine position, and the electronic stethoscope was fixed in a cross shape using masking tape at a position 9 cm to the right of the navel. The electronic stethoscope has a heart sound mode and a respiratory sound mode, and was set to the respiratory sound mode, which can record a wider frequency range. The sampling frequency of the acoustic data during collection was 44,100 Hz, and the digital resolution was 16 bits/sample.

In the evaluation device 3, a trained neural network model was used as the extraction model M to extract multiple bowel sounds from each piece of acoustic data. Specifically, the acoustic data was segmented with a segment length of 64 ms and a shift size of 16 ms, and a PNCC, which is a frequency-related feature, was extracted from each segment. The extracted PNCC was then input into the neural network model, and a determination was made as to whether each segment contained bowel sounds based on the output of the neural network. When there were consecutive segments containing bowel sounds (BS episodes), the consecutive segments were treated as one BS episode.

Next, the representative frequency of each bowel sound was calculated to generate distribution data. Specifically, the distribution data was generated in the following procedure.
(1) As a preprocessing step, the sampling frequency of the bowel sounds was downsampled to 4000 Hz, and a band-pass filter was applied that passed the 80 to 1000 Hz band, which is the main frequency band of bowel sounds.
(2) Each bowel sound was segmented with a segment length of 64 ms (256 samples) and a shift size of 16 ms (64 samples).
(3) An FFT was performed for each segment using a window of 128 ms (512 samples) to determine the peak frequency.
(4) Among the peak frequencies obtained for each segment, the frequency with the maximum power was selected as the representative frequency of bowel sounds.
(5) Using representative values of bowel sounds generated within a certain period of time, histogram distribution data was generated in a frequency range of 0 to 1000 Hz with 42 frequency bins.

As a result, four pieces of distribution data were generated from the acoustic data obtained from each subject: two pieces of distribution data generated from acoustic data acquired before and after water intake in the first test, and two pieces of distribution data generated from acoustic data acquired before and after carbonated water intake in the second test. Here, the timing at which acoustic data was acquired before water intake in the first test is referred to as BWI (Before water intake), and the timing at which acoustic data was acquired after water intake in the first test is referred to as AWI (After water intake). In addition, the timing at which acoustic data was acquired before carbonated water intake in the second test is referred to as BSI (Before soda intake), and the timing at which acoustic data was acquired after carbonated water intake in the second test is referred to as ASI (After soda intake). In other words, the four pieces of distribution data are distribution data generated from acoustic data acquired in BWI, AWI, BSI, and ASI, and in this embodiment, these distribution data are simply referred to as BWI, AWI, BSI, and ASI.

Next, the similarity of the distribution data relating to the following three combinations from the four distribution data was calculated as Pearson's correlation coefficient R.
・BWI-AWI
・BSI-ASI
・BSI-BWI

Figure 9 is a graph showing the average value for each subject of the degree of similarity of the distribution data for these three combinations. The correlation coefficient R (BWI-AWI), which indicates the degree of similarity between BWI and AWI, is close to 1, suggesting that there is little change in gastrointestinal condition before and after water intake. Previous research has shown that water intake hardly increases gastrointestinal motility, and the results of this example show a similar tendency to those of the previous research.

The correlation coefficient R (BSI-ASI), which indicates the similarity between the BSI and ASI, is significantly smaller than the correlation coefficient R (BWI-AWI) described above. This result indicates that the change in gastrointestinal state before and after ingesting carbonated water is greater than the change in gastrointestinal state before and after ingesting water. This suggests that the present disclosure makes it possible to grasp the change in gastrointestinal state caused by ingesting carbonated water.

Furthermore, the correlation coefficient R (BSI-BWI), which indicates the similarity between the BSI and BWI, is significantly smaller than the two correlation coefficients R mentioned above. This result indicates that the change in intestinal condition between before water intake in the first test and before carbonated water intake in the second test is greater due to the influence of lifestyle and other factors caused by the different measurement dates than the change in gastrointestinal condition before and after water intake and before and after carbonated water intake. This suggests that the present disclosure can capture not only the change in gastrointestinal condition when carbonated water is taken, but also the change in gastrointestinal condition due to the influence of lifestyle and other factors.

Comparative Example
As a comparative example, bowel sounds were extracted from the acoustic data obtained from each of the 19 subjects described above, and further, BS power, BS length, BS frequency (No. of BS detected), and BS occurrence interval (Sound to Sound Interval, SSI) were extracted as time domain features of the bowel sounds.

Figures 10 (a)-(d) are box plots showing the BS power, BS length, BS frequency, and BS occurrence interval extracted from the acoustic data obtained before carbonated water intake (BSI), before water intake (BWI), after carbonated water intake (ASI), and after water intake (AWI), respectively. From Figure 10, it can be seen that the time domain features are significantly different before and after carbonated water intake (BSI-ASI), but no significant difference was observed before and after water intake (BWI-AWI). Furthermore, no significant difference was observed in the time domain features before water intake (BWI) and before carbonated water intake (BSI), i.e., before and after the passage of three days.

Thus, while time domain features of bowel sounds can sensitively detect changes in gastrointestinal motility due to the ingestion of carbonated water, they are thought to be less sensitive in detecting changes in gastrointestinal condition due to lifestyle, etc.

Reference Signs List 1 Evaluation system 2 Sound collection device 3 Evaluation device 30 Auxiliary storage device 31 Acoustic data acquisition unit 32 Extraction unit 33 Feature amount calculation unit 331 Division unit 332 Peak frequency calculation unit 333 Representative frequency selection unit 34 Distribution data generation unit 35 Evaluation unit 351 Reference data acquisition unit 352 Similarity calculation unit A Distribution data B Reference distribution data M Extraction model P Evaluation program

Claims (13)

An evaluation device for evaluating a gastrointestinal condition of a subject, comprising:
an extracting unit for extracting a plurality of bowel sounds from acoustic data obtained from the subject;
a feature calculation unit that calculates a feature associated with a frequency domain of each of the bowel sounds;
a distribution data generating unit that generates distribution data indicating a distribution of each of the feature amounts;
an evaluation unit for evaluating a gastrointestinal condition of the subject based on the distribution data;
Equipped with
The evaluation unit is
a reference data acquisition unit that acquires reference distribution data indicating a distribution of the feature amounts of a plurality of bowel sounds included in acoustic data obtained from the subject when the gastrointestinal condition of the subject is in a predetermined state;
a similarity calculation unit that calculates a similarity between the distribution data generated by the distribution data generation unit and the reference distribution data;
Equipped with
An evaluation device that evaluates the state based on the predetermined state and the similarity . An evaluation device for evaluating a gastrointestinal condition of a subject, comprising:
an extracting unit for extracting a plurality of bowel sounds from acoustic data obtained from the subject;
a feature calculation unit that calculates a feature associated with a frequency domain of each of the bowel sounds;
a distribution data generating unit that generates distribution data indicating a distribution of each of the feature amounts;
an evaluation unit for evaluating a gastrointestinal condition of the subject based on the distribution data;
Equipped with
the feature amount is a representative frequency,
The feature amount calculation unit is
a division unit for dividing each of the bowel sounds into a plurality of segments at regular time intervals;
A peak frequency calculation unit that calculates a peak frequency of each of the segments;
a representative frequency selection unit that selects a maximum peak frequency from the peak frequencies as the representative frequency;
An evaluation device comprising: The evaluation device according to claim 1 , wherein the feature amount is a representative frequency. The feature amount calculation unit is
a division unit for dividing each of the bowel sounds into a plurality of segments at regular time intervals;
A peak frequency calculation unit that calculates a peak frequency of each of the segments;
a representative frequency selection unit that selects a maximum peak frequency from the peak frequencies as the representative frequency;
The evaluation device according to claim 3 . The evaluation device according to claim 1 or 4 ,
a sound collection device for obtaining the acoustic data from the subject;
An evaluation system comprising: 1. A method for evaluating a gastrointestinal condition of a subject by a computer , comprising:
extracting a plurality of bowel sounds from acoustic data obtained from the subject;
a feature calculation step of calculating a feature associated with a frequency domain of each of the bowel sounds;
a distribution data generating step of generating distribution data indicating a distribution of each of the feature amounts;
evaluating a gastrointestinal condition of the subject based on the distribution data;
having
The evaluation step includes:
a reference data acquiring step of acquiring reference distribution data indicating a distribution of the feature amounts of a plurality of bowel sounds included in acoustic data obtained from the subject when the gastrointestinal condition of the subject is in a predetermined state;
a similarity calculation step of calculating a similarity between the distribution data generated in the distribution data generation step and the reference distribution data;
having
The evaluation method includes evaluating the state based on the predetermined state and the similarity . 1. A method for evaluating a gastrointestinal condition of a subject by a computer, comprising:
extracting a plurality of bowel sounds from acoustic data obtained from the subject;
a feature calculation step of calculating a feature associated with a frequency domain of each of the bowel sounds;
a distribution data generating step of generating distribution data indicating a distribution of each of the feature amounts;
evaluating a gastrointestinal condition of the subject based on the distribution data;
having
the feature amount is a representative frequency,
The feature amount calculation step includes:
a segmentation step of dividing each of the bowel sounds into a plurality of segments at regular time intervals;
a peak frequency calculation step of calculating a peak frequency of each of the segments;
a representative frequency selection step of selecting a maximum peak frequency from the peak frequencies as the representative frequency;
The evaluation method has the following features: The evaluation method according to claim 6 , wherein the feature amount is a representative frequency. The feature amount calculation step includes:
a segmentation step of dividing each of the bowel sounds into a plurality of segments at regular time intervals;
a peak frequency calculation step of calculating a peak frequency of each of the segments;
a representative frequency selection step of selecting a maximum peak frequency from the peak frequencies as the representative frequency;
The evaluation method according to claim 8 , comprising: 1. An assessment program for assessing a gastrointestinal condition of a subject, comprising:
extracting a plurality of bowel sounds from acoustic data obtained from the subject;
a feature calculation step of calculating a feature associated with a frequency domain of each of the bowel sounds;
a distribution data generating step of generating distribution data indicating a distribution of each of the feature amounts;
evaluating a gastrointestinal condition of the subject based on the distribution data;
Run the following on your computer :
The evaluation step includes:
a reference data acquiring step of acquiring reference distribution data indicating a distribution of the feature amounts of a plurality of bowel sounds included in acoustic data obtained from the subject when the gastrointestinal condition of the subject is in a predetermined state;
a similarity calculation step of calculating a similarity between the distribution data generated in the distribution data generation step and the reference distribution data;
having
an evaluation program that evaluates the state based on the predetermined state and the similarity degree ; 1. An assessment program for assessing a gastrointestinal condition of a subject, comprising:
extracting a plurality of bowel sounds from acoustic data obtained from the subject;
a feature calculation step of calculating a feature associated with a frequency domain of each of the bowel sounds;
a distribution data generating step of generating distribution data indicating a distribution of each of the feature amounts;
evaluating a gastrointestinal condition of the subject based on the distribution data;
Run the following on your computer:
the feature amount is a representative frequency,
The feature amount calculation step includes:
a segmentation step of dividing each of the bowel sounds into a plurality of segments at regular time intervals;
a peak frequency calculation step of calculating a peak frequency of each of the segments;
a representative frequency selection step of selecting a maximum peak frequency from the peak frequencies as the representative frequency;
An evaluation program. The evaluation program according to claim 10 , wherein the characteristic amount is a representative frequency. The feature amount calculation step includes:
a segmentation step of dividing each of the bowel sounds into a plurality of segments at regular time intervals;
a peak frequency calculation step of calculating a peak frequency of each of the segments;
a representative frequency selection step of selecting a maximum peak frequency from the peak frequencies as the representative frequency;
The evaluation program according to claim 12 , comprising: