Estimation device, index statistical value calculation method, program, and recording medium

Abstract

An estimation device provided with an indicator calculation unit for calculating a time series indicator relating to the condition and movements of a subject on the basis of vital data relating to the condition of the circulatory organs of the subject and movement data relating to physical oscillations and angles; a statistic calculation unit for calculating an indicator statistic, which is a statistic for the time series indicator, on the basis of the time series indicator; and an independence estimation unit for estimating the independence of the subject on the basis of the indicator statistic.

Description

[Technical Field]

[0001] The present invention provides an estimation device, Calculation method of indicator statistics , programs and Recording media Regarding. [Background technology]

[0002] In rehabilitation medicine, when a patient suffers from paralysis due to cerebrovascular disease or other causes, the degree of independence is used as a measure to evaluate the disability. The Functional Independence Measure (FIM) is frequently used in clinical settings not only in Japan but also overseas to measure independence. While the FIM is scored visually by medical professionals, a method has been proposed to automatically estimate the FIM using machine learning from measurements taken with a wearable device (Non-Patent Document 1).

[0003] Non-Patent Document 2 describes an example of applying measurement results from a wearable device to rehabilitation. Non-Patent Document 3 describes imputation methods in statistics. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Publication No. 2020-036781 [Non-patent literature]

[0005] [Non-Patent Document 1] Jason Conci, Utilizing Consumer-grade Wearable Sensors for Unobtrusive Rehabilitation Outcome Prediction, 2019. [Non-patent document 2] "Biometric Sensing and Its Application to Human State Estimation," Technical Information Association, 2020 [Non-patent document 3] https: / / ja.wikipedia.org / wiki / Substitution method_(Statistics), last updated April 16, 2021 13:13 (UTC) Summary of the Invention [Problem to be solved by the invention]

[0006] In Non-Patent Document 1, the wearable device measures physical activity, heart rate, and sleep quality. However, hospitalized patients often suffer from sleep disorders, and the correlation between sleep quality and independence is low, making it difficult to accurately estimate independence. This is because sleep, which is one state related to nervous system activity, was measured using a type of wearable device that can only measure information related to the circulatory system and physical movement. Therefore, this problem can be solved by appropriately basing the measurement on information related to the circulatory system and physical movement. An object of the present invention is to provide an estimation device, an estimation method, a program, and a storage medium that can estimate the degree of independence with high accuracy. [Means for solving the problem]

[0007] One aspect of the present invention is an estimation device that includes an index calculation unit that calculates a time series index regarding the condition and movement of a person being measured based on vital data regarding the state of the person's circulatory system and movement data regarding physical vibrations and angles, a statistical value calculation unit that calculates an index statistical value, which is a statistical value of the time series index, based on the time series index, and an independence estimation unit that estimates the degree of independence of the person being measured based on the index statistical value.

[0008] One aspect of the present invention is an estimation method having an index calculation step for calculating a time series index regarding the condition and movement of a subject based on vital data related to the state of the circulatory system of the subject and movement data related to physical vibrations and angles; a statistical value calculation step for calculating an index statistical value, which is a statistical value of the time series index, based on the time series index; and an independence estimation step for estimating the degree of independence of the subject based on the index statistical value.

[0009] One aspect of the present invention is a program for causing a computer to function as the above-described estimation device.

[0010] One aspect of the present invention is a computer-readable recording medium on which a program for causing a computer to function as the above-described estimation device is recorded. [Effects of the Invention]

[0011] According to the present invention, the degree of independence can be estimated with high accuracy. [Brief explanation of the drawings]

[0012] [Figure 1] FIG. 1 is a diagram illustrating an estimation system according to a first embodiment. [Figure 2] FIG. 1 is a diagram illustrating a configuration of an estimation device. [Figure 3] 10 is a flowchart showing the operation of the estimation system. [Figure 4] FIG. 2 illustrates an example of a hardware configuration of an estimation apparatus. [Figure 5] FIG. 2 is a diagram illustrating a configuration of an estimation device according to a modified example of the first embodiment. [Figure 6] FIG. 2 is a diagram illustrating a configuration of an estimation device according to a modified example of the first embodiment. [Figure 7] FIG. 1 is a diagram showing explanatory variables and correlation coefficients. DETAILED DESCRIPTION OF THE INVENTION

[0013] First Embodiment FIG. 1 is a diagram showing an estimation system 1 according to a first embodiment. The estimation system 1 includes an estimation device 2, a sensor terminal 202, and a relay terminal 203. In the estimation system 1, the sensor terminal 202 is attached to the trunk of a person being measured 201, for example. Measurement results by the sensor terminal 202 are relayed by the relay terminal 203 and transmitted to the estimation device 2. The sensor terminal 202 measures vital data and movement data of the person being measured 201. The vital data is a feature amount related to the state of the circulatory system of the person being measured 201. The movement data is a feature amount related to the physical vibration and angle of the person being measured 201. The vital data is, for example, a cardiac potential, heart rate, pulse rate, RRI, and body temperature. The movement data is, for example, acceleration and angular velocity. The sensor terminal 202 may be a computing device such as a smartphone or tablet.

[0014] The relay terminal 203 transmits the data received from the sensor terminal 202 to the estimation device 2. The relay terminal 203 is connected to the sensor terminal 202 via Bluetooth (registered trademark), for example, but is not limited to, and to the estimation device 2 via Wi-Fi. The relay terminal 203 may be a computer device such as a smartphone or tablet, and may process the data received from the sensor terminal 202 and transmit it to the estimation device 2.

[0015] 2 is a diagram showing the configuration of the estimation device 2. The estimation device 2 includes a receiving unit 10, a received data storage unit 11, an index calculation unit 12, a statistical value calculation unit 14, an independence estimation model storage unit 16, an independence estimation unit 18, and a presentation unit 20.

[0016] The receiving unit 10 receives vital data and motion data, which are measurement results by the sensor terminal 202, from the relay terminal 203.

[0017] The received data storage unit 11 stores data received by the receiving unit 10. The received data storage unit 11 stores data in association with the time at which the data was measured. The received data storage unit 11 stores, for example, time-series data of the cardiac potential and acceleration of the subject 201.

[0018] The index calculation unit 12 calculates time-series indexes indicating changes over time in the state and movement of the subject 201 based on the data stored in the received data storage unit 11. The time-series index is, for example, a change over time in %HRR (percent heart rate reserve), which indicates a value obtained by normalizing the heart rate calculated based on the cardiac potential by the maximum and minimum heart rates of the subject 201. The time-series index is, for example, the activity time for each posture (lying, sitting, standing, walking) calculated based on acceleration. The time-series index is, for example, a change over time in body movement, which is the standard deviation per predetermined time (for example, 1 second) of a value obtained by combining three-axis acceleration calculated based on acceleration. The time-series index is, for example, a change over time in the number of steps per predetermined time (for example, 1 minute) calculated based on acceleration. The time-series index is, for example, a change over time in an activity cost index, which is a value obtained by dividing %HRR calculated based on the cardiac potential and acceleration by body movement. The index calculation unit 12 may calculate at least one of the time change in %HRR, the activity time for each posture, the time change in body movement, the time change in the number of steps per hour, and the time change in the activity cost index.

[0019] The index calculation section 12 does not need to calculate a time series index indicating sleep as a time series index indicating a time change in the state or movement of the subject 201. The time series index indicating sleep is, for example, sleep time or sleep quality.

[0020] The index calculation unit 12 may normalize the time series data stored in the received data storage unit 11. For example, when the time series data stored in the received data storage unit 11 is time series data spanning 24 hours or more, the index calculation unit 12 may normalize the time series data to generate time series data spanning 24 hours. The index calculation unit 12 may generate time series data spanning 24 hours by taking an ensemble average of the time series data at the same time.

[0021] The statistical value calculation unit 14 calculates a statistical value of the index based on the time-series index calculated by the index calculation unit 12. The statistical value is, for example, an average value, a quantile, or a deviation. The statistical value of the index may be a statistical value of the value of the time-series index for a predetermined time period (for example, 30 minutes). When the time-series index is a time change in %HRR over 24 hours, for example, 48 average values ​​of %HRR are calculated for each 30 minutes, and the final average value of the 48 average values ​​is calculated as the statistical value of the index. When the time-series index is an activity time by posture, for example, the average value of the time for each posture for each 30 minutes is calculated as the statistical value of the index. When the time-series index is a time change in body movement over 24 hours, for example, 48 average values ​​of body movement for each 30 minutes are calculated, and the final average value of the 48 average values ​​is calculated as the statistical value of the index. When the time-series index is a time change in step count, for example, the average number of steps for each 30 minutes is calculated as the statistical value of the index. When the time series index is the time change of the activity cost index over 24 hours, for example, 48 average values ​​of the activity cost index every 30 minutes are calculated, and the final average value of the 48 average values ​​is calculated as the statistical value of the index.

[0022] The independence degree estimation model storage unit 16 stores an independence degree estimation model. The independence degree estimation model is a model that receives the statistical values ​​of the indices calculated by the statistical value calculation unit 14 as input and outputs the independence degree. The independence degree estimation model is created by machine learning using a dataset of the statistical values ​​of the indices and the independence degree. In machine learning, the explanatory variable is the statistical value of the indices, and the objective variable is the independence degree. The independence degree is a measure for evaluating the disability of a person with a disability, particularly an elderly person with a disability. The independence degree is an index indicating how independently a person with a disability can perform daily activities and is also an index indicating whether a person with a disability is bedridden. The independence degree is, for example, the FIM. The independence degree is, for example, the sum of scores for 13 items related to motor function included in the FIM. The machine learning method is not limited, and examples include neural networks, random forests, support vector machines, logistic regression, and ensemble learning.

[0023] The statistical value of the index may be a value of the time-series index for each predetermined time period. When the time-series index is the change in %HRR over a 24-hour period, the statistical value of the index may be 48 average values ​​of %HRR for each 30-minute period. When the time-series index is the activity time for each posture, the statistical value of the index may be the time for each posture for each 30-minute period. When the time-series index is the change in body movement over a 24-hour period, the statistical value of the index may be 48 average values ​​of body movement for each 30-minute period. When the time-series index is the change in number of steps over a 24-hour period, the statistical value of the index may be the number of steps for each 30-minute period. When the time-series index is the change in activity cost index over a 24-hour period, the statistical value of the index may be 48 average values ​​of activity cost index for each 30-minute period. In this case, the number of inputs to the independence estimation model increases, which increases the calculation load during estimation, but improves the estimation accuracy.

[0024] The statistical values ​​of the indices may be time-series indices. In this case, the number of inputs to the independence estimation model increases, which increases the calculation load during estimation, but improves the estimation accuracy.

[0025] The independence degree estimation unit 18 estimates the degree of independence from the statistical values ​​of the indexes using an independence degree estimation model. The independence degree estimation unit 18 estimates the degree of independence by inputting the statistical values ​​of the indexes into the independence degree estimation model and outputting the independence degree.

[0026] The presentation unit 20 presents the degree of independence estimated by the independence degree estimation unit 18. The presentation unit 20 presents the degree of independence by outputting data to a display device such as a display.

[0027] 3 is a flowchart showing the operation of the estimation system 1. The sensor terminal 202 measures vital data and movement data (step S101). The relay terminal 203 relays the vital data and movement data measured by the sensor terminal 202 and transmits them to the estimation device 2 (step S102). The receiver 10 of the estimation device 2 receives the vital data and movement data from the relay terminal 203 (step S201). The index calculation unit 12 calculates a time-series index based on the vital data and movement data (step S202). The statistical value calculation unit 14 calculates a statistical value of the index based on the time-series index (step S203). The independence degree estimation unit 18 estimates the independence degree by inputting the statistical value of the index into an independence degree estimation model (step S204). The presentation unit 20 presents the estimated independence degree (step S205).

[0028] 4 is a diagram showing an example of the hardware configuration of the estimation device 2. The estimation device 2 can be realized by, for example, a computer including an arithmetic unit 102 having a CPU 103 and a main memory device 104 connected via a bus 101, a communication interface 105, an external memory device 107, a clock 108, and a display device 109, and a program that controls these hardware resources.

[0029] The CPU 103 and the main storage device 104 constitute the arithmetic device 102. The main storage device 104 stores in advance programs for the CPU 103 to perform various controls and calculations. The arithmetic device 102 realizes each function of the estimation device 2 shown in FIG. 2.

[0030] The communication interface 105 is an interface and control device for connecting the estimation device 2 to various external electronic devices such as the relay terminal 203 via a communication network. The estimation device 2 may receive data on heart rate, electrocardiogram waveform, and acceleration from the relay terminal 203 via the communication interface 105 and the communication network.

[0031] The communication interface 105 may be, for example, an arithmetic interface and an antenna compatible with wireless data communication standards such as LTE, 3G, wireless LAN, Bluetooth, etc. The communication interface 105 implements the receiving unit 10 in FIG.

[0032] The external storage device 107 is composed of a readable / writable storage medium and a drive for reading and writing various information such as programs and data from and to the storage medium. The storage medium for the external storage device 107 can be a hard disk or a semiconductor memory such as a flash memory.

[0033] The external storage device 107 may have a storage area for storing vital data and movement data measured by the sensor terminal 202, a program storage unit for storing programs for the estimation device 2 to perform analysis processing of vital data and movement data, and other storage devices (not shown), such as a storage device for backing up programs and data stored in the external storage device 107. The external storage device 107 realizes the received data storage unit 11 and the independence estimation model storage unit 16 in FIG. 2 .

[0034] The clock 108 measures time and is configured by an internal clock or the like provided in the estimation device 2. The time information obtained by the clock 108 is used for sampling vital data and movement data and for data analysis processing.

[0035] The display device 109 functions as the presentation unit 20 of the estimation device 2. The display device 109 is realized by a liquid crystal display or the like.

[0036] <Variations> The independence estimation model may be a model that receives as input the statistical values ​​of the indices calculated by the statistical value calculation unit 14 and the profile of the subject 201, and outputs the independence. In other words, the independence estimation model is generated using the statistical values ​​of the indices and the profile of the subject 201 as explanatory variables, and the independence as a response variable. The profile of the subject 201 is, for example, the age, height, weight, or sex of the subject 201. The profile of the subject 201 may be data that combines some of the age, height, weight, and sex of the subject 201.

[0037] 5 is a diagram showing the configuration of an estimation device 2 according to a modification of the first embodiment. The estimation device 2 includes a subject profile acquisition unit 22. The subject profile acquisition unit 22 acquires a profile of the subject 201. The independence degree estimation unit 18 inputs the statistical values ​​of the indices and the profile of the subject 201 acquired by the subject profile acquisition unit 22 into an independence degree estimation model and outputs the independence degree, thereby estimating the independence degree.

[0038] The independence degree estimation model may be a model that receives as input the statistical values ​​of the indices calculated by the statistical value calculation unit 14 and the past independence degree of the subject 201, and outputs the independence degree. In other words, the independence degree estimation model is generated using the statistical values ​​of the indices and the past independence degree of the subject 201 as explanatory variables and the independence degree as a response variable.

[0039] 6 is a diagram showing the configuration of an estimation device 2 according to a modification of the first embodiment. The estimation device 2 includes a past independence degree acquisition unit 24. The past independence degree acquisition unit 24 acquires the past independence degree of the subject 201. The independence degree estimation unit 18 inputs the statistical values ​​of the indices and the past independence degree of the subject 201 acquired by the past independence degree acquisition unit 24 into an independence degree estimation model and outputs the independence degree, thereby estimating the independence degree.

[0040] The independence degree estimation model may be generated using the statistical values ​​of the indices, the profile of the subject 201, and the past independence degree of the subject 201 as explanatory variables, and the independence degree as a response variable. In this case, the independence degree estimation unit 18 inputs the statistical values ​​of the indices, the profile of the subject 201, and the past independence degree of the subject 201 into the independence degree estimation model and outputs the independence degree, thereby estimating the independence degree.

[0041] <Experimental Example> The explanatory variables used to generate the independence estimation model were varied, and the correlation coefficient between the estimated independence level and the actual independence level was calculated. Figure 7 shows the explanatory variables and correlation coefficients. Under the first condition, the correlation coefficient was calculated by performing a five-fold cross-validation on the statistical values ​​of each of the time-series indicators of the subject 201 (%HRR, time in a lying position, time between standing and sitting, walking time, number of steps, and activity cost index) in 30-minute increments. The explanatory variables under the second condition included the explanatory variables under the first condition as well as the profile of the subject 201 (age, height, weight, and gender). Under the third condition, the explanatory variables were the profile of the subject 201 (age, height, weight, and gender) and the subject 201's past independence level. The subject 201's past independence level was the independence level during the first week of hospitalization, and the independence level of the objective variable was the independence level during the fifth week of hospitalization. In other words, the past independence level was the independence level four weeks prior to the estimated independence level. The explanatory variables for the fourth condition include the explanatory variables for the first condition as well as the past independence of the subject 201. The explanatory variables for the fifth condition include the explanatory variables for the fourth condition as well as the profile of the subject 201 (age, height, weight, and gender). For the second to fifth conditions as well, a 5-fold cross-validation was performed on the statistical values ​​to calculate the correlation coefficient. In k-fold cross-validation, k is the number of datasets used for evaluation, and the value of k can be 2, 5, or 10.

[0042] The correlation coefficient exceeded 0.6 under all conditions. As a result, the estimation device 2 can estimate the degree of independence with high accuracy.

[0043] Furthermore, if the subject 201 is an inpatient, the subject 201 often suffers from sleep disorders and the quality of their sleep is often poor regardless of their degree of independence. Therefore, if the correlation between sleep data and the degree of independence is low and the time-series indices showing the time-series changes in the state and behavior of the subject 201 include time-series indices showing sleep, the accuracy of the estimation device 2's estimation of the degree of independence will deteriorate. Therefore, if the time-series indices showing the time-series changes in the state and behavior of the subject 201 include time-series indices showing sleep, the estimation device 2 can estimate the degree of independence without being affected by the degree of independence, leading to improved estimation accuracy.

[0044] Second Embodiment The estimation device 2 according to the second embodiment includes a defect complementing unit 30 in addition to the components of the estimation device 2 according to the first embodiment.

[0045] The missing data complementing unit 30 complements missing data. Missing data is data that the estimation device 2 was unable to receive due to a malfunction of the sensor terminal 202 or the communication conditions between the relay terminal 203 and the estimation device 2. The missing data may include data whose value exceeds an upper threshold or falls below a lower threshold based on a predetermined upper threshold and a predetermined lower threshold. The defect complementing unit 30 may invalidate the missing data. When the index calculating unit 12 generates time series data spanning 24 hours by taking the ensemble average of the time series data at the same time, the defect complementing unit 30 may calculate the average value of the time series excluding the missing data as the ensemble average.

[0046] The missing data complementing section 30 may complement missing data by applying a multiple imputation method to the profile of the subject 201 and the subject's past independence level.

[0047] The estimation device 2 according to the second embodiment can complement missing data, handle a larger number of data sets with explanatory variables that are free of missing data, and improve the reliability of estimation of the degree of independence.

[0048] Other Embodiments The explanatory variables of the independence estimation model are not limited to the statistical values ​​of the indicators, the profile of the subject 201, or the past independence of the subject 201. For example, the explanatory variables of the independence estimation model may include disease information of the subject 201. The disease information of the subject 201 is, for example, information on whether or not the subject 201 has a disease such as cerebrovascular disease, spinal injury, or femoral fracture. The independence estimation unit 18 may include the disease information of the subject 201 in the input to the independence estimation model in accordance with the explanatory variables.

[0049] The estimation device 2 may include an estimated independence degree storage unit 40, which stores the independence degree estimated by the independence degree estimation unit 18. The estimated independence degree storage unit 40 may also store the statistical values ​​of the indices and the independence degree in association with the profile of the subject 201, the subject 201's past independence degree, or disease information of the subject 201.

[0050] The estimation device 2 may include an estimated statistical value calculation unit 42 that calculates statistical values ​​of the estimated independence degrees based on the estimated independence degrees stored in the estimated independence degree storage unit 40. The estimated statistical value calculation unit 42 may calculate statistical values ​​of the estimated independence degrees based on a profile of the subject 201. For example, the estimated statistical value calculation unit 42 calculates average values ​​and standard deviations, which are statistical values ​​of the estimated independence degrees for each age group, such as those in their 50s and 60s. The presentation unit 20 may present the results calculated by the estimated statistical value calculation unit 42.

[0051] By using different explanatory variables and objective variables in the independence estimation model, it is possible to create a model that estimates a characteristic other than the independence level. For example, an estimation model that estimates one item in the profile of the subject 201 may be created by using one item in the profile of the subject 201 as the objective variable and at least one of the statistical value of the index and the item in the profile of the subject 201 that is not the objective variable, the independence level of the subject 201, and disease information of the subject 201 as the explanatory variables. By inputting explanatory variables into the estimation model, an estimated value of one item in the profile of the subject 201 is output. For example, when an explanatory variable is input into the estimation model, an estimated value of the age of the subject 201 is output. If the estimated age is lower than the subject's actual age, the subject 201 can understand that the statistical values ​​of the indicators included in the explanatory variables and the degree of independence are good.

[0052] For example, an estimation model for estimating a disease may be created by using the disease of the subject 201 as the objective variable and at least one of the statistical value of an index, an item in the profile of the subject 201, or the degree of independence of the subject 201 as the explanatory variable.

[0053] When the independence estimation model is an autoencoder, the presentation unit 20 may present data compressed by the independence estimation model. The compression by the autoencoder reduces the dimensions of the explanatory variables, leaving only essential information necessary for learning, allowing the essential information to be grasped.

[0054] The estimation device 2 and the sensor terminal 202 may be implemented by the same device. In this case, the estimation device 2 and the sensor terminal 202 do not need to communicate via the relay terminal 203. The relay terminal 203 may also perform some of the functions of the estimation device 2. For example, the relay terminal 203 may perform some of the functions of the index calculation unit 12, the statistical value calculation unit 14, the independence estimation unit 18, and the presentation unit 20, and transmit the processed data to the estimation device 2. [Explanation of symbols]

[0055] 1...estimation system, 2...estimation device, 10...receiving unit, 11...received data storage unit, 12...index calculation unit, 14...statistical value calculation unit, 16...independence degree estimation model storage unit, 18...independence degree estimation unit, 20...presentation unit, 22...subject profile acquisition unit, 24...past independence degree acquisition unit, 101...bus, 102...arithmetic unit, 103...CPU, 104...main storage unit, 105...communication interface, 107...external storage unit, 108...clock, 109...display unit, 201...subject, 202...sensor terminal, 203...relay terminal

Claims

1. an index calculation unit that calculates a time series index regarding the state and movement of the subject based on vital data regarding the state of the circulatory system of the subject and movement data regarding physical vibrations and angles; a statistical value calculation unit that calculates an index statistical value, which is a statistical value of the time-series index, based on the time-series index; an independence level estimation unit that estimates the independence level of the subject based on the index statistical value; Equipped with The time series index includes a time change in the activity time for each posture of the subject and / or a time change in the number of steps of the subject per predetermined time. Estimation device.

2. the independence degree estimation unit estimates the independence degree based on a profile of the subject. The estimation device according to claim 1 .

3. the independence degree estimation unit estimates the independence degree based on the subject's past independence degree; The estimation device according to claim 1 or 2.

4. a missing data complementing unit that complements missing data in the vital data and the motion data; The estimation device according to claim 1 , further comprising:

5. the independence degree estimating unit estimates the independence degree of the subject based on disease information of the subject. The estimation device according to any one of claims 1 to 4.

6. an estimated statistical value calculation unit that calculates a statistical value of the degree of independence based on the degree of independence estimated by the independence degree estimation unit; a presentation unit that presents the statistical value calculated by the estimated statistical value calculation unit; The estimation device according to claim 1 , further comprising:

7. A method for calculating an index statistic for estimating a subject's degree of independence, comprising: an index calculation step of calculating a time series index regarding the state and movement of the subject based on vital data regarding the state of the circulatory system of the subject and movement data regarding physical vibrations and angles; a statistical value calculation step of calculating an index statistic, which is a statistical value of the time-series index, based on the time-series index; and The time series index includes a time change in the activity time for each posture of the subject and / or a time change in the number of steps of the subject per predetermined time. Method for calculating indicator statistics.

8. A program for causing a computer to function as the estimation device according to any one of claims 1 to 6.

9. A computer-readable recording medium storing a program for causing a computer to function as the estimation device according to any one of claims 1 to 6.

Images