Biological information acquisition device, processing device, and computer program

Abstract

To enhance interpretation property of a processing result of biological information acquired from a subject.SOLUTION: A reception device 11 receives waveform data WD corresponding to measurement waveforms of an electrocardiogram of a subject 30 from an electrode 20, and acquires values PV of a plurality of feature parameters associated with the measurement waveforms on the basis of the waveform data WD. A processing device 12 acquires an inference result RS about at least one of a plurality of classes into which the waveform data WD are classified by inputting the values PV of the plurality of feature parameters to a machine learned model 123, and specifies the importance of each of the plurality of feature parameters to the inference result RS. An output device 13 outputs an index showing at least one name of the plurality of feature parameters and the importance specified about the at least one feature parameter. The output device 13 outputs the index in a mode in which the importance related to a feature parameter positively contributing to the inference result and the importance related to a feature parameter negatively contributing to the inference result can be distinguished in addition to the inference result.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a biometric information acquisition device that acquires biometric information of a subject, a processing device that processes the biometric information of a subject, and a computer program that can be executed by a processing unit installed in the processing device. [Background technology]

[0002] Patent Document 1 discloses a device for acquiring a pulse wave, which is an example of biological information of a subject. If it is determined that the measured waveform of the pulse wave contains noise above a predetermined level, the device notifies the user. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2009-100934 Summary of the Invention [Problem to be solved by the invention]

[0004] An object of the present invention is to improve the interpretability of the processing results of biological information obtained from a subject. [Means for solving the problem]

[0005] One aspect of the present invention to achieve the above object is a biological information acquisition device for acquiring biological information of a subject, comprising: an accepting device that accepts waveform data corresponding to a measured waveform of the biological information from a sensor and acquires values of a plurality of characteristic parameters associated with the measured waveform based on the waveform data; a processing device that inputs values of the plurality of feature parameters into a machine-learned model to obtain an inference result for at least one of a plurality of classes into which the waveform data is classified, and identifies the importance of each of the plurality of feature parameters with respect to the inference result; an output device that outputs an index indicating the name of at least one of the plurality of feature parameters and the importance level determined for the at least one feature parameter; It is equipped with:

[0006] One aspect of the present invention to achieve the above object is a processing device for processing biological information of a subject, comprising: an interface that receives values of a plurality of characteristic parameters that are acquired based on waveform data corresponding to a measured waveform of the biological information and that are associated with the measured waveform; a processor that inputs values of the plurality of feature parameters into a machine-learned model to obtain an inference result for at least one of a plurality of classes into which the waveform data is classified, identifies the importance of each of the plurality of feature parameters with respect to the inference result, and outputs to an output device an index indicating the name of at least one of the plurality of feature parameters and the identified importance of the at least one feature parameter; It is equipped with:

[0007] One aspect of the present invention to achieve the above object is a computer program executable by a processor installed in a processing device that processes biological information of a subject, the computer program comprising: When executed, the processing device: receiving values of a plurality of characteristic parameters associated with the measured waveform, the characteristic parameters being acquired based on waveform data corresponding to the measured waveform of the biological information; inputting the values of the plurality of feature parameters into a machine-learned model to obtain an inference result for at least one of a plurality of classes into which the waveform data is classified; Identifying the importance of each of the plurality of feature parameters to the inference result; An output device is caused to output an index indicating the name of at least one of the plurality of feature parameters and the importance level determined for the at least one feature parameter.

[0008] When unstructured data such as waveform data corresponding to the measured waveform of biological information is subjected to inference processing, it is generally difficult to visualize the basis of the inference. However, according to the configurations of the above aspects, by using a method for visualizing feature parameters that can be treated as structured data, it is possible to allow a user to recognize, through indicators, which feature parameters in the measured waveform of the biological information contributed to the inference. Therefore, it is possible to improve the interpretability of the processing results of the biological information. [Brief explanation of the drawings]

[0009] [Figure 1] 1 illustrates an example of a functional configuration of an electrocardiograph according to an embodiment. [Figure 2] 2 illustrates an example of a process flow executed by the processing device of FIG. 1. [Figure 3] 2 shows an example of an index output by the output device of FIG. 1. [Figure 4] 10 shows another example of an index output by the output device of FIG. [Figure 5] 10 shows another example of an index output by the output device of FIG. [Figure 6] 10 shows another example of an index output by the output device of FIG. [Figure 7] 10 shows another example of an index output by the output device of FIG. [Figure 8] 10 shows another example of an index output by the output device of FIG. DETAILED DESCRIPTION OF THE INVENTION

[0010] Examples of embodiments are described in detail below with reference to the accompanying drawings.

[0011] FIG. 1 illustrates the functional configuration of an electrocardiograph 10 according to one embodiment. The electrocardiograph 10 is a device that acquires an electrocardiogram of a subject 30 through electrodes 20. The electrocardiogram is an example of biological information. The electrocardiograph 10 is an example of a biological information acquisition device. The electrodes 20 are an example of sensors.

[0012] The electrocardiograph 10 includes a receiving device 11. The receiving device 11 is configured to receive waveform data WD corresponding to the measured waveform of an electrocardiogram from the electrodes 20. The waveform data WD may be in the form of analog data or digital data. When the waveform data WD is in the form of analog data, the receiving device 11 includes an appropriate conversion circuit including an A / D converter.

[0013] 2, the reception device 11 is configured to acquire, based on the waveform data WD, values PV of a plurality of characteristic parameters associated with the measured waveform of an electrocardiogram. The characteristic parameters are quantities that represent the shape characteristics of the measured waveform. Examples of the characteristic parameters of the measured waveform of an electrocardiogram include the polarity and amplitude of the P wave, the width of the QRS wave, the polarity and amplitude of the T wave, the polarity and amplitude of the U wave, the width of the P wave, the width of the Q wave, the width of the R wave, the width of the S wave, the interval between the P wave and the Q wave, the interval between the Q wave and the T wave, the interval between the P wave and the P wave, and the interval between the R wave and the R wave.

[0014] Specifically, based on a general electrocardiogram measurement algorithm, an average waveform is calculated for each lead based on waveform data WD obtained within a predetermined time, and the above-mentioned characteristic parameter values are obtained for the average waveform.

[0015] 1, the electrocardiograph 10 includes a processing device 12. The processing device 12 includes an input interface 121, a processor 122, a machine-learned model 123, and an output interface 124.

[0016] The input interface 121 is configured to accept the values PV of the plurality of characteristic parameters acquired by the acceptance device 11.

[0017] As illustrated in FIG. 2, the processor 122 is configured to input the values PV of the plurality of feature parameters received by the input interface 121 to the machine-learned model 123.

[0018] The machine-learned model 123 is an algorithm that outputs an inference as to whether or not the subject 30 is at risk of developing paroxysmal atrial fibrillation (pAF) based on the input parameter values PV. The inference results "risk of developing pAF" and "no risk of developing pAF" are examples of multiple classes into which the waveform data WD are classified.

[0019] The inference result RS of the machine-learned model 123 can take various forms. Only a classification result relating to a single class, "at risk of developing pAF," may be output, or a classification result relating to a single class together with the probability, "the risk of developing pAF is 84%." Alternatively, a classification result relating to multiple classes, "the probability of developing pAF is 84% and the probability of not developing pAF is 16%," may be output together with the probability.

[0020] The machine-learned model 123 is an inference algorithm generated through machine learning. The machine-learned model 123 may be generated through machine learning using a neural network such as deep learning, or through other machine learning algorithms. Examples of other machine learning algorithms include decision trees, random forests, and support vector machines.

[0021] The processor 122 is configured to determine the importance of each of the plurality of characteristic parameter values PV for the inference result RS. In this example, the processor 122 is configured to calculate a SHAP (SHapley Additive exPlanations) value for each of the plurality of characteristic parameter values PV. That is, the SHAP value is an example of the importance.

[0022] The SHAP value is obtained by calculating how the presence of a certain feature value changes the prediction result for all combinations of feature values that do not include that value. If the presence of a certain feature value significantly changes the prediction result, the SHAP value will be large. In this case, it can be determined that the feature value contributes significantly to the prediction. In other words, by calculating the SHAP value for each of multiple feature parameters, the contribution (importance) of each feature parameter to the inference result RS can be identified. The method of calculating the SHAP value itself is well known, so a detailed explanation will be omitted.

[0023] The processor 122 is configured to generate an index data ID for outputting an index indicating the name of at least one of the plurality of characteristic parameters and the SHAP value determined for the at least one characteristic parameter, and output the generated index data from the output interface 124. The index data ID may be in the form of analog data or digital data. When the index data ID is in the form of analog data, the output interface 124 includes an appropriate conversion circuit including a D / A converter.

[0024] As illustrated in FIG. 1, the electrocardiograph 10 includes an output device 13. The output device 13 can be realized as a display that displays an index based on the index data ID output from the processing device 12. The display of the index is an example of outputting the index. The output form of the index may be other visual presentation such as image projection or printing, or may be auditory presentation in addition to or instead of the visual presentation.

[0025] 3 shows an example of an index output by the output device 13. In this example, the index indicates the class "at risk of developing pAF" included in the inference result, along with its probability. In addition, the index indicates the names of multiple feature parameters in the measured waveform of the electrocardiogram, along with the SHAP value identified for each feature parameter and the measurement value of each feature parameter. In this example, the magnitude of the SHAP value corresponds to the length of the bar extending in the left-right direction.

[0026] In this example, it is shown that the "area under the negative P waveform in lead V1" in the measured waveform of the electrocardiogram contributed most to the inferred result that "the probability of developing pAF is 84%." On the other hand, it is shown that the "R wave width in lead V6" in the measured waveform of the electrocardiogram contributed most to the inferred result that "there is a possibility that pAF will not occur."

[0027] When unstructured data such as waveform data WD corresponding to electrocardiogram waveforms is subjected to inference processing, it is generally difficult to visualize the basis of the inference. However, with the above-described configuration, by using a method for visualizing feature parameters that can be treated as structured data, it is possible to allow a user to recognize, through indicators, which feature parameters in the electrocardiogram waveform contributed to the inference. Therefore, it is possible to improve the interpretability of the electrocardiogram processing results.

[0028] In particular, when the machine-learned model 123 is generated through machine learning using a neural network, it is fundamentally difficult to obtain a clear basis for the inference results. In the medical field, there is a tendency to avoid ambiguity in the basis for judgments, so in such cases, the effect of improving the interpretability of the electrocardiogram processing results becomes more pronounced.

[0029] In this example, the index shows multiple SHAP values determined for multiple characteristic parameters in a manner that allows comparison of their relative importance. For example, for the inference result that "the probability of developing pAF is 84%," it shows that the "area over the negative P waveform in lead V1" in the measured waveform of the electrocardiogram has a higher importance than the "amplitude of the positive P wave in lead V3."

[0030] This configuration allows the user to recognize, through the indicator, which feature parameter of the measured waveform of the electrocardiogram contributed more significantly to the inference than the other feature parameter, thereby further improving the interpretability of the electrocardiogram processing results.

[0031] If it is possible to display the multiple SHAP values determined for multiple feature parameters in a manner that allows comparison of their relative importance, the indicator output by the output device 13 may take the form shown in Fig. 4. In this example, the names of the multiple feature parameters whose importance has been determined are arranged vertically. The magnitude of the SHAP value determined for each feature parameter corresponds to the length of the bar arranged beside the name.

[0032] On the other hand, the index output by the output device 13 may be configured to indicate the importance of a single feature parameter. For example, the name of the feature parameter that contributed most to the inference result may be displayed. In this case, the visibility of the index can be improved and the user can be made aware of information of relatively high importance.

[0033] For example, in the index shown in Fig. 3, the SHAP value determined for each characteristic parameter is shown together with the value of the characteristic parameter. For example, it is shown that the "positive P wave amplitude in lead V3" is 85 mV and the "positive P wave amplitude in lead V1" is 70 mV.

[0034] Since medical professionals generally recognize the possible values of each feature parameter, the above configuration allows them to consider the appropriateness of the importance of a specific feature parameter to the inference result indicated by the index, taking into account the value of the feature parameter.

[0035] However, the display of at least one value of the plurality of characteristic parameters may be omitted as appropriate in consideration of the visibility of the indicator.

[0036] As illustrated in FIG. 5, the output device 13 can display the name of at least one of the plurality of characteristic parameters and an indicator indicating the SHAP value identified for the at least one characteristic parameter, superimposed on the measured electrocardiogram waveform WF.

[0037] In this example, the measured waveform WF is an average electrocardiogram waveform acquired by the reception device 11. In addition, the names of multiple characteristic parameters, namely, "positive P wave amplitude," "P wave and Q wave interval," "ST elevation," and "T wave amplitude," for the estimated result "possible occurrence of pAF," and indices including the importance levels identified for each characteristic parameter are displayed superimposed on the measured waveform WF.

[0038] More specifically, an indicator indicating the importance of a certain feature parameter is displayed as a band overlapping the portion of the measured waveform WF to which the feature parameter relates. The numerical value displayed overlapping the band represents the value of the feature parameter. For example, an indicator related to "positive P-wave amplitude" is displayed as a band overlapping the position of the P wave in the measured waveform WF, along with the value of "positive P-wave amplitude" of "60 (mV)." In this example, the magnitude of the SHAP value corresponds to the intensity of the color of the band.

[0039] With this configuration, it becomes easier to intuitively understand which part of the measured waveform WF the feature parameters related to contributed to the inference result made on the electrocardiogram.

[0040] In addition, if multiple feature parameters are associated with a specific location in the measured waveform WF and multiple levels of importance are identified for the multiple feature parameters, the name of the feature parameter with the greatest level of importance and an indicator indicating that greatest level of importance may be displayed.

[0041] For example, if a SHAP value is identified for each of the "positive P wave amplitude" and "area under the positive P waveform" associated with the P wave portion of the measured waveform WF, and the SHAP value for the "positive P wave amplitude" is higher, only the indicator corresponding to the SHAP value identified for the "positive P wave amplitude" is displayed superimposed on the measured waveform WF.

[0042] With this configuration, it is possible to prevent a decrease in visibility due to multiple indices being displayed in a limited area of the measurement waveform WF.

[0043] Alternatively, as illustrated in FIG. 6, the output device 13 may display the name of at least one of the plurality of characteristic parameters and an indicator indicating the SHAP value identified for the at least one characteristic parameter so as not to overlap with the measured waveform WF of the electrocardiogram.

[0044] In this example, a measured electrocardiogram waveform WF corresponding to multiple heartbeats is displayed, but an average waveform may also be displayed, as in the example shown in FIG. 5. Furthermore, as shown in this example, an index indicating importance determined using gradient-weighted class activation mapping (Grad-CAM) may be displayed superimposed on the measured waveform WF as needed. In this example, a band-shaped index is displayed superimposed on waveform portions determined to have high importance for the inference result "possibly causing pAF." The level of importance corresponds to the intensity of the band's color.

[0045] With this configuration, the validity of the output index can be examined while referring to the measured waveform WF of the electrocardiogram. Also, other indices indicating importance can be used in combination as needed.

[0046] As illustrated in Figure 7, the output device 13 may display indices associated with the atria and ventricles. The atria and ventricles are an example of multiple body parts in the subject.

[0047] Specifically, an index associated with the atrium is generated by adding up the SHAP values determined for the feature parameters associated with the P wave of the electrocardiogram. Similarly, an index associated with the ventricle is generated by adding up the SHAP values determined for the feature parameters associated with the QRST wave of the electrocardiogram. The addition may be performed for all feature parameters, or may be performed for a predetermined number of feature parameters in descending order of SHAP value. Multiplication by a coefficient may be performed as appropriate. In this example, the magnitude of the sum of the SHAP values corresponds to the length of the bar extending in the left-right direction.

[0048] 8, the output device 13 can display indices associated with each of leads V1 to V6. In the figure, the shape of the heart as seen from below the human body is displayed in a schematic manner, and indices associated with each of leads V1 to V6 are arranged so as to correspond to positions on the heart. In other words, the positions from which leads V1 to V6 are acquired are examples of multiple body parts of the subject.

[0049] Specifically, the SHAP values identified for the feature parameters related to lead V1 are added together to generate an index associated with the acquisition position of lead V1. Similar processing is performed for each of leads V2 to V6. The addition may be performed for all feature parameters, or may be performed for a predetermined number of feature parameters in descending order of SHAP value. Multiplication by a coefficient may be performed as appropriate. In this example, the magnitude of the sum of the SHAP values corresponds to the distance from the center of the figure simulating the heart. The greater the distance from the center, the higher the identified importance.

[0050] The above configuration makes it easy to intuitively understand which part of the heart's feature parameters contributed to the inference result made on the electrocardiogram. In the example shown in Fig. 7, the inference result "possibly causing pAF" can be analyzed to suspect some abnormality in the atrium. In the example shown in Fig. 8, the inference result "possibly causing pAF" can be analyzed to suspect some abnormality in the lateral wall of the heart.

[0051] The processor 122 of the processing device 12 having the above-described functions may be realized by a general-purpose microprocessor operating in cooperation with general-purpose memory. Examples of the general-purpose microprocessor include a CPU, an MPU, and a GPU. Examples of the general-purpose memory include a ROM and a RAM. In this case, a computer program for executing the above-described processing may be stored in the ROM. The ROM is an example of a non-transitory computer-readable medium for storing a computer program. The general-purpose microprocessor specifies at least a portion of the program stored in the ROM, expands it in the RAM, and executes the above-described processing in cooperation with the RAM. The computer program may be pre-installed in the general-purpose memory, or may be downloaded from an external server via a communication network and then installed in the general-purpose memory. In this case, the external server is an example of a non-transitory computer-readable medium for storing a computer program.

[0052] The processor 122 of the processing device 12 having the above-described functions may be realized by a dedicated integrated circuit capable of executing the above-described computer program, such as a microcontroller, an ASIC, or an FPGA. In this case, the computer program is pre-installed in a memory element included in the dedicated integrated circuit. The memory element is an example of a computer-readable medium storing a computer program. The processor 122 of the processing device 12 having the above-described functions may also be realized by a combination of a general-purpose microprocessor and a dedicated integrated circuit.

[0053] The above-described embodiments are merely examples for facilitating understanding of the present invention, and the configurations according to the above-described embodiments may be appropriately modified or improved without departing from the spirit and scope of the present invention.

[0054] When generating the indexes to be output by the output device 13, the LIME (Local Interpretable Model-agnostic Explanation) algorithm may be used instead of the SHAP algorithm.

[0055] In the above embodiment, the output device 13 constitutes a part of the electrocardiograph 10. However, the output device 13 may be a device provided independently of the electrocardiograph 10, as long as the output device 13 is capable of receiving the index data ID from the processing device 12 via wired or wireless communication.

[0056] A part of the processing performed by the reception device 11 in the above embodiment may be executed by the processor 122 of the processing device 12. For example, the processing of obtaining the value PV of each characteristic parameter from the waveform data WD may be executed by the processor 122.

[0057] In the above embodiment, the processing device 12 infers whether or not the subject 30 is likely to develop paroxysmal atrial fibrillation. However, in addition to or instead of paroxysmal atrial fibrillation, the processing device 12 may infer whether or not the subject 30 is likely to develop other cardiac diseases. Examples of other cardiac diseases include atrial premature contractions, paroxysmal supraventricular tachycardia, ventricular premature contractions, ventricular tachycardia, ventricular fibrillation, and myocardial infarction.

[0058] The possibility of developing a disease other than cardiac disease may also be the subject of inference. An example of the other disease is epileptic seizures. When the possibility of developing an epileptic seizure is the subject of inference, characteristic parameters related to alpha waves, beta waves, theta waves, etc. are acquired based on the electroencephalogram waveform acquired from the subject by an electroencephalogram (EEG), and the importance of the characteristic parameters relative to the inference result is determined. In other words, the electroencephalogram waveform is an example of bioinformation. The electroencephalogram (EEG) is an example of a bioinformation acquisition device.

[0059] Other examples of other diseases include hypotension and hypertension. When the possibility of developing hypotension or hypertension is to be inferred, characteristic parameters such as the systolic peak time, the diastolic time, and the cardiac output are acquired based on the blood pressure waveform acquired from the subject by a blood pressure monitor, and the importance of the characteristic parameters to the inference result is determined. In other words, the blood pressure waveform is an example of biometric information. The blood pressure monitor is an example of a biometric information acquisition device.

[0060] Another example of the other disease is sleep apnea syndrome. When the possibility of developing sleep apnea syndrome is to be inferred, characteristic parameters such as exhalation time, inhalation time, and breathing time are acquired based on a respiratory waveform acquired from a subject by a respirometer, and the importance of the characteristic parameters to the inference result is determined. In other words, the respiratory waveform is an example of bioinformation. The respirometer is an example of a bioinformation acquisition device.

[0061] The expression "at least one of A and B" used in this specification with respect to two entities A and B includes cases where only A is specified, where only B is specified, and where both A and B are specified. Each entity A and B may be singular or plural unless otherwise specified.

[0062] The expression "at least one of A, B, and C" used in this specification with respect to three entities A, B, and C includes cases where only A is specified, where only B is specified, where only C is specified, where A and B are specified, where B and C are specified, where A and C are specified, and where all of A, B, and C are specified. Each of the entities A, B, and C may be singular or plural unless otherwise specified. The same applies when there are four or more entities being described. [Explanation of symbols]

[0063] 10: Electrocardiograph, 11: Reception device, 12: Processing device, 121: Input interface, 122: Processor, 123: Machine learning model, 13: Output device, 20: Electrode, 30: Subject, ID: Index data, PV: Value of feature parameter, RS: Inference result, WD: Waveform data, WF: Measured waveform

Claims

1. A biological information acquisition device for acquiring biological information of a subject, an accepting device that accepts waveform data corresponding to a measured waveform of the biological information from a sensor and acquires values of a plurality of characteristic parameters associated with the measured waveform based on the waveform data; a processing device that inputs the values of the plurality of feature parameters into a machine-learned model to obtain an inference result regarding the probability that the waveform data will be classified into one of a plurality of classes, and identifies the importance of each of the plurality of feature parameters with respect to the inference result; an output device that outputs an index indicating the name of at least one of the plurality of feature parameters and the importance level determined for the at least one feature parameter; It is equipped with the output device outputs the index in addition to the inference result in a manner that allows a distinction to be made between the importance of a feature parameter that positively contributed to the inference result and the importance of a feature parameter that negatively contributed to the inference result. Biometric information acquisition device.

2. the index is configured to enable a comparison of the relative importance levels determined for at least two of the plurality of feature parameters. The biometric information acquisition device according to claim 1 .

3. The index is configured to indicate the importance identified using SHAP (SHapley Additive exPlanations). The biometric information acquisition device according to claim 2 .

4. A biological information acquisition device for acquiring biological information of a subject, an accepting device that accepts waveform data corresponding to a measured waveform of the biological information from a sensor and acquires values of a plurality of characteristic parameters associated with the measured waveform based on the waveform data; a processing device that inputs values of the plurality of feature parameters into a machine-learned model to obtain an inference result for at least one of a plurality of classes into which the waveform data is classified, and identifies the importance of each of the plurality of feature parameters with respect to the inference result; an output device that outputs an index indicating a result of adding the importance levels determined for a plurality of feature parameters associated with each of a plurality of body parts of the subject; Equipped with Biometric information acquisition device.

5. the index includes a value of the at least one characteristic parameter. The biometric information acquisition device according to claim 1 .

6. The machine-learned model is generated by machine learning using a neural network. The biometric information acquisition device according to claim 1 .

7. A processing device for processing biological information of a subject, an interface that receives values of a plurality of characteristic parameters that are acquired based on waveform data corresponding to a measured waveform of the biological information and that are associated with the measured waveform; a processor that inputs the values of the plurality of feature parameters into a machine-learned model to obtain an inference result regarding the probability that the waveform data will be classified into one of a plurality of classes, identifies the importance of each of the plurality of feature parameters with respect to the inference result, and outputs, to an output device, index data corresponding to an index indicating the name of at least one of the plurality of feature parameters and the importance identified for the at least one feature parameter; It is equipped with the processor causes the output device to output the index in addition to the inference result in a manner that allows the importance of a feature parameter that positively contributed to the inference result to be distinguished from the importance of a feature parameter that negatively contributed to the inference result. Processing equipment.

8. A computer program executable by a processor installed in a processing device that processes biological information of a subject, When executed, the processing device: receiving values of a plurality of characteristic parameters that are acquired based on waveform data corresponding to a measured waveform of the biological information and that are associated with the measured waveform; inputting the values of the plurality of feature parameters into a machine-learned model to obtain an inference result regarding the probability that the waveform data will be classified into one of a plurality of classes; Identifying the importance of each of the plurality of feature parameters to the inference result; outputting, to an output device, index data corresponding to an index indicating the name of at least one of the plurality of feature parameters and the importance level specified for the at least one feature parameter; outputting the index to the output device in a manner that allows the importance of a feature parameter that positively contributed to the inference result to be distinguished from the importance of a feature parameter that negatively contributed to the inference result in addition to the inference result; Computer program.

9. A processing device for processing biological information of a subject, an interface that receives values of a plurality of characteristic parameters that are acquired based on waveform data corresponding to a measured waveform of the biological information and that are associated with the measured waveform; a processor that inputs the values of the plurality of feature parameters into a machine-learned model to obtain an inference result for at least one of a plurality of classes into which the waveform data is classified, identifies the importance of each of the plurality of feature parameters with respect to the inference result, and outputs to an output device an index indicating the result obtained by adding up the identified importance of the plurality of feature parameters associated with each of a plurality of body parts of the subject; Equipped with Processing equipment.

10. A computer program executable by a processor installed in a processing device that processes biological information of a subject, When executed, the processing device: receiving values of a plurality of characteristic parameters associated with the measured waveform, the characteristic parameters being acquired based on waveform data corresponding to the measured waveform of the biological information; inputting the values of the plurality of feature parameters into a machine-learned model to obtain an inference result for at least one of a plurality of classes into which the waveform data is classified; Identifying the importance of each of the plurality of feature parameters to the inference result; outputting, to an output device, an index indicating a result of adding up the importance levels determined for a plurality of feature parameters associated with each of a plurality of body parts of the subject; Computer program.

Images