Exercise tolerability estimation method, exercise tolerability estimation device, and computer program

Abstract

To provide an exercise tolerability estimation method, an exercise tolerability estimation device, and a computer program capable of estimating exercise tolerability that reflects a subject's physical condition in detail, without imposing exercise load on the subject.SOLUTION: An exercise tolerability estimation method comprises: a subject information acquisition step of acquiring subject information related to a subject; an estimation model storage step of storing estimation models generated on the basis of subject information of other subjects and exercise tolerability; an exercise tolerability estimation step of estimating the subject's exercise tolerability on the basis of the subject information acquired in the subject information acquisition step and the estimation models stored in the estimation model storage step; and an output step of outputting the exercise tolerability estimated in the exercise tolerability estimation step.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present invention relates to a method for estimating exercise tolerance, an apparatus for estimating exercise tolerance, and a computer program.

Background Art

[0002] Estimating the state based on information obtained by observing a living body such as a human body is widely performed. An example is the estimation of exercise tolerance. Estimation of exercise tolerance is essential for setting optimal exercise goals according to an individual's condition in effective training for improving cardiopulmonary function and in prescribing rehabilitation after treatment of heart disease. In Non-Patent Document 1, it has been proposed to observe the resting heart rate and determine the target heart rate during exercise. Further, in Non-Patent Document 2, a method for evaluating exercise tolerance by a cardiopulmonary exercise test (hereinafter referred to as CPX) has been proposed.

Prior Art Documents

Non-Patent Documents

[0003]

Non-Patent Document 1

Non-Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, the method described in Non-Patent Document 1 estimates an approximate target heart rate during exercise based on the resting heart rate and age, and does not reflect the physical condition in detail. Therefore, it is difficult to evaluate the exercise tolerance that reflects the physical condition of each subject in detail by the method described in Non-Patent Document 1.

[0005] In addition, the method described in Non-Patent Document 2 generally requires a large-scale CPX test device and test system, and it is necessary to impose an exercise load close to the limit on the subject in the test. Therefore, in the method described in Non-Patent Document 2, it may be difficult to conduct the test when the test device and test system are not in place, when the cardiopulmonary function of the subject is reduced, especially immediately after the treatment of the disease.

[0006] In view of the above circumstances, an object of the present invention is to provide an exercise tolerance estimation method, an exercise tolerance estimation device, and a computer program that estimate the exercise tolerance that reflects the physical condition of a subject in detail without imposing an exercise load on the subject.

Means for Solving the Problems

[0007] One aspect of the present invention includes a subject information acquisition step of acquiring subject information regarding a subject, an estimation model accumulation step of accumulating an estimation model generated based on subject information other than the subject and exercise tolerance, and based on the subject information acquired in the subject information acquisition step and the estimation model accumulated in the estimation model accumulation step, an exercise tolerance estimation step of estimating the exercise tolerance of the subject, and an output step of outputting the exercise tolerance estimated in the exercise tolerance estimation step. It is an exercise tolerance estimation method having.

[0008] Also, one aspect of the present invention is an exercise tolerance estimation device including: a subject information acquisition unit that acquires subject information regarding a subject; an estimation model storage unit that stores an estimation model generated based on subject information other than the subject and exercise tolerance; an exercise tolerance estimation unit that estimates the exercise tolerance of the subject based on the subject information acquired by the subject information acquisition unit and the estimation model stored in the estimation model storage unit; and an output unit that outputs the exercise tolerance estimated by the exercise tolerance estimation unit.

Advantages of the Invention

[0009] According to the present invention, it is possible to estimate the exercise tolerance that reflects the physical state of the subject in detail without imposing an exercise load on the subject.

Brief Description of the Drawings

[0010]

Figure 1

Figure 2

Figure 3

Figure 4A

Figure 4B

Figure 4C

Figure 4D

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Embodiments for Carrying Out the Invention

[0011] Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[0012] [First Embodiment] First, the first embodiment of the present invention will be described.

[0013] FIG. 1 is a schematic configuration diagram of an exercise tolerance estimation system 10a according to a first embodiment of the present invention. The exercise tolerance estimation system 10a includes an exercise tolerance estimation device 100a and an estimation model generation device 200a. The exercise tolerance estimation device 100a and the estimation model generation device 200a are connected by wire or wirelessly, and it is possible to transmit and receive data between the estimation model generation device 200a and the exercise tolerance estimation device 100a. Here, the exercise tolerance estimation device 100a and the estimation model generation device 200a are regarded as different devices, but they may be configured as a single device. Note that the exercise tolerance means the ability of the body to withstand exercise to what extent, and it is an index for examining the functions of the whole body such as the heart, lungs, and muscles.

[0014] The exercise tolerance estimation device 100a includes a blood test value acquisition unit 101 (also referred to as a subject information acquisition unit), an exercise tolerance estimation unit 102, an estimation model storage unit 103, and an output unit 104. The blood test value acquisition unit 101 includes an input device such as a keyboard or a touch panel. The blood test value acquisition unit 101 is connected to the exercise tolerance estimation unit 102. The blood test value acquisition unit 101 acquires blood test values (also referred to as subject information) of a subject for whom exercise tolerance is to be estimated based on the operation of the user.

[0015] The blood test values acquired by the blood test value acquisition unit 101 include values related to at least one of CRP, NT-pro-BNP, BNP, CK, HDL cholesterol, triglyceride, creatinine, blood glucose, albumin, white blood cell count, Hb (hemoglobin), platelet count, and hemoglobin A1c. CRP is C-reactive protein. If the value of CRP increases too much, there may be an inflammatory disease or necrosis of human body tissues.

[0016] NT-pro-BNP is a type of hormone secreted by the heart. If the value of NT-pro-BNP increases too much, it may indicate a heavy burden on the heart. BNP is a hormone secreted by the heart to protect the heart. If the value of BNP increases too much, it may indicate a decline in heart function. CK refers to creatine kinase, which is an enzyme present in muscles and other tissues. If the value of CK increases too much, it may indicate that muscle cells are damaged.

[0017] HDL cholesterol is "good cholesterol" and functions to prevent arteriosclerosis by collecting excess cholesterol in the blood and returning it to the liver. If the value of HDL cholesterol increases too much, arteriosclerosis may be progressing. Triglycerides are one of the lipids present in the body and serve as an energy source when a person is active. If the value of triglycerides increases too much, it may cause health problems.

[0018] Creatinine is a waste product after muscles produce the energy needed for movement and is excreted by the kidneys. If there is an abnormality in the creatinine value, the kidney function may be declining. Blood sugar refers to glucose contained in the blood, which is transported throughout the body by the blood and serves as an energy source for the cells that make up the body. If an abnormality occurs in the blood sugar value, there is a possibility of developing diabetes or hypoglycemia. Albumin is a protein mainly produced by the liver. If an abnormality occurs in the albumin value, the liver function may be declining.

[0019] The white blood cell count represents the number of white blood cells in the blood. If the white blood cell count is too high, there may be a risk of developing bacterial infections, cancer, leukemia, etc. On the other hand, if the white blood cell count is too low, there may be a risk of developing severe infections or aplastic anemia. Hb (hemoglobin) is a substance present in the blood. If the Hb (hemoglobin) value is too high, there may be a risk of developing polycythemia, and if it is too low, there may be a risk of developing anemia. The platelet count represents the number of platelets in the blood. If the platelet count is too high, blood clots are more likely to form in the blood vessels, increasing the risk of developing myocardial infarction, cerebral infarction, etc. On the other hand, if the platelet count is too low, bleeding is difficult to stop. Hemoglobin A1c is a combination of hemoglobin in red blood cells and glucose in the blood. If the Hemoglobin A1c value is too high, there may be a risk of developing diabetes.

[0020] Returning to the description of FIG. 1, the exercise tolerance estimation unit 102 is composed of a CPU (Central Processing Unit) etc., and is connected to the blood test value acquisition unit 101, the estimation model storage unit 103, and the output unit 104. The exercise tolerance estimation unit 102 estimates the exercise tolerance of the subject based on the blood test values of the subject and the estimation model. Here, the exercise tolerance estimation unit 102 estimates the exercise tolerance by using an estimation model learned by machine learning with the previously performed CPX test results as teacher data.

[0021] The exercise tolerance estimated by the exercise tolerance estimation unit 102 includes at least one of the heart rate, oxygen uptake, maximum oxygen uptake, VE (minute ventilation) - VCO2 (carbon dioxide excretion) SLOPE, systolic blood pressure, diastolic blood pressure, end-expiratory oxygen concentration, and gas exchange ratio at 1 minute before AT (Anaerobic Threshold), at AT, and during maximum exercise. Note that the exercise tolerance estimated by the exercise tolerance estimation unit 102 may include other data other than the heart rate, oxygen uptake, maximum oxygen uptake, VE - VCO2 SLOPE, systolic blood pressure, diastolic blood pressure, end-expiratory oxygen concentration, and gas exchange ratio.

[0022] Here, AT means the anaerobic work threshold, that is, the boundary between anaerobic exercise and aerobic exercise. That is, the heart rate at AT means the heart rate when reaching AT during exercise load. Also, VE-VCO2SLOPE is a value indicating the increasing rate of ventilation volume with respect to carbon dioxide emission amount when performing a CPX (cardiopulmonary exercise load test) or the like.

[0023] Returning to the description of FIG. 1, the estimated model storage unit 103 is composed of a memory or the like, is connected to the exercise tolerance estimation unit 102, and is also connected to an estimated model generation device 200a installed outside the exercise tolerance estimation device 100a. The estimated model storage unit 103 stores an estimated model used for estimating basic numerical values (such as blood test numerical values) related to the exercise tolerance of the subject.

[0024] The output unit 104 is composed of a display or the like and is connected to the exercise tolerance estimation unit 102. The output unit 104 outputs (displays) the exercise tolerance of the subject on the screen of the display. The estimated model generation device 200a generates an estimated model learned by machine learning using the already performed CPX test results as teacher data. When the estimated model generation device 200a generates an estimated model, known machine learning methods can be used. As known machine learning methods, for example, linear regression, decision tree, deep learning, etc. can be used.

[0025] In the first embodiment, a case where the estimation model generation device 200a uses a decision tree when generating an estimation model will be described. A decision tree is a method of making a case-by-case distinction based on the values of features and obtaining an estimated value while sequentially repeating this, and it can be created by a known method. Here, an example in which a decision tree is learned using a method called gradient boosting will be described. In this learning, the data of 8225 people who actually underwent the CPX test was used as a sample. Without duplication, 64% was used for learning, 16% for validation, and 20% for testing. That is, for the sample data, the allocation as learning, validation, and testing was randomly changed 30 times for testing. FIG. 4A shows an example of subject information and the feature amounts related to blood test numerical values. Each data was composed of a plurality of feature amounts including those shown in FIG. 4A and one of the basic numerical values related to the exercise tolerance included in the CPX test result, the heart rate at AT. The feature amounts shown in FIG. 4A include the above-described blood test numerical values.

[0026] The learning of the estimation model by the estimation model generation device 200a was performed by sequentially selecting features. First, the selected feature set was initialized with an empty set. Next, among the features that have not yet been selected, when the feature is included in the feature set and the value of the feature that has not been selected is set as missing, the operation of selecting the feature that achieves the minimum estimation error using the trained model was repeated until all features were selected. As a result, the features were ranked in the order considered to be important for estimation. For each of the features thus selected, a branching point was set based on the distribution of the learning data, and a decision tree was sequentially constructed. The estimation model storage unit 103 stores the decision tree thus generated as an estimation model.

[0027] The estimation model generation device 200a can acquire data on the blood test values of a large number of humans and data on the exercise tolerance of those large number of humans, and based on those data, generate an estimation model that estimates, by machine learning the relationship between blood test values and exercise tolerance, what exercise load is preferable for what blood test values.

[0028] Figure 2 is a flowchart showing the processing of the exercise tolerance estimation device 100a according to the first embodiment of the present invention. First, the estimation model storage unit 103 of the exercise tolerance estimation device 100a acquires and stores the estimation model generated by the estimation model generation device 200a (step S11). Before the processing of the exercise tolerance estimation device 100a according to the first embodiment shown in Figure 2 is performed, in the estimation model generation device 200a, data on features related to blood test values (for example, CRP, NT-pro-BNP (or BNP), CK, HDL cholesterol, triglycerides, creatinine, blood glucose, albumin, white blood cell count, Hb (hemoglobin), platelet count, hemoglobin A1c) as shown in Figure 4A are acquired for a large number of humans, and for those large number of humans, data related to exercise tolerance are acquired. In the estimation model generation device 200a, based on those data, an estimation model is generated that estimates, by machine learning the relationship between blood test values and exercise tolerance, what exercise load is preferable for what blood test values.

[0029] Next, the blood test value acquisition unit 101 of the exercise tolerance estimation device 100a acquires the blood test values of the subject for whom the exercise tolerance is to be estimated (step S12). The process of step S12 is, for example, to display an image as shown in FIG. 3 on the display of the exercise tolerance estimation device 100a, and the user of the exercise tolerance estimation device 100a inputs the necessary items (for example, age, gender, height, weight, current smoking history, past smoking history, disease name, systolic blood pressure, diastolic blood pressure, resting heart rate, blood test) while looking at the image. In FIG. 3, the case where the values of NT-pro-BNP and BNP are input as the blood test values acquired by the blood test value acquisition unit 101 is shown. However, it is also preferable to input other values (that is, CRP, CK, HDL cholesterol, triglyceride, creatinine, blood sugar, albumin, white blood cell count, Hb (hemoglobin), platelet count, hemoglobin A1c) shown in FIG. 4A. By increasing the number of items to be input, it becomes possible to more accurately estimate the exercise tolerance of the subject.

[0030] However, as the blood test values acquired by the blood test value acquisition unit 101, it is not always necessary to input all of the feature amounts related to the blood test values as shown in FIG. 4A (that is, CRP, NT-pro-BNP (or BNP), CK, HDL cholesterol, triglyceride, creatinine, blood sugar, albumin, white blood cell count, Hb (hemoglobin), platelet count, hemoglobin A1c), and at least one of them may be input. By doing so, it becomes possible to estimate the exercise tolerance even if the data on all the feature amounts related to the blood test values as shown in FIG. 4A is not complete.

[0031] Here, although the case where the user views an image as shown in FIG. 3 displayed on the display of the exercise tolerance estimation device 100a and inputs numerical values such as blood test values has been described, it is not limited to this. For example, by accessing an electronic record (such as a database or an electronic medical record) in which past blood test values are stored with the exercise tolerance estimation device 100a, and obtaining numerical values such as blood test values, the blood test value acquisition unit 101 may obtain blood test values and the like.

[0032] Returning to the description of FIG. 2, after the process of step S12 is performed, the exercise tolerance estimation unit 102 of the exercise tolerance estimation device 100a estimates the exercise tolerance based on the blood test values of the subject acquired by the blood test value acquisition unit 101 and the estimation model stored in the estimation model storage unit 103 (step S13). By using the estimation model stored in the estimation model storage unit 103, it is possible to estimate what kind of exercise load is preferable for what kind of blood test values. Therefore, the exercise tolerance estimation unit 102 can estimate the preferable exercise tolerance for the blood test values of the subject acquired by the blood test value acquisition unit 101 by using the estimation model.

[0033] Next, the output unit 104 of the exercise tolerance estimation device 100a outputs the exercise tolerance of the subject estimated in step S13 (step S14). For example, the output unit 104 performs the process of step S14 by displaying an image as shown in FIG. 5 on the display of the exercise tolerance estimation device 100a as the estimated exercise tolerance of the subject. In FIG. 5, as the exercise tolerance of the subject, the load amount, VO2, METs, and heart rate at 1 minute before AT (that is, the time 1 minute before reaching AT), at the time of AT (that is, the time of reaching AT), and at the time of maximum exercise are displayed, but other items (for example, oxygen uptake, maximum oxygen uptake, VE-VCO2 SLOPE, systolic blood pressure, diastolic blood pressure, end-expiratory oxygen concentration, gas exchange ratio) may be displayed.

[0034] Here, the load amount shown in FIG. 5 is the load amount during exercise that is estimated to be preferable for the subject for whom the blood test numerical value has been input to the blood test numerical value acquisition unit 101. Also, the VO2 shown in FIG. 5 is the oxygen uptake amount during exercise that is estimated to be preferable for the subject for whom the blood test numerical value has been input to the blood test numerical value acquisition unit 101. Further, the METs shown in FIG. 5 indicates the intensity of exercise that is estimated to be preferable for the subject for whom the blood test numerical value has been input to the blood test numerical value acquisition unit 101, and shows how many times more energy is consumed compared to when the resting state is set to 1. Also, the heart rate shown in FIG. 5 is the heart rate during exercise that is estimated to be preferable for the subject for whom the blood test numerical value has been input to the blood test numerical value acquisition unit 101.

[0035] Here, the case where the output unit 104 displays the exercise tolerance as shown in FIG. 5 on the display of the exercise tolerance estimation device 100a has been described, but it is not limited thereto. For example, the output unit 104 may output the exercise tolerance as electronic data or a printed matter. Further, the output unit 104 may graphically display the exercise tolerance including the transition from the past and the comparison with a group set under predetermined conditions.

[0036] According to the first embodiment described above, it is possible to estimate the exercise tolerance preferable for the subject only by acquiring the blood test numerical value of the subject without imposing an exercise load on the subject for whom the exercise tolerance is to be estimated. For example, according to the first embodiment described above, even without performing CPX, it is possible to estimate the exercise tolerance, which is a main numerical value obtained by CPX, from the blood test numerical values generally obtained without causing the subject to exercise. Therefore, even when there is no CPX testing device or testing system, or when it is difficult to perform CPX, it is possible to estimate the exercise tolerance for each subject, and it is possible to select an appropriate exercise menu or exercise prescription compared to known methods.

[0037] [Second Embodiment] Next, a second embodiment of the present invention will be described. Regarding the points where the exercise tolerance estimation system 10b according to the second embodiment is the same as the exercise tolerance estimation system 10a according to the first embodiment, their descriptions will be omitted.

[0038] FIG. 6 is a schematic configuration diagram of an exercise tolerance estimation system 10b according to a second embodiment of the present invention. The exercise tolerance estimation system 10b includes an exercise tolerance estimation device 100b and an estimation model generation device 200b. The exercise tolerance estimation device 100b includes a blood test value acquisition unit 201 (also referred to as a subject information acquisition unit), an ultrasonic examination result acquisition unit 202, an exercise tolerance estimation unit 203, an estimation model storage unit 204, and an output unit 205. Since the processing contents performed by the blood test value acquisition unit 201, the estimation model storage unit 204, and the output unit 205 included in the exercise tolerance estimation device 100b according to the second embodiment are the same as the processing contents performed by the blood test value acquisition unit 101, the estimation model storage unit 103, and the output unit 104 included in the exercise tolerance estimation device 100a according to the first embodiment, the descriptions thereof are omitted.

[0039] The ultrasonic examination result acquisition unit 202 includes an input device such as a keyboard or a touch panel. The ultrasonic examination result acquisition unit 202 is connected to the exercise tolerance estimation unit 203. The ultrasonic examination result acquisition unit 202 acquires the ultrasonic examination result of the subject for whom the exercise tolerance is to be estimated based on the operation of the user. The ultrasonic examination result acquired by the ultrasonic examination result acquisition unit 202 includes at least one of the subject's vena cava diameter, TAPSE (Tricuspid Annular Plane Systolic Excursion), ventricular septal thickness, right ventricular end-diastolic area, ascending aortic diameter, right ventricular end-systolic area, maximum left atrial volume, maximum left atrial volume coefficient, left ventricular outflow tract, TRPG (Tricuspid Regurgitation Peak Gradient), and the examination image.

[0040] Returning to the description of FIG. 6, the exercise tolerance estimating unit 203 is configured by a CPU or the like and is connected to the blood test value acquisition unit 201, the ultrasonic examination result acquisition unit 202, the estimation model storage unit 204, and the output unit 205. The exercise tolerance estimating unit 203 estimates the exercise tolerance of the subject based on the blood test values of the subject, the ultrasonic examination results, and the estimation model. Here, the exercise tolerance estimating unit 203 estimates the exercise tolerance by using an estimation model learned by machine learning with the already performed CPX examination results and ultrasonic examination results as teacher data.

[0041] The estimation model generation device 200b generates an estimation model learned by machine learning with the already performed CPX examination results and ultrasonic examination results as teacher data. When the estimation model generation device 200b generates an estimation model, known machine learning methods can be used. The estimation model generation device 200b acquires data on blood test values of a large number of people, data on ultrasonic examination results, and data on the exercise tolerance of those large number of people, and based on those data, by machine learning the relationship between the blood test values, the ultrasonic examination results, and the exercise tolerance, it is possible to generate an estimation model for estimating what kind of exercise load is preferable for what kind of blood test values and ultrasonic examination results.

[0042] FIG. 7 is a flowchart showing the processing of the exercise tolerance estimation device 100b according to the second embodiment of the present invention. First, the estimation model storage unit 204 of the exercise tolerance estimation device 100b acquires and stores the estimation model generated by the estimation model generation device 200b (step S21). Note that before the processing of the exercise tolerance estimation device 100b according to the second embodiment shown in FIG. 7 is performed, in the estimation model generation device 200b, data on features related to ultrasonic examination results as shown in FIG. 4B (for example, diameter of the vena cava, TAPSE, ventricular septal thickness, right ventricular end-diastolic area, ascending aortic diameter, right ventricular end-systolic area, maximum left atrial volume, maximum left atrial volume coefficient, left ventricular outflow tract, TRPG, etc.) are acquired for a large number of people, and for those large number of people, data related to exercise tolerance are acquired. In the estimation model generation device 200b, based on those data, a machine learning is performed on the relationship between blood test values, ultrasonic examination results, and exercise tolerance, and an estimation model is generated to estimate what kind of exercise load is preferable in the case of what kind of blood test values and ultrasonic examination results.

[0043] Next, the blood test value acquisition unit 201 of the exercise tolerance estimation device 100b acquires the blood test values of the subject for whom exercise tolerance is to be estimated (step S22). Next, the ultrasonic examination result acquisition unit 202 of the exercise tolerance estimation device 100b acquires the ultrasonic examination results of the subject for whom exercise tolerance is to be estimated (step S23). Next, the exercise tolerance estimation unit 203 of the exercise tolerance estimation device 100b estimates the exercise tolerance based on the blood test values of the subject acquired by the blood test value acquisition unit 201, the ultrasonic examination results acquired by the ultrasonic examination result acquisition unit 202, and the estimation model stored in the estimation model storage unit 204 (step S24).

[0044] If the estimation model stored in the estimation model storage unit 204 is used, it is possible to estimate what kind of exercise load is preferable for any blood test values and ultrasonic test results. Therefore, the exercise tolerance estimation unit 203 can estimate the preferable exercise tolerance for the blood test values and ultrasonic test results of the subject acquired by the blood test value acquisition unit 201 using the estimation model. Next, the output unit 205 of the exercise tolerance estimation device 100b outputs the exercise tolerance of the subject estimated in step S24 (step S25).

[0045] According to the second embodiment described above, it is possible to estimate the preferable exercise tolerance for the subject only by acquiring the blood test values and ultrasonic test results of the subject without imposing an exercise load on the subject for whom the exercise tolerance is to be estimated. For example, according to the second embodiment described above, even without performing CPX, it is possible to estimate the exercise tolerance, which is the main value obtained by CPX, from the blood test values and ultrasonic test results generally obtained without exercising the subject. Therefore, even when there is no CPX test device or test system, or when it is difficult to perform CPX, it is possible to estimate the exercise tolerance for each subject, and it is possible to select an appropriate exercise menu or exercise prescription compared to known methods.

[0046] [Third Embodiment] Next, a third embodiment of the present invention will be described. Regarding the points where the exercise tolerance estimation system 10c according to the third embodiment is the same as the exercise tolerance estimation system 10a according to the first embodiment, their descriptions will be omitted.

[0047] FIG. 8 is a schematic configuration diagram of an exercise tolerance estimation system 10c according to the third embodiment of the present invention. The exercise tolerance estimation system 10c includes an exercise tolerance estimation device 100c and an estimation model generation device 200c. The exercise tolerance estimation device 100c includes a blood test value acquisition unit 301 (also referred to as a subject information acquisition unit), a drug information acquisition unit 302, an exercise tolerance estimation unit 303, an estimation model storage unit 304, and an output unit 305. Since the processing contents performed by the blood test value acquisition unit 301, the estimation model storage unit 304, and the output unit 305 included in the exercise tolerance estimation device 100c according to the third embodiment are the same as the processing contents performed by the blood test value acquisition unit 101, the estimation model storage unit 103, and the output unit 104 included in the exercise tolerance estimation device 100a according to the first embodiment, the descriptions thereof are omitted.

[0048] The drug information acquisition unit 302 includes an input device such as a keyboard or a touch panel. The drug information acquisition unit 302 is connected to the exercise tolerance estimation unit 303. The drug information acquisition unit 302 acquires drug information of a subject for whom exercise tolerance is to be estimated based on the operation of the user. The drug information includes information such as the presence or absence of a prescription for a beta-blocker in the subject and the potency of the prescribed beta-blocker. Note that the drug information acquisition unit 302 acquires the potency of the beta-blocker by using a conversion table that converts the prescribed dose into potency.

[0049] The exercise tolerance estimation unit 303 is configured by a CPU or the like and is connected to the blood test value acquisition unit 301, the drug information acquisition unit 302, the estimation model storage unit 304, and the output unit 305. The exercise tolerance estimation unit 303 estimates the exercise tolerance of the subject based on the blood test values of the subject, the drug information, and the estimation model. Here, the exercise tolerance estimation unit 303 estimates the exercise tolerance by using an estimation model learned by machine learning using the previously performed CPX test results and drug information as teacher data.

[0050] The estimation model generation device 200c generates an estimation model learned by machine learning using the already performed CPX test results and drug information as teacher data. When the estimation model generation device 200c generates an estimation model, a known machine learning method can be used. The estimation model generation device 200c acquires data on the blood test values of a large number of people, data on drug information, and data on the exercise tolerance of those large number of people, and based on those data, by machine learning the relationship between the blood test values, drug information, and exercise tolerance, it is possible to generate an estimation model that estimates what exercise load is preferable for what blood test values and drug information.

[0051] Figure 9 is a flowchart showing the processing of the exercise tolerance estimation device 100c according to the third embodiment of the present invention. First, the estimation model storage unit 304 of the exercise tolerance estimation device 100c acquires and stores the estimation model generated by the estimation model generation device 200c (step S31). Before the processing of the exercise tolerance estimation device 100c according to the third embodiment shown in Figure 9 is performed, in the estimation model generation device 200c, data on features related to drug information (for example, drugs, etc.) as shown in Figure 4C is acquired for a large number of people, and for those large number of people, data related to exercise tolerance is acquired. In the estimation model generation device 200c, based on those data, by machine learning the relationship between the blood test values, drug information, and exercise tolerance, an estimation model is generated that estimates what exercise load is preferable for what blood test values and drug information.

[0052] Next, the blood test value acquisition unit 301 of the exercise tolerance estimation device 100c acquires the blood test values of the subject for whom the exercise tolerance is to be estimated (step S32). Next, the drug information acquisition unit 302 of the exercise tolerance estimation device 100c acquires the drug information of the subject for whom the exercise tolerance is to be estimated (step S33). Next, the exercise tolerance estimation unit 303 of the exercise tolerance estimation device 100c estimates the exercise tolerance based on the blood test values of the subject acquired by the blood test value acquisition unit 301, the drug information acquired by the drug information acquisition unit 302, and the estimation model stored in the estimation model storage unit 304 (step S34).

[0053] By using the estimation model stored in the estimation model storage unit 304, it is possible to estimate what kind of exercise load is preferable for any blood test values and drug information. Therefore, the exercise tolerance estimation unit 303 can estimate the preferable exercise tolerance for the blood test values and drug information of the subject acquired by the blood test value acquisition unit 301 by using the estimation model. Next, the output unit 305 of the exercise tolerance estimation device 100c outputs the exercise tolerance of the subject estimated in step S34 (step S35).

[0054] According to the above-described third embodiment, it is possible to estimate the preferable exercise tolerance for the subject only by acquiring the blood test values and drug information of the subject without imposing an exercise load on the subject for whom the exercise tolerance is to be estimated. For example, according to the above-described third embodiment, even without performing CPX, it is possible to estimate the exercise tolerance, which is the main value obtained by CPX, from the blood test values and drug information that can be generally obtained without causing the subject to exercise. Therefore, even when there is no CPX testing device or testing system, or when it is difficult to perform CPX, it is possible to estimate the exercise tolerance for each subject, and it is possible to select an appropriate exercise menu or exercise prescription compared to known methods.

[0055] [Fourth Embodiment] Next, a fourth embodiment of the present invention will be described. Regarding the points where the exercise tolerance estimation system 10d according to the fourth embodiment is the same as the exercise tolerance estimation systems 10a, 10b, and 10c according to the first to third embodiments, their descriptions will be omitted.

[0056] FIG. 10 is a schematic configuration diagram of an exercise tolerance estimation system 10d according to a fourth embodiment of the present invention. The exercise tolerance estimation system 10d includes an exercise tolerance estimation device 100d and an estimation model generation device 200d. The exercise tolerance estimation device 100d includes a blood test value acquisition unit 401 (also referred to as a subject information acquisition unit), an ultrasonic examination result acquisition unit 402, a drug information acquisition unit 403, an exercise tolerance estimation unit 404, an estimation model storage unit 405, and an output unit 406.

[0057] Since the processing contents performed by the blood test value acquisition unit 401, the estimation model storage unit 405, and the output unit 406 included in the exercise tolerance estimation device 100d according to the fourth embodiment are the same as the processing contents performed by the blood test value acquisition unit 101, the estimation model storage unit 103, and the output unit 104 included in the exercise tolerance estimation device 100a according to the first embodiment, their descriptions will be omitted.

[0058] Also, since the processing content performed by the ultrasonic examination result acquisition unit 402 included in the exercise tolerance estimation device 100d according to the fourth embodiment is the same as the processing content performed by the ultrasonic examination result acquisition unit 202 included in the exercise tolerance estimation device 100b according to the second embodiment, its description will be omitted. Also, since the processing content performed by the drug information acquisition unit 403 included in the exercise tolerance estimation device 100d according to the fourth embodiment is the same as the processing content performed by the drug information acquisition unit 302 included in the exercise tolerance estimation device 100c according to the third embodiment, its description will be omitted.

[0059] The exercise tolerance estimation unit 404 is composed of a CPU or the like and is connected to the blood test value acquisition unit 401, the ultrasonic examination result acquisition unit 202, the drug information acquisition unit 403, the estimation model storage unit 405, and the output unit 406. The exercise tolerance estimation unit 404 estimates the exercise tolerance of the subject based on the blood test values of the subject, the ultrasonic examination results, the drug information, and the estimation model. Here, the exercise tolerance estimation unit 404 estimates the exercise tolerance by using an estimation model learned by machine learning with the CPX test results, ultrasonic examination results, and drug information that have already been performed as teacher data.

[0060] The estimation model generation device 200d generates an estimation model learned by machine learning with the CPX test results, ultrasonic examination results, and drug information that have already been performed as teacher data. When the estimation model generation device 200d generates an estimation model, a known machine learning method can be used. The estimation model generation device 200d acquires data on the blood test values of a large number of people, data on the ultrasonic examination results, data on the drug information, and data on the exercise tolerance of those large number of people, and based on those data, learns the relationship between the blood test values, the ultrasonic examination results, the drug information, and the exercise tolerance by machine learning, so as to generate an estimation model that estimates what exercise load is preferable for what blood test values, ultrasonic examination results, and drug information.

[0061] FIG. 11 is a flowchart showing the processing of the exercise tolerance estimation device 100d according to the fourth embodiment of the present invention. First, the estimation model storage unit 405 of the exercise tolerance estimation device 100d acquires and stores the estimation model generated by the estimation model generation device 200d (step S41). Note that before the processing of the exercise tolerance estimation device 100d according to the fourth embodiment shown in FIG. 11 is performed, in the estimation model generation device 200d, data on features related to drug information as shown in FIGS. 4B and 4C is acquired for a large number of people, and for those large number of people, data related to exercise tolerance is acquired. In the estimation model generation device 200c, based on those data, by machine learning the relationship between blood test values, ultrasonic examination results, drug information, and exercise tolerance, an estimation model is generated that estimates what exercise load is preferable for what blood test values, ultrasonic examination results, and drug information.

[0062] Next, the blood test value acquisition unit 401 of the exercise tolerance estimation device 100d acquires the blood test values of the subject for whom exercise tolerance is to be estimated (step S42). Next, the ultrasonic examination result acquisition unit 402 of the exercise tolerance estimation device 100d acquires the ultrasonic examination results of the subject for whom exercise tolerance is to be estimated (step S43). Next, the drug information acquisition unit 403 of the exercise tolerance estimation device 100d acquires the drug information of the subject for whom exercise tolerance is to be estimated (step S44).

[0063] Next, the exercise tolerance estimation unit 404 of the exercise tolerance estimation device 100d estimates the exercise tolerance based on the blood test values of the subject acquired by the blood test value acquisition unit 401, the ultrasonic examination results acquired by the ultrasonic examination result acquisition unit 402, the drug information acquired by the drug information acquisition unit 403, and the estimation model stored in the estimation model storage unit 405 (step S45).

[0064] If the estimation model stored in the estimation model storage unit 405 is used, it is possible to estimate what exercise load is preferable for any blood test values, ultrasonic test results, and drug information. Therefore, the exercise tolerance estimation unit 404 can estimate the preferable exercise tolerance for the blood test values, ultrasonic test results, and drug information of the subject acquired by the blood test value acquisition unit 401 using the estimation model. Next, the output unit 406 of the exercise tolerance estimation device 100d outputs the exercise tolerance of the subject estimated in step S45 (step S46).

[0065] According to the above-described fourth embodiment, it is possible to estimate the preferable exercise tolerance for a subject only by acquiring the blood test values, ultrasonic test results, and drug information of the subject without imposing an exercise load on the subject for whom the exercise tolerance is to be estimated. For example, according to the above-described fourth embodiment, even without performing CPX, it is possible to estimate the exercise tolerance, which is a main value obtained by CPX, from the blood test values, ultrasonic test results, and drug information that can be generally obtained without exercising the subject. Therefore, even when there is no CPX testing device or testing system, or when it is difficult to perform CPX, it is possible to estimate the exercise tolerance for each subject, and it is possible to select an appropriate exercise menu or exercise prescription compared to known methods.

[0066] Note that in the fourth embodiment, the blood test value acquisition unit 401 acquires feature quantities (such as CRP, NT-pro-BNP (or BNP), CK, HDL cholesterol, triglyceride, creatinine, blood glucose, albumin, white blood cell count, Hb (hemoglobin), platelet count, hemoglobin A1c, etc.) as shown in FIG. 4A, the ultrasonic examination result acquisition unit 402 uses feature quantities (such as vena cava diameter, TAPSE, ventricular septal thickness, right ventricular end-diastolic area, ascending aortic diameter, right ventricular end-systolic area, maximum left atrial volume, maximum left atrial volume coefficient, left ventricular outflow tract, TRPG, etc.) as shown in FIG. 4B, and the drug information acquisition unit 403 uses feature quantities (such as beta-blockers, etc.) as shown in FIG. 4C, but it is not limited thereto. For example, the estimated model generation device 200d may generate an estimated model for estimating exercise tolerance using feature quantities (such as resting heart rate, disease name, age, resting systolic blood pressure, resting diastolic blood pressure, weight, height) as shown in FIG. 4D in addition to the feature quantities as shown in FIGS. 4A, 4B, and 4C. By doing so, it is possible to further estimate exercise tolerance with higher accuracy. That is, when estimating exercise tolerance, the feature quantities shown in FIG. 4D may be used as subject information.

[0067] Next, an experiment for confirming the effect when the exercise tolerance estimation system 10d according to the fourth embodiment is used will be described. The exercise tolerance to be estimated is the heart rate at AT, which is obtained using the learning data and test data described above. The estimation accuracy was evaluated by MAE (Mean Absolute Error).

[0068] FIG. 12 shows the distribution of the heart rate at AT according to the present invention for the test subjects. For each subject, the true value (heart rate at AT measured by the actual CPX test) is taken on the horizontal axis, and the estimated value according to the present invention is plotted on the vertical axis, and the distribution is shown by contour lines. The estimated value corresponds to the exercise tolerance estimated by the exercise tolerance estimation unit 404. In FIG. 12, the straight line with a slope of 1 represents a state without estimation error. However, in a wide range, the subjects are distributed evenly around it. Moreover, most of the subjects are within the range of an error of about 10 heartbeats. It is shown that extremely useful results can be obtained by using the exercise tolerance estimation system 10d according to the fourth embodiment.

[0069] On the other hand, FIG. 13 shows the distribution of the heart rate at AT obtained by a known method (Non-Patent Document 1) for the same subjects as in FIG. 12 for comparison, and the heart rate at AT obtained by the following formula.

[0070] (220 - age - resting heart rate) × 0.7 + resting heart rate

[0071] Comparing FIG. 12 (the fourth embodiment) and FIG. 13 (Non-Patent Document 1), it can be said that the estimation according to FIG. 12 (the fourth embodiment) is more accurate.

[0072] FIG. 14 is a graph showing the relationship between the number of input features and MAE in the fourth embodiment of the present invention. In FIG. 14, the number of features used for estimation is shown on the horizontal axis, and MAE is shown on the vertical axis. FIG. 14 shows that when the number of features used for estimation is 10, an MAE of 8 or less is obtained. Since MAE decreases as the number of features used for estimation increases, the number of features used in the exercise tolerance estimation system 10d is preferably 10 or more, more preferably 15 or more, and even more preferably 20 or more.

[0073] In the exercise tolerance estimation systems 10a, 10b, 10c, and 10d in the above-described first to fourth embodiments, the case of estimating the exercise tolerance of a subject has been described. This exercise tolerance includes various measurement values representing the current physical state of the subject. Note that in the exercise tolerance estimation systems 10a, 10b, 10c, and 10d, as the exercise tolerance of the subject, an index of an exercise intensity that is safe and effective in cardiac rehabilitation or cardiopulmonary function training may be used. The exercise intensity is a target value indicating what kind of exercise is good for the subject to do. As the exercise intensity, a target value indicating what kind of exercise is best for the subject to do or a target value indicating what kind of exercise is appropriate for the subject to do can be used. By using the exercise intensity as the exercise tolerance, even non-medical workers with little medical knowledge can specifically grasp what kind of exercise should be performed, and medical workers with medical knowledge can clearly grasp what kind of exercise should be made the subject perform.

[0074] The exercise tolerance estimation apparatuses 100a, 100b, 100c, 100d and the estimation model generation apparatuses 200a, 200b, 200c, 200d included in the exercise tolerance estimation systems 10a, 10b, 10c, 10d in the above-described first to fourth embodiments may be implemented by a computer. In that case, a program for realizing this function may be recorded on a computer-readable recording medium, and the program recorded on this recording medium may be read into a computer system and executed to realize it. Here, the "computer system" includes hardware such as an OS (Operating System) and peripheral devices. Further, the "computer-readable recording medium" refers to a portable medium such as a flexible disk, a magneto-optical disk, a ROM (Read Only Memory), a CD-ROM (Compact Disk Read Only Memory), or a storage device such as a hard disk built into a computer system. Furthermore, the "computer-readable recording medium" also includes something that dynamically holds a program for a short time, such as a communication line when transmitting a program via a network such as the Internet or a communication line such as a telephone line, and something that holds a program for a certain period of time, such as volatile memory inside a computer system serving as a server or a client in that case. Also, the above program may be for realizing a part of the aforementioned functions, and may further be something that can be realized in combination with a program already recorded in the computer system, or may be realized using a programmable logic device such as an FPGA (Field Programmable Gate Array).

[0075] As described above, the embodiments of this invention have been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and designs and the like within the scope not departing from the gist of this invention are also included.

Industrial Applicability

[0076] The present invention can be applied to an exercise tolerance estimation method, an exercise tolerance estimation device, a computer program, etc. that are required to estimate the exercise tolerance that reflects the physical condition of a subject in detail without imposing an exercise load on the subject.

Explanation of symbols

[0077] 10a ··· Exercise tolerance estimation system, 10b ··· Exercise tolerance estimation system, 10c ··· Exercise tolerance estimation system, 10d ··· Exercise tolerance estimation system, 100a ··· Exercise tolerance estimation device, 100b ··· Exercise tolerance estimation device, 100c ··· Exercise tolerance estimation device, 100d ··· Exercise tolerance estimation device, 101 ··· Blood test value acquisition unit, 102 ··· Exercise tolerance estimation unit, 103 ··· Estimation model storage unit, 104 ··· Output unit, 200a ··· Estimation model generation device, 200b ··· Estimation model generation device, 200c ··· Estimation model generation device, 200d ··· Estimation model generation device, 201 ··· Blood test value acquisition unit, 202 ··· Ultrasonic examination result acquisition unit, 203 ··· Exercise tolerance estimation unit, 204 ··· Estimation model storage unit, 205 ··· Output unit, 301 ··· Blood test value acquisition unit, 302 ··· Drug information acquisition unit, 303 ··· Exercise tolerance estimation unit, 304 ··· Estimation model storage unit, 305 ··· Output unit, 401 ··· Blood test value acquisition unit, 402 ··· Ultrasonic examination result acquisition unit, 403 ··· Drug information acquisition unit, 404 ··· Exercise tolerance estimation unit, 405 ··· Estimation model storage unit, 406 ··· Output unit

Claims

1. A subject information acquisition step of acquiring subject information regarding a subject; An estimated model accumulation step of accumulating an estimated model generated based on subject information other than the subject and exercise tolerance; An exercise tolerance estimation step of estimating the exercise tolerance of the subject based on the subject information acquired in the subject information acquisition step and the estimated model accumulated in the estimated model accumulation step; An output step of outputting the exercise tolerance estimated in the exercise tolerance estimation step; An exercise tolerance estimation method having the above.

2. In the subject information acquisition step, as the subject information, blood test values of the subject are acquired. In the estimated model accumulation step, blood test values other than the subject are used as subject information other than the subject. The exercise tolerance estimation method according to Claim 1.

3. Further having an ultrasonic examination result acquisition step of acquiring an ultrasonic examination result of the subject; In the exercise tolerance estimation step, based on the subject information acquired in the subject information acquisition step, the estimated model accumulated in the estimated model accumulation step, and the ultrasonic examination result acquisition step, the exercise tolerance of the subject is estimated. The exercise tolerance estimation method according to Claim 1 or 2.

4. In the ultrasonic examination result acquisition step, as the ultrasonic examination result, at least one of the subject's vena cava diameter, TAPSE, interventricular septum thickness, right ventricular end-diastolic area, ascending aortic diameter, right ventricular end-systolic area, maximum left atrial volume, maximum left atrial volume coefficient, left ventricular outflow tract, TRPG, and examination image is acquired. The exercise tolerance estimation method according to Claim 3.

5. Further having a drug information acquisition step of acquiring at least one of the presence or absence of prescription of a beta blocker for the subject and the potency of the beta blocker as drug information; In the exercise tolerance estimation step, based on the subject information acquired in the subject information acquisition step, the estimated model accumulated in the estimated model accumulation step, and the drug information acquired in the drug information acquisition step, the exercise tolerance of the subject is estimated. The exercise tolerance estimation method according to Claim 1 or 2.

6. The exercise tolerance of the subject estimated in the exercise tolerance estimation step includes at least one of heart rate, oxygen uptake, maximum oxygen uptake, VE-VCO ２ SLOPE, systolic blood pressure, diastolic blood pressure, end-tidal oxygen concentration, and respiratory exchange ratio The exercise tolerance estimation method according to Claim 1.

7. The subject information of the subject obtained in the subject information acquisition step includes numerical values related to at least one of CRP, NT-pro-BNP, BNP, CK, HDL cholesterol, triglyceride, creatinine, blood glucose, albumin, white blood cell count, Hb (hemoglobin), platelet count, and hemoglobin A1c of the subject. The exercise tolerance estimation method according to claim 1.

8. A subject information acquisition unit that acquires subject information regarding a subject, An estimation model storage unit that stores an estimation model generated based on subject information and exercise tolerance other than the subject, An exercise tolerance estimation unit that estimates the exercise tolerance of the subject based on the subject information acquired by the subject information acquisition unit and the estimation model stored in the estimation model storage unit, An output unit that outputs the exercise tolerance estimated by the exercise tolerance estimation unit, An exercise tolerance estimation device comprising:

9. A computer program for causing a computer to function as the exercise tolerance estimation device according to claim 8.

Images