Information processing apparatus, method, and program

Abstract

To provide an information processing apparatus, a method, and a program capable of easily and accurately determining a disease of at least one of cognition and movement.SOLUTION: A processor: identifies a first section and a second section, each having a different detection target for a subject's movement, based on the measurement values of the subject's movement measured by a device wearable by the subject; acquires at least one feature value representing a characteristic of a disease related to at least one of cognition and movement, based on each of the measurement values of the first section and the second section; and acquires a disease assessment result based on the feature value and a predetermined assessment criterion.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present disclosure relates to an information processing device, a method, and a program. [Background technology]

[0002] With the advent of a fully aging society, it is becoming increasingly important to accurately predict the progression of cognitive or movement-related disorders such as dementia or Parkinson's disease and to develop optimal treatment strategies based on the predictions. For example, Patent Document 1 proposes a method for assessing cognitive and movement disorders by having a subject use a mobile device to perform tasks such as manually drawing or squeezing shapes, and comparing the measurement data obtained from the tasks with reference values. Patent Document 2 also proposes a method for determining whether a subject has mild dementia by determining factor information, such as an estimated Mini Mental State Examination (MMSE) score and an estimated cerebral glucose metabolic rate, which are factors that contribute to dementia, and using data acquired by a sensor worn by the subject and the factor information. In Patent Document 2, the factor information is acquired using a regression model generated by machine learning, with sensor data and attribute data (healthy or mildly dementia) as explanatory variables and the factor information as a target variable. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Special Publication No. 2019-531569 [Patent Document 2] Patent Publication No. 2021-029692 Summary of the Invention [Problem to be solved by the invention]

[0004] The method described in Patent Document 1 requires the subject to perform actions such as drawing, which places a heavy burden on the subject. The method described in Patent Document 2 requires determining factor information to obtain a diagnosis result, which increases processing costs. Furthermore, the method described in Patent Document 2 requires a two-stage determination process of determining factor information to determine mild cognitive impairment. As a result, the final diagnosis result of mild cognitive impairment is likely to contain errors, which may result in a decrease in diagnosis accuracy.

[0005] The present disclosure has been made in consideration of the above circumstances, and aims to enable easy and accurate diagnosis of at least one of cognitive and behavioral disorders. [Means for solving the problem]

[0006] An information processing device according to the present disclosure includes at least one processor, The processor Identifying a first section and a second section in which detection targets of the subject's movement are different based on measurements of the subject's movement measured by a device wearable on the subject; acquiring at least one feature value representing a disease characteristic relating to at least one of cognition and movement based on each of the measurement values of the first section and the measurement values of the second section; A disease determination result is obtained based on the feature amount and predetermined determination criteria.

[0007] In the information processing device according to the present disclosure, the first section may be a section for detecting the movement of the subject's entire body, and the second section may be a section for detecting the movement of the subject's head.

[0008] In the information processing device according to the present disclosure, the measurement value includes a change in acceleration of the subject in a vertical downward direction, The processor may derive landing information indicating that the subject has landed while moving based on changes in acceleration, and may identify a walking section of the subject as a first section and a non-walking section of the subject as a second section based on the continuity of the landing information.

[0009] In the information processing device according to the present disclosure, the measurement value includes the acceleration of the head in the front-back direction in the second section, The processor may acquire a feature quantity representing a characteristic of a cognitive disorder based on the forward / backward acceleration of the head in the second interval.

[0010] In the information processing device according to the present disclosure, the measurement value includes a head rotation speed in the second section, The processor may acquire a feature quantity representing a characteristic of a movement-related disease based on the rotation speed of the head in the second section.

[0011] In the information processing device according to the present disclosure, the measurement values include a left-right movement of the subject in the first section, The processor may acquire a feature quantity representing a characteristic of a cognitive disorder based on the left-right movement of the subject in the first section.

[0012] In the information processing device according to the present disclosure, the measurement values include a movement of the subject in the front-back direction in the first section, The processor may acquire a feature quantity representing a characteristic of a disease related to movement based on the forward and backward movement of the subject in the first section.

[0013] In addition, in the information processing device according to the present disclosure, the processor may derive the determination criterion.

[0014] In the information processing device according to the present disclosure, the determination criterion may be a reference value for distinguishing between disease and non-disease.

[0015] In the information processing device according to the present disclosure, the determination criterion may be a feature distribution estimated based on a plurality of feature amounts representing characteristics of a disease.

[0016] In addition, in the information processing device according to the present disclosure, the determination criterion may be a discrimination model that has been trained to output a score representing the possibility of disease when a feature amount is input.

[0017] In addition, in the information processing device according to the present disclosure, the disease may be at least one of mild cognitive impairment, dementia, and Parkinson's disease.

[0018] In the information processing device according to the present disclosure, the device may have an acceleration sensor and an angular velocity sensor.

[0019] In the information processing device according to the present disclosure, the device may further include an electro-oculography sensor.

[0020] The information processing method according to the present disclosure includes a step of: a computer identifying a first section and a second section in which detection targets of the subject's movement are different, based on measurement values of the subject's movement measured by a device wearable on the subject; acquiring at least one feature value representing a disease characteristic relating to at least one of cognition and movement based on each of the measurement values of the first section and the measurement values of the second section; A disease determination result is obtained based on the feature amount and predetermined determination criteria.

[0021] The information processing program according to the present disclosure includes steps of identifying a first section and a second section in which detection targets of the subject's movement are different, based on measurement values of the subject's movement measured by a device wearable on the subject; acquiring at least one feature value representing a disease characteristic related to at least one of cognition and movement based on each of the measurement values in the first section and the measurement values in the second section; a procedure for obtaining a disease determination result based on the feature amount and a predetermined determination criterion; to be executed by the computer. [Effects of the Invention]

[0022] According to the present disclosure, brain diseases can be diagnosed easily and accurately. [Brief explanation of the drawings]

[0023] [Figure 1] FIG. 1 is a diagram showing a schematic configuration of a disease detection system to which an information processing device according to an embodiment of the present invention is applied. [Figure 2] FIG. 1 is a diagram showing the hardware configuration of a glasses-type device according to an embodiment of the present invention. [Figure 3] FIG. 1 is a diagram showing the hardware configuration of an analytical server, which is an information processing device according to an embodiment of the present invention. [Figure 4] A functional configuration diagram of an analytical server, which is an information processing device according to this embodiment. [Figure 5] Figure showing the detection results of landing candidates [Figure 6] A diagram for explaining the derivation of the distribution of feature quantities using the kernel density estimation method. [Figure 7] Diagram showing the discrimination model [Figure 8] FIG. 1 is a diagram for explaining disease determination when the determination criterion is a representative value of a feature amount. [Figure 9] FIG. 1 is a diagram for explaining disease determination when the determination criterion is the distribution of feature amounts. [Figure 10] A diagram showing a display screen of the judgment result [Figure 11] A flowchart showing the processing performed in this embodiment [Figure 12] FIG. 10 is a diagram for explaining correction of detection results of landing candidates; DETAILED DESCRIPTION OF THE INVENTION

[0024] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. First, the configuration of a disease detection system to which an information processing device according to this embodiment is applied will be described. FIG. 1 is a diagram showing a schematic configuration of the disease detection system. In the disease detection system 100 shown in FIG. 1, an eyeglass-type device 1 and an analytical server 2, which is an information processing device according to this embodiment, are connected in a communicable state via a network 3. Note that, in addition to the analytical server 2 and eyeglass-type device 1, the disease detection system 100 may further include a mobile terminal 4, such as a smartphone or tablet terminal, and a general-purpose computer 5.

[0025] In this embodiment, the analytical server 2 derives a disease assessment result relating to at least one of the cognition and movement of the subject H. In this embodiment, the analytical server 2 acquires disease assessment results relating to both the cognition and movement of the subject H. Specifically, the analytical server 2 acquires an assessment result for Mild Cognitive Impairment (MCI, hereinafter referred to as MCI), which is a cognitive disease, and an assessment result for Parkinson's disease, which is a movement disease.

[0026] First, the eyeglass-type device 1 will be described. The eyeglass-type device 1 is a wearable device that is worn by a subject H who is a target for disease detection and is capable of measuring the movements of the subject H. FIG. 2 is a diagram showing the hardware configuration of the eyeglass-type device 1. As shown in FIG. 2, the eyeglass-type device 1 includes a memory 16 as a temporary storage area, a communication I / F (Interface) 17, and a sensor 18. The memory 16, the communication I / F 17, and the sensor 18 are connected to a bus 19. An example of such an eyeglass-type device is "JINS MEME" (registered trademark) manufactured by JINS Co., Ltd. Note that the eyeglass-type device 1 may also include a CPU, a screen for displaying various information, and the like, like smart glasses.

[0027] The sensor 18 is provided on the eyebrows, nose pads, etc. of the eyeglass device 1 and detects the movement of the subject H. Specifically, the sensor 18 is a six-axis sensor having a three-axis acceleration sensor and a three-axis angular velocity sensor so as to be able to detect three-axis acceleration and angular velocity around three axes. The sensor 18 may also be a nine-axis sensor having a three-axis axis sensor in addition to the three-axis acceleration sensor and the three-axis angular velocity sensor. The sensor 18 may also include an electro-oculography sensor that detects the electro-oculography.

[0028] In this embodiment, the subject H wears the eyeglass-type device 1 and moves freely. Then, the sensor 18 measures the movements of the subject H while the subject H is moving freely and acquires measurement values. In this embodiment, the sensor 18 is a six-axis sensor, and thus six measurement values are acquired while the subject H is moving freely. Therefore, in this embodiment, the sensor 18 acquires time-series measurement values over a period during which the subject H is moving freely.

[0029] The three axes measured by the sensor 18 are the front-to-back, left-to-right, and up-down directions of the subject H, and the acceleration sensor acquires measured values of acceleration of the subject H in the front-to-back, left-to-right, and up-down directions. The measured values of acceleration of the subject H in the front-to-back, left-to-right, and up-down directions represent the front-to-back, left-to-right, and up-down directions of the subject H. In addition, the angular velocity sensor acquires measured values of angular velocity by rotation around an axis extending from front to back, an axis extending left to right, and an axis extending up and down of the subject H. The measured values of angular velocity represent the rotational speed of the subject H around an axis extending from front to back, an axis extending left to right, and an axis extending up and down.

[0030] In this embodiment, the measurement value data acquired by the sensor 18 is temporarily stored in the memory 16, and is transmitted as a measurement value from the communication I / F 17 via the network 3 to the analytical server 2 after the measurement period has ended.

[0031] The communication I / F 17 may be wirelessly connected to the network 3 to perform wireless communication, or may be short-range wireless communication such as Bluetooth (registered trademark). A plurality of communication I / Fs 17 may be provided to enable both wireless communication and short-range wireless communication. When the communication I / F 17 performs short-range wireless communication, the measurement values acquired by the eyeglass device 1 are transmitted to a portable terminal 4 of, for example, the subject H, and then transmitted from the portable terminal 4 to the analysis server 2 via the network 3. Alternatively, the measurement values are transmitted from the portable terminal 4 to the computer 5 via short-range wireless communication or a wired connection, and then transmitted from the computer 5 to the analysis server 2 via the network 3. When the measurement values are transmitted to the portable terminal 4 or the computer 5, the measurement values may be confirmed in the portable terminal 4 or the computer 5.

[0032] Note that the measurement values acquired by the sensor 18 contain fine vibration components that can be considered as noise in addition to the values representing the movements to be acquired. For this reason, the eyeglass device 1 may perform a noise removal process on the data acquired by the sensor 18, and transmit the noise-removed data as measurement values to the analysis server 2. Examples of noise removal processes include, but are not limited to, filtering using a low-pass filter.

[0033] The network 3 may be a local area network (LAN) that locally connects the eyeglass-type device 1 and the analysis server 2, or a wide area network (WAN: Wide Area Network) that connects the analysis server 2 and the eyeglass-type device 1 over a wide area via a public line network or a dedicated line network.

[0034] Next, the analytical server 2 will be described. Fig. 3 is a diagram showing the hardware configuration of the analytical server 2, which is an information processing device. As shown in Fig. 3, the analytical server 2 includes a CPU 21, non-volatile storage 23, and memory 26 as a temporary storage area. The analytical server 2 also includes a display 24 such as an LCD display, an input device 25 consisting of a keyboard, a pointing device such as a mouse, and the like, and a communication I / F 27 connected to the network 3. The CPU 21, storage 23, display 24, input device 25, memory 26, and communication I / F 27 are connected to a bus 29. The CPU 21 is an example of a processor in the present disclosure.

[0035] The storage 23 is realized by a hard disk drive (HDD), an SSD, a flash memory, etc. The storage 23 as a storage medium stores an information processing program 22. The CPU 21 reads the information processing program 22 from the storage 23, loads it into the memory 26, and executes the loaded information processing program 22.

[0036] Next, the functional configuration of the analytical server 2, which is an information processing device according to this embodiment, will be described. Fig. 4 is a diagram showing the functional configuration of the analytical server 2, which is an information processing device according to this embodiment. As shown in Fig. 4, the analytical server 2 includes a measurement value acquisition unit 31, a section identification unit 32, a feature amount acquisition unit 33, a criterion derivation unit 34, a disease determination unit 35, and an output control unit 36. When the CPU 21 executes the information processing program 22, the CPU 21 functions as the measurement value acquisition unit 31, the section identification unit 32, the feature amount acquisition unit 33, the criterion derivation unit 34, the disease determination unit 35, and the output control unit 36.

[0037] The measurement value acquisition unit 31 acquires, via the communication I / F 27, the measurement values transmitted from the glasses-type device 1 via the network 3.

[0038] The section identification unit 32 identifies a first section and a second section, which have different detection targets for the movement of the subject H, based on the measurement values acquired by the measurement value acquisition unit 31. In this embodiment, the first section is a section in which the detection target is the movement of the entire body of the subject H, and the second section is a section in which the detection target is the movement of the head of the subject H. The movement of the entire body of the subject H occurs when the subject H is moving, particularly while walking. The movement of the head of the subject H occurs when the subject H is not moving, particularly when not walking. For this reason, in this embodiment, the first section is a walking section, and the second section is a non-walking section.

[0039] In this embodiment, the measurement values measured by the eyeglass device 1 include the vertical acceleration of the subject H, as described above. When a person is walking, a large change in acceleration occurs in the vertical downward direction when one foot lands on the ground. Therefore, in this embodiment, the section identification unit 32 detects, based on the vertical acceleration included in the measurement values, the time when the change in acceleration in the vertical downward direction becomes equal to or greater than a predetermined threshold, as a landing candidate.

[0040] Fig. 5 is a diagram showing the detection results of landing candidates. In Fig. 5, the horizontal axis indicates the time when the measurement value was acquired, and the vertical axis indicates an index indicating whether a landing candidate was detected. As shown in Fig. 5, the detection results of landing candidates include a section T1 where landing candidates are continuously detected over a certain period of time, and a section T2 where no landing candidates are detected at all over a certain period of time.

[0041] Here, when a person is walking, landings are made continuously, and therefore landing candidates are continuously detected. As shown in FIG. 5 , the section identification unit 32 identifies a section T1 in which landing candidates are continuously detected for a predetermined period throughout the entire period in which measurement values are acquired as a walking section. Also, as shown in FIG. 5 , the section identification unit 32 identifies a section T2 in which no landing candidates are detected for a predetermined period as a non-walking section. In this embodiment, the measurement value acquisition unit 31 acquires time-series measurement values. Therefore, the section identification unit 32 identifies the start time and end time of section T1 as a walking section, and the start time and end time of section T2 as a non-walking section. Instead of identifying the start time and end time, the walking section and non-walking section may be identified using the start time and the length of time from when the section starts to when it ends.

[0042] The feature acquisition unit 33 acquires at least one feature representing a characteristic of a disease related to at least one of cognition and movement based on each of the measurement values of the first section and the measurement values of the second section. Specifically, the feature acquisition unit 33 acquires multiple feature representing the characteristics of MCI and Parkinson's disease based on each of the walking section and non-walking section identified by the section identification unit 32 as described above.

[0043] Regarding head movement, MCI patients tend to have greater variability in forward and backward sway compared to healthy individuals, and Parkinson's disease patients tend to have slower head rotation, such as shaking their head from side to side, compared to healthy individuals.

[0044] In terms of overall body movement, MCI patients tend to have worse left-right balance than healthy people. Because overall body movement in Parkinson's disease patients is slower, their forward-backward movement tends to be smaller than that of healthy people.

[0045] Therefore, in the non-walking section where the movement of the subject H's head is to be detected, the feature acquisition unit 33 acquires information about the acceleration in the front-to-back direction of the subject H and information about the rotational speed as feature amounts. For example, statistical information about the acceleration component in the front-to-back direction in the non-walking section included in the measurement values can be used as the information about the acceleration in the front-to-back direction. For example, statistical values related to the dispersion of values of the acceleration component in the front-to-back direction, such as the standard deviation and quartile deviation, can be used as the statistical information about the acceleration component in the front-to-back direction. Furthermore, statistical values representative of the acceleration component in the front-to-back direction, such as the maximum value, minimum value, mean value, median, mode, and quartiles, can be used as the statistical information about the acceleration component in the front-to-back direction. Furthermore, information about the shape of the distribution of the acceleration component, such as the kurtosis and skewness of the acceleration component in the front-to-back direction, can be used as the statistical information about the acceleration component in the front-to-back direction. However, the statistical information about the acceleration component in the front-to-back direction is not limited to these.

[0046] Information relating to rotational speed can be, for example, statistical information on angular velocity components in the front-to-back and left-to-right directions of subject H in the non-walking section. Statistical information on angular velocity components in the front-to-back and left-to-right directions of subject H can include, but is not limited to, statistical values such as the average, median, and quartiles of angular velocity components in rotations around an axis extending from front to back, left to right, and up to down of subject H, as well as information such as the 90th, 95th, and 97th percentiles. A percentile is a percentage that indicates where a certain measurement value lies within the whole.

[0047] On the other hand, in the walking section where the movement of the entire body of the subject H is to be detected, the feature acquisition unit 33 acquires information about the left-right movement of the subject H and information about the forward-backward movement as feature amounts. The information about the left-right movement can be, for example, statistical information about the left-right acceleration components in the walking section. The statistical information about the left-right acceleration components can be, for example, the difference between the average values of the left-right acceleration components, the difference between the cumulative values of the left-right acceleration components, or the difference between the cumulative values of the left-right acceleration components at the timing when a large change in the vertical acceleration component occurs (i.e., the timing when a landing candidate appears). Furthermore, the statistical information about the left-right acceleration components can be, for example, statistical values related to the variation of values, such as the standard deviation and quartile deviation of the left-right acceleration components in the walking section. Furthermore, the statistical information about the left-right acceleration components can be, for example, statistical values representative of the acceleration components, such as the maximum value, minimum value, mean value, median, mode, and quartile of the left-right acceleration components. Furthermore, as statistical information on the acceleration component in the left-right direction, information on the shape of the distribution of acceleration component values, such as kurtosis and skewness of the acceleration component in the left-right direction, can be used. As information on the movement in the front-back direction, the same statistical information as that described above for the acceleration component in the front-back direction can also be used. Note that the statistical information is not limited to the above.

[0048] If sensor 18 includes an electro-oculography sensor, the measurement values acquired by measurement value acquisition unit 31 include the measurement values of the electro-oculography sensor. Parkinson's disease patients tend to blink less frequently than healthy individuals. Therefore, if sensor 18 includes an electro-oculography sensor, feature amount acquisition unit 33 acquires the time when a blink occurred from the change pattern of the electro-oculography, and acquires the number of blinks per unit time as a feature amount.

[0049] The determination criterion derivation unit 34 derives in advance determination criteria that the disease determination unit 35 (described later) uses to obtain disease determination results, and stores the derived criteria in the storage 23. When deriving the determination criterion, the determination criterion derivation unit 34 acquires feature quantities similar to those of the feature acquisition unit 33 described above from measurement values of a large number of subjects H, including healthy individuals, MCI patients, and Parkinson's disease patients. That is, in a non-walking section in which head movements of the subjects H are to be detected, the determination criterion derivation unit 34 acquires, as feature quantities, information on the forward / backward acceleration and rotational speed of the large number of subjects H. Furthermore, in a walking section in which movement of the entire body of the subjects H is to be detected, the determination criterion derivation unit 34 acquires, as feature quantities, information on the left-right movement and the forward / backward movement of the large number of subjects H. The determination criterion derivation unit 34 acquires feature quantities in association with the attributes of the patients, i.e., the attributes of healthy individuals, MCI, and Parkinson's disease.

[0050] The determination criterion derivation unit 34 derives a determination criterion by associating each acquired feature with an attribute. For example, the determination criterion derivation unit 34 derives a representative value of the feature for each attribute as the determination criterion. For example, in a non-walking section where head movement is the detection target, the determination criterion derivation unit 34 derives a representative value of the feature that is information about the forward / backward acceleration of many MCI patients as the determination criterion for MCI. Furthermore, in a walking section where whole-body movement is the detection target, the determination criterion derivation unit 34 derives a representative value of the feature that is information about the left-right movement of many MCI patients as the determination criterion for MCI.

[0051] The determination criterion derivation unit 34 derives, as a determination criterion for Parkinson's disease, a representative value of the feature amount, which is information about the rotation speed of many Parkinson's disease patients, in a non-walking section where head movement is the detection target. The determination criterion derivation unit 34 also derives, as a determination criterion for Parkinson's disease, a representative value of the feature amount, which is information about the forward and backward movement of many Parkinson's disease patients, in a walking section where whole-body movement is the detection target.

[0052] For able-bodied persons, in the non-walking section, the determination criterion derivation unit 34 derives a representative value of a feature amount that is information about the acceleration in the forward and backward directions of a large number of able-bodied persons and a representative value of a feature amount that is information about the rotational speed of a large number of able-bodied persons as the determination criterion for able-bodied persons.In the walking section, the determination criterion derivation unit 34 derives a representative value of a feature amount that is information about the movement in the left and right directions of a large number of able-bodied persons and a representative value of a feature amount that is information about the movement in the forward and backward directions of a large number of able-bodied persons as the determination criterion for able-bodied persons.

[0053] The representative value may be, but is not limited to, a maximum value, a minimum value, an average value, a median value, a mode value, a quartile value, or the like.

[0054] Furthermore, instead of the representative value of each feature, the distribution of the feature can be used as the criterion. The distribution of the feature can be determined by applying, for example, kernel density estimation to a large number of feature values. Kernel density estimation is a technique for estimating a feature distribution that smooths the changes in the distribution by applying a kernel function to the distribution of raw feature data, such as a histogram of the feature values. Figure 6 is a diagram for explaining the derivation of the distribution of the feature using kernel density estimation. In Figure 6, the horizontal axis represents the value of the feature, and the vertical axis represents the frequency of occurrence of the feature. By applying kernel density estimation to the distribution of raw feature data shown by the solid line in Figure 6, the distribution of the feature shown by the dashed line in Figure 6 can be derived. Note that in the distribution of the feature, the vertical axis represents the probability of occurrence. The probability can be derived by normalizing the distribution of the criterion, with the peak of the criterion set to, for example, 0.8.

[0055] When using such a distribution of feature values as the criterion for determination, the determination criterion derivation unit 34 derives, in the non-walking section, the distribution of feature values that are information regarding the forward / backward acceleration of a large number of MCI patients as the criterion for determination of MCI. Furthermore, in the walking section, the determination criterion derivation unit 34 derives, as the criterion for determination of MCI, the distribution of feature values that are information regarding the left / right movement of a large number of MCI patients. Furthermore, in the non-walking section, the determination criterion derivation unit 34 derives, as the criterion for determination of Parkinson's disease, the distribution of feature values that are information regarding the rotational speed of a large number of Parkinson's disease patients. Furthermore, in the walking section, the determination criterion derivation unit 34 derives, as the criterion for determination of Parkinson's disease, the distribution of feature values that are information regarding the forward / backward movement of a large number of Parkinson's disease patients.

[0056] For able-bodied persons, in the non-walking section, the determination criterion derivation unit 34 derives the distribution of feature values that are information about the acceleration in the forward and backward directions of a large number of able-bodied persons and the distribution of feature values that are information about the rotational speeds of a large number of able-bodied persons as the determination criteria for able-bodied persons.In the walking section, the determination criterion derivation unit 34 derives the distribution of feature values that are information about the movement in the left and right directions of a large number of able-bodied persons and the distribution of feature values that are information about the movement in the forward and backward directions of a large number of able-bodied persons as the determination criteria for able-bodied persons.

[0057] As a criterion for determining Parkinson's disease, in addition to the distribution of features related to the rotational speed of a large number of Parkinson's disease patients in the non-walking section, a distribution of features related to the forward-backward acceleration of a large number of Parkinson's disease patients may be derived. As a criterion for determining Parkinson's disease, in addition to the distribution of features related to the forward-backward movement of a large number of Parkinson's disease patients in the walking section, a distribution of features related to the left-right movement of a large number of Parkinson's disease patients may be derived. As a criterion for determining MCI, in addition to the distribution of features related to the forward-backward acceleration of a large number of MCI patients in the non-walking section, a distribution of features related to the rotational speed of a large number of MCI patients may be derived. As a criterion for determining MCI, in addition to the distribution of features related to the left-right movement of a large number of MCI patients in the walking section, a distribution of features related to the forward-backward movement of a large number of MCI patients may be derived.

[0058] Furthermore, as the judgment criteria, a discrimination model for distinguishing between MCI patients and healthy individuals, and a discrimination model for distinguishing between Parkinson's disease patients and healthy individuals can be used. For example, a discrimination model for distinguishing between MCI patients and healthy individuals can be constructed by machine learning a neural network using feature values acquired from MCI patients and feature values acquired from healthy individuals as training data, respectively, so that when the feature values of subject H are input, a score indicating MCI is output. The score ranges, for example, from 0 to 1, with the score closer to 1 indicating a higher likelihood of MCI. Note that the feature values used in constructing a discrimination model for determining MCI are information related to forward / backward acceleration in non-walking sections and information related to left / right movement in walking sections.

[0059] A discrimination model for distinguishing between Parkinson's disease patients and healthy individuals can be constructed by machine learning a neural network using feature values acquired from Parkinson's disease patients and feature values acquired from healthy individuals as training data, respectively, so that when the feature values of subject H are input, a score indicating Parkinson's disease is output. The score ranges, for example, from 0 to 1, with the score closer to 1 indicating a higher likelihood of Parkinson's disease. The feature values used in constructing a discrimination model for determining Parkinson's disease are information about rotation speed in non-walking sections and information about forward and backward movement in walking sections.

[0060] When deriving a score for a healthy individual, the score for MCI or the score for Parkinson's disease can be subtracted from 1.

[0061] Any machine learning method can be used, such as linear discriminant analysis, support vector machine, random forest, etc. As a determination criterion, a discrimination model 40 for identifying MCI patients and a discrimination model 41 for identifying Parkinson's disease patients may be derived separately, as shown in Fig. 7, or only one discrimination model 42 that outputs scores for both MCI patients and Parkinson's disease patients may be constructed.

[0062] Note that the subject from whom the features are acquired when deriving the determination criterion may suffer from both dementia and Parkinson's disease. The features of such subject H may have a different tendency from those of subjects suffering from only dementia and only Parkinson's disease. Therefore, when deriving the discrimination model as the determination criterion, only the features acquired from subjects suffering from only dementia and only Parkinson's disease may be used.

[0063] The disease determination unit 35 obtains a determination result as to whether the subject H is a healthy individual, MCI, or Parkinson's disease, based on the feature amounts obtained by the feature amount obtaining unit 33 and the determination criteria derived by the determination criteria derivation unit 34. When the determination criteria are representative values of the feature amounts, the disease determination unit 35 compares the representative values for the three attributes of healthy individual, MCI, and Parkinson's disease with the feature amounts obtained by the feature amount obtaining unit 33 for the subject H. Specifically, of the determination criteria for each attribute, such as healthy individual, MCI, or Parkinson's disease, the disease determination unit 35 obtains the attribute that serves as the determination criterion closest to the feature amounts of the subject H as the disease determination result for the subject H.

[0064] FIG. 8 is a diagram illustrating disease determination when the determination criterion is a representative value of a feature. As shown in FIG. 8, for MCI, the feature F0 of subject H is compared with the representative value R10, which is the determination criterion for a healthy subject, and the representative value R11, which is the determination criterion for an MCI patient. In this case, the feature F0 of subject H is close to the representative value R11. For Parkinson's disease, the feature F1 of subject H is compared with the representative value R20, which is the determination criterion for a healthy subject, and the representative value R21, which is the determination criterion for a Parkinson's disease patient. In this case, the feature F1 of subject H is close to the representative value R20. Therefore, the disease determination unit 35 obtains a determination result indicating that subject H is highly likely to have MCI.

[0065] When the determination criterion is the distribution of features for each attribute of healthy individuals, MCI, and Parkinson's disease, the disease determination unit 35 derives the occurrence probability of the feature of the subject H from the distribution of the feature. Then, the attribute with the highest occurrence probability of the feature is acquired as the disease determination result for the subject H.

[0066] FIG. 9 is a diagram illustrating disease determination when the determination criterion is the distribution of feature quantities. As shown in FIG. 9, for a certain feature quantity, a distribution 51 for healthy individuals, a distribution 52 for MCI, and a distribution 53 for Parkinson's disease are acquired as determination criteria. The disease determination unit 35 applies the feature quantity F2 of subject H to the distribution of the three feature quantities to derive the probability α1 of being healthy, the probability α2 of being MCI, and the probability α3 of being Parkinson's disease. In this case, α2 > α1 > α3. Therefore, the disease determination unit 35 obtains a determination result indicating that subject H is most likely to have MCI, which has the highest probability of occurrence.

[0067] The probability α1 of occurrence of a healthy individual, the probability α2 of occurrence of MCI, and the probability α3 of occurrence of Parkinson's disease may be used as the determination results.

[0068] When the determination criterion is a discriminant model, the disease determination unit 35 inputs the feature amounts of the subject H into the discriminant model and acquires a score α11 for MCI and a score α12 for Parkinson's disease as the determination results. The disease determination unit 35 may acquire the attribute with the larger score for the subject H as the determination result. For example, when the score α12 for Parkinson's disease is larger than the score α11 for MCI, the disease determination unit 35 may acquire a determination result that the subject H is highly likely to have Parkinson's disease.

[0069] The output control unit 36 displays the determination result output by the disease determination unit 35 on the display 24. FIG. 10 is a diagram showing a display screen of the determination result. As shown in FIG. 10, subject information 61, such as the name, gender, and date of birth of subject H, and a determination result 62 are displayed on a display screen 60. In FIG. 10, the determination result indicating that the subject is highly likely to have MCI is displayed. The output control unit 36 may notify and display the determination result using a pop-up or the like, instead of displaying the display screen 60 shown in FIG. 10. The output control unit 36 may output the determination result by voice, instead of or in addition to displaying the determination result. The output control unit 36 may output the determination result by transmitting it to an external terminal device, such as the mobile terminal 4 of subject H, instead of displaying it on the display 24. In this case, the output control unit 36 may output the determination result by transmitting it to the terminal device by email or via an application on the terminal device. The terminal device may display the received determination result by email or via an application.

[0070] In this embodiment, when the representative value of the feature and the distribution of the feature are used as the assessment criteria, information on the forward / backward acceleration in the non-walking section and information on the left / right movement in the walking section are used to assess MCI. Furthermore, information on the rotational speed in the non-walking section and information on the forward / backward movement in the walking section are used to assess Parkinson's disease. Therefore, assessment results based on two assessment criteria are obtained for MCI and Parkinson's disease, respectively. Generally, the assessment results using the two assessment criteria are the same depending on the disease suffered by subject H. However, there are cases where the assessment results using the two assessment criteria differ. In such cases, the disease assessment unit 35 simply obtains a result indicating that assessment is impossible. In this case, the output control unit 36 simply displays the impossible assessment result on the display 24. Alternatively, of the assessment results based on the two assessment criteria, the one closest to the representative value of the feature or with a higher probability of occurrence in the distribution of the feature may be used as the assessment result for subject H's disease.

[0071] Next, the processing performed in this embodiment will be described. Fig. 11 is a flowchart showing the processing performed in this embodiment. It is assumed that the judgment criterion is derived by the judgment criterion derivation unit 34 and stored in the storage 23. First, the measurement value acquisition unit 31 acquires the measurement value transmitted from the eyeglass-type device 1 (step ST1). Next, the section identification unit 32 identifies a first section and a second section in which the detection targets of the movement of the subject H are different (section identification; step ST2).

[0072] Next, the feature acquisition unit 33 acquires at least one feature representing a characteristic of a disease related to at least one of cognition and movement based on each of the measurement values of the first section and the second section (step ST3). That is, a plurality of feature representing a characteristic of MCI and Parkinson's disease is acquired for each of the walking section and non-walking section identified by the section identification unit 32.

[0073] Next, the disease determination unit 35 obtains a determination result indicating whether the subject H is a healthy subject, MCI, or Parkinson's disease based on the feature amount obtained by the feature amount obtaining unit 33 and the determination criterion derived by the determination criterion derivation unit 34 (step ST4). Then, the output control unit 36 displays the determination result on the display 24 (step ST5), and the process ends.

[0074] As described above, in this embodiment, measurement values are acquired from a wearable device such as the eyeglass-type device 1 worn by the subject H, a first section and a second section in which the detection target of the subject H's movement is different are identified, at least one feature amount representing a disease characteristic related to at least one of cognition and movement is acquired based on each of the measurement values of the first section and the measurement values of the second section, and a disease assessment result is acquired based on the feature amount and predetermined assessment criteria. Therefore, the measurement values necessary for disease assessment can be acquired through free movement without the subject H performing any operation himself / herself. This reduces the burden on the subject H who is assessing the disease.

[0075] Furthermore, to obtain a judgment result, it is only necessary to identify the first and second intervals and derive the feature amounts for each interval, and there is no need to judge factor information as in the method described in Patent Document 2. This reduces the processing cost for judgment and improves the judgment accuracy.

[0076] In the above embodiment, the landing candidates detected by the section identification unit 32 include sections in which landing candidates appear randomly, in addition to sections T1 and T2, as shown in FIG. 5 . For this reason, the relationship between time and landing candidates shown in FIG. 5 may be displayed on the display 24, and the operator of the analysis server 2 may be allowed to modify the sections. FIG. 12 is a diagram for explaining section modification. The relationship between time and landing candidates in the upper part of FIG. 12 is the same as the relationship between time and landing candidates shown in FIG. 5 . In the relationship 70 in the upper part of FIG. 12 , there is a section between sections T1 and T2 in which landing candidates appear sparsely. Five landing candidates 71 that appear sparsely are indicated by arrows. These sparsely appearing landing candidates can be considered to be some kind of noise that occurred while the subject H was not walking.

[0077] Therefore, the operator issues an instruction to delete five landing candidates 71 that appear sparsely in the displayed relationship 70. As a result, the section identification unit 32 deletes the five landing candidates 71 for which deletion was instructed. Then, by deleting the landing candidates 71, the section T2 expands as shown in the relationship 72 at the bottom of FIG. 12. Therefore, the section identification unit 32 identifies the section expanded by the correction as a new non-walking section T3. This makes it possible to more accurately identify walking sections and non-walking sections.

[0078] Note that instead of deleting sparse landing candidates as described above, the walking section may be specified by adding landing candidates. Also, the walking section may be specified by deleting sections that are not required as walking sections.

[0079] In the above embodiment, the glasses-type device 1 may perform noise removal processing on the data detected by the sensor 18, but this is not limited to this. The glasses-type device 1 may transmit the measurement values to the analysis server 2 without performing noise removal processing, and the analysis server 2 may perform noise removal processing on the measurement values.

[0080] Furthermore, in the above embodiment, the analytical server 2 is equipped with the criterion derivation unit 34, but this is not limited to this. The criterion derivation unit 34 may be omitted from the analytical server 2, and the analytical server 2 may acquire and store in the storage 23 criterion derived by an external server.

[0081] Furthermore, in the above embodiment, MCI and Parkinson's disease are diagnosed, but this is not limiting. Dementia may also be diagnosed in addition to MCI and Parkinson's disease. Dementia is a disease resulting from the progression of MCI, and when the above-mentioned feature values are calculated, feature values with different values or distributions than those of MCI are obtained. For this reason, the criterion derivation unit 34 may derive a criterion using a large number of feature values for dementia patients, and determine whether subject H has MCI, dementia, or Parkinson's disease using the feature values acquired from the measurement values acquired from subject H and the dementia criterion.

[0082] In the above embodiment, the eyeglass-type device 1 is worn by the subject H to acquire the measurement values of the subject H, but the present invention is not limited to this. In addition to the eyeglass-type device 1, a sensor may be worn on the waist or arm of the subject H to acquire the measurement values representing the movement of the waist or arm of the subject.

[0083] In addition, in the above embodiment, the eyeglass-type device 1 is worn by the subject H to acquire the measurement values of the subject H, but this is not limited to this. Instead of the eyeglass-type device 1, an earring-type device may be worn by the subject H to acquire the measurement values representing the movement of the subject's head.

[0084] In the above embodiment, the following various processors can be used as the hardware structure of the processing units that perform various processes, such as the measurement value acquisition unit 31, the section identification unit 32, the feature acquisition unit 33, the criterion derivation unit 34, the disease determination unit 35, and the output control unit 36. As described above, the various processors include a CPU, which is a general-purpose processor that executes software (programs) and functions as various processing units, as well as dedicated electrical circuits that are processors having a circuit configuration specifically designed to perform specific processes, such as a programmable logic device (PLD), which is a processor whose circuit configuration can be changed after manufacture, such as an FPGA (Field Programmable Gate Array), and an ASIC (Application Specific Integrated Circuit).

[0085] A single processing unit may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs or a combination of a CPU and an FPGA). Also, multiple processing units may be configured with a single processor.

[0086] Examples of configuring multiple processing units with a single processor include, first, a form in which one processor is configured with a combination of one or more CPUs and software, and this processor functions as multiple processing units, as typified by computers such as client and server. Second, a form in which a processor is used to realize the functions of an entire system including multiple processing units with a single IC (Integrated Circuit) chip, as typified by systems on chips (SoCs). In this way, various processing units are configured using one or more of the above-mentioned various processors as a hardware structure.

[0087] Furthermore, more specifically, the hardware structure of these various processors can be an electric circuit that combines circuit elements such as semiconductor elements.

[0088] The following are appendices to the present disclosure. (Additional note 1) at least one processor; The processor: identifying a first section and a second section in which a detection target of the subject's movement is different based on a measurement value of the subject's movement measured by a device wearable on the subject; acquiring at least one feature value representing a disease characteristic related to at least one of cognition and movement based on each of the measurement values in the first section and the measurement values in the second section; An information processing device that acquires a diagnosis result of the disease based on the feature amount and a predetermined diagnosis criterion. (Additional note 2) 2. The information processing device according to claim 1, wherein the first section is a section for detecting movement of the subject's entire body, and the second section is a section for detecting movement of the subject's head. (Additional note 3) the measurement value includes a change in acceleration of the subject in a vertical downward direction, The information processing device described in Appendix 2, wherein the processor derives landing information indicating that the subject has landed while moving based on the change in acceleration, and identifies a walking section of the subject as the first section and a non-walking section of the subject as the second section based on the continuity of the landing information. (Additional note 4) the measured value includes the acceleration of the head in the front-back direction in the second section, 4. The information processing device according to any one of claims 1 to 3, wherein the processor acquires a feature quantity representing a characteristic of the cognitive disorder based on the forward / backward acceleration of the head in the second interval. (Additional note 5) the measured value includes a head rotation speed in the second section, 5. The information processing device according to any one of claims 1 to 4, wherein the processor acquires a feature quantity representing a characteristic of a disease related to the movement based on a rotation speed of the head in the second section. (Additional note 6) the measurement value includes a left-right movement of the subject in the first section, 6. The information processing device according to any one of claims 1 to 5, wherein the processor acquires a feature quantity representing a characteristic of the cognitive disorder based on the left-right movement of the subject in the first section. (Additional note 7) the measured value includes a front-to-back movement of the subject in the first section; 7. The information processing device according to any one of appendices 1 to 6, wherein the processor acquires a feature quantity representing a characteristic of a disease related to the movement based on the forward / backward movement of the subject in the first section. (Additional note 8) 8. The information processing device according to claim 1, wherein the processor derives the determination criterion. (Additional note 9) 9. The information processing device according to any one of appendices 1 to 8, wherein the determination criterion is a reference value for distinguishing between the disease and non-disease. (Additional note 10) 9. The information processing device according to any one of appendices 1 to 8, wherein the determination criterion is a feature distribution estimated based on a plurality of feature amounts representing characteristics of the disease. (Additional note 11) 9. The information processing device according to any one of appendices 1 to 8, wherein the judgment criterion is a discrimination model that has been trained to output a score representing the possibility of the disease when the feature amount is input. (Additional note 12) 12. The information processing device according to any one of claims 1 to 11, wherein the disease is at least one of mild cognitive impairment, dementia, and Parkinson's disease. (Additional note 13) 13. The information processing device according to any one of claims 1 to 12, wherein the device has an acceleration sensor and an angular velocity sensor. (Additional note 14) 14. The information processing device according to any one of claims 1 to 13, wherein the device further includes an electrooculography sensor. (Additional note 15) a computer, based on a measurement value of the subject's movement measured by a device wearable on the subject, identifying a first section and a second section in which a detection target of the subject's movement is different; acquiring at least one feature value representing a disease characteristic related to at least one of cognition and movement based on each of the measurement values in the first section and the measurement values in the second section; An information processing method for obtaining a diagnosis result of the disease based on the feature amount and predetermined criteria. (Additional note 16) a step of identifying a first section and a second section in which detection targets of the subject's movement are different, based on a measurement value of the subject's movement measured by a device wearable on the subject; acquiring at least one feature value representing a disease characteristic related to at least one of cognition and movement based on each of the measurement values in the first section and the measurement values in the second section; obtaining a disease diagnosis result based on the feature amount and a predetermined diagnosis criterion; An information processing program that causes a computer to execute the above. [Explanation of symbols]

[0089] 1 Eyeglass-type device 1 2. Analysis server (information processing device) 3 Network 4. Mobile devices 5. Computer 21 CPU 23 Storage 25 Input Devices 16,26 memory 17,27 Communication I / F 18 Sensors 19,29 Bus 22 Information Processing Program 24 displays 31 Measurement value acquisition unit 32 Section Identification Unit 33 Feature acquisition unit 34 Judgment criteria derivation part 35 Disease determination department 36 Output control section 41-43 Identification Model 51-53 Distribution of Features 60 display screen 61 Subject information 62 Judgment result 70,72 Relationship between time and landing candidate 71 Landing Candidates 100 Disease Detection System F0~F2 features R10, R11, R20, R21 Representative values of feature quantities T1 Walking section T2 Non-pedestrian section

Claims

1. at least one processor; The processor: identifying a first section and a second section in which a detection target of the subject's movement is different based on a measurement value of the subject's movement measured by a device wearable on the subject; acquiring at least one feature amount representing a characteristic of a disease related to at least one of cognition and movement based on each of the measurement values of the first section and the measurement values of the second section; An information processing device that acquires a diagnosis result of the disease based on the feature amount and a predetermined diagnosis criterion.

2. The information processing device according to claim 1 , wherein the first section is a section for detecting movement of the subject's entire body, and the second section is a section for detecting movement of the subject's head.

3. the measurement value includes a change in acceleration of the subject in a vertical downward direction, 3. The information processing device according to claim 2, wherein the processor derives landing information indicating that the subject has landed while moving based on the change in acceleration, and identifies a walking section of the subject as the first section and a non-walking section of the subject as the second section based on the continuity of the landing information.

4. the measured value includes acceleration of the head in the front-rear direction in the second section, The information processing device according to claim 1 , wherein the processor acquires a feature quantity representing a characteristic of the cognitive disorder based on the acceleration of the head in the forward / backward direction in the second section.

5. the measured value includes a head rotation speed in the second section, The information processing device according to claim 1 , wherein the processor acquires a feature quantity representing a characteristic of a disease related to the movement based on the rotation speed of the head in the second section.

6. the measurement value includes a left-right movement of the subject in the first section, The information processing device according to claim 1 , wherein the processor acquires a feature quantity representing a characteristic of the cognitive disorder based on the left-right movement of the subject in the first section.

7. the measured value includes a front-to-back movement of the subject in the first section; The information processing device according to claim 1 , wherein the processor acquires a feature quantity representing a characteristic of a disease related to the movement based on the forward / backward movement of the subject in the first section.

8. The information processing apparatus according to claim 1 , wherein the processor derives the criterion.

9. The information processing device according to claim 1 , wherein the determination criterion is a reference value for distinguishing between the diseased state and the non-disease state.

10. The information processing apparatus according to claim 1 , wherein the determination criterion is a feature distribution estimated based on a plurality of feature quantities representing characteristics of the disease.

11. The information processing device according to claim 1 , wherein the determination criterion is a discrimination model that has been trained to output a score representing the possibility of the disease when the feature amount is input.

12. The information processing device according to claim 1 , wherein the disease is at least one of mild cognitive impairment, dementia, and Parkinson's disease.

13. The information processing apparatus according to claim 1 , wherein the device has an acceleration sensor and an angular velocity sensor.

14. The information processing apparatus according to claim 1 , wherein the device further comprises an electrooculography sensor.

15. a computer, based on a measurement value of the subject's movement measured by a device wearable on the subject, identifying a first section and a second section in which a detection target of the subject's movement is different; acquiring at least one feature amount representing a characteristic of a disease related to at least one of cognition and movement based on each of the measurement values of the first section and the measurement values of the second section; An information processing method for obtaining a diagnosis result of the disease based on the feature amount and predetermined criteria.

16. a step of identifying a first section and a second section in which detection targets of the subject's movement are different, based on a measurement value of the subject's movement measured by a device wearable on the subject; acquiring at least one feature value representing a characteristic of a disease related to at least one of cognition and movement based on each of the measurement values of the first section and the measurement values of the second section; obtaining a disease diagnosis result based on the feature amount and a predetermined diagnosis criterion; An information processing program that causes a computer to execute the above.

Images