JP2023038746A - Classification system, classification method, and program - Google Patents

Abstract

To classify pulses based on a way of thinking of pulse diagnosis in the oriental medicine by using a general-purpose pulse measuring device used in wearable devices of smart watches, etc.SOLUTION: A classification system includes: an acquisition part for acquiring pulse wave data measured at a measurement state corresponding to each of superficial level-feeling, middle level-feeling, and heavy level-feeling; a classification part for performing, on the pulse wave data acquired by the acquisition part, each of a first classification of classifying the pulse wave data on the basis of a pulse rate, a second classification of classifying the pulse wave data on the basis of a variation degree of the pulse duration, a third classification of classifying the pulse wave data on the basis of the difference in the pulse wave data in each of the measurement states of the superficial level-feeling, the middle level-feeling, and the heavy level feeling, and a fourth classification of classifying the pulse wave data on the basis of a waveform feature value; and a determination part for determining the type of the pulse wave data on the basis of classification results classified by the classification part.SELECTED DRAWING: Figure 3

Description

The present invention relates to a classification system, a classification method, and a program.

In oriental medicine, pulse diagnosis has been performed. In the pulse diagnosis, the condition of the disease is diagnosed by the specialist touching the patient's pulse and grasping the strength, speed, hardness, thickness, ups and downs of the pulse, and the like. For example, in pulse diagnosis, a specialist judges which of a plurality of types (for example, 28 types) of pulse patterns (pulse types) the patient's pulse corresponds to, and the examination is performed based on the judgment result. Such pulse diagnosis has been handed down by human senses and requires a long period of time to master, so it was difficult for an inexperienced person to judge the pulse pattern. Patent Literature 1 discloses a technique for quantitatively determining a pulse phenomenon using a signal processing technique so that even an inexperienced person can determine the pulse phenomenon.

On the other hand, Patent Document 2 discloses a photoelectric pulse measuring device. A photoelectric pulse measuring device irradiates a human body with light and measures the pulse by utilizing the difference in absorbance due to the pulsation of blood. Such photoelectric pulse measuring devices are currently widely used in wearable devices such as smart watches.

JP 2015-16268 A JP-A-9-122090

However, in Patent Document 1, the cuff of the sphygmomanometer is wrapped around the arm for measurement. Therefore, there is a problem that the measurement is troublesome compared to wearable devices such as smart watches. In addition, although the pulse measurement device of Patent Document 2 is compact, it is incorporated in a wearable device and only measures heart rate (pulse rate), blood oxygen saturation (SpO2), etc., and is based on the concept of pulse diagnosis. It has hardly been used as a pulse measuring device for judging the state of physical condition.

The present invention has been made in view of this situation, and uses a general-purpose pulse measuring device that is used in wearable devices such as smart watches to classify pulses based on the concept of pulse diagnosis in oriental medicine. To provide a classification system, a classification method, and a program capable of

In order to solve the above-described problems of the present invention, the present invention provides an acquisition unit that acquires pulse wave data measured in measurement states corresponding to each of flotation, mediation, and sinking, and acquisition by the acquisition unit The obtained pulse wave data is classified into a first classification based on the pulse rate, a second classification based on the degree of variation in pulse time, floating, intermediate and sinking. a classification unit that performs a third classification for classifying the pulse wave data based on differences in the pulse wave data in each measurement state, and a fourth classification for classifying the pulse wave data based on the characteristic amount of the waveform; and a determination unit that determines the type of the pulse wave data based on the result of classification by the classification unit.

The present invention is a classification method performed by a computer, wherein an acquisition unit acquires pulse wave data measured in measurement states corresponding to each of flotation, mediation, and sinking, and the classification unit performs In the pulse wave data acquired by, a first classification that classifies the pulse wave data based on the pulse rate, a second classification that classifies the pulse wave data based on the degree of variation in pulse time, floating, middle and a third classification for classifying the pulse wave data based on the difference in the pulse wave data in each measurement state of sedimentation, and a fourth classification for classifying the pulse wave data based on the feature value of the waveform. The classification method, wherein the determination unit determines the type of the pulse wave data based on the classification results of the classification by the classification unit.

The present invention causes a computer to acquire pulse wave data measured in measurement states corresponding to each of floating, medial, and submerged, and adds the pulse wave data to the acquired pulse wave data based on the pulse rate. A first classification for classifying the wave data, a second classification for classifying the pulse wave data based on the degree of variation in the pulse time, based on the difference in the pulse wave data in each of the floating, intermediate and submerged measurement states A third classification for classifying the pulse wave data and a fourth classification for classifying the pulse wave data based on the feature amount of the waveform are performed, and the type of the pulse wave data is determined based on the classified results. It is a program that allows

According to the present invention, it is possible to classify pulses based on the concept of pulse diagnosis in oriental medicine using a general-purpose pulse measuring device that is used in wearable devices such as smart watches.

1 is a block diagram showing an example configuration of a classification system 1 according to an embodiment; FIG. 1 is a block diagram showing an example of the configuration of a measuring terminal 10 according to an embodiment; FIG. 2 is a block diagram showing an example of the configuration of a classification server 20 according to the embodiment; FIG. It is a figure which shows the example of the pulse wave information 220 which concerns on embodiment. It is a figure explaining the pulse phenomenon which concerns on embodiment. It is a figure explaining the pulse phenomenon which concerns on embodiment. It is a figure explaining the pulse phenomenon which concerns on embodiment. It is a figure explaining the pulse phenomenon which concerns on embodiment. It is a figure which shows the example of the pulse wave data which concern on embodiment. It is a figure explaining the pulse phenomenon which concerns on embodiment. 4 is a flow chart showing the flow of processing performed by the classification server 20 according to the embodiment. It is a block diagram which shows the example of a structure of the classification system 1 which concerns on the modification 1 of embodiment. FIG. 11 is a block diagram showing an example of the configuration of a learning server 30 according to modification 1 of the embodiment; FIG. 11 is a sequence diagram showing the flow of processing performed by the classification system 1 according to Modification 1 of the embodiment;

Hereinafter, embodiments of the invention will be described with reference to the drawings.

[Overview of classification system 1]
An outline of the classification system 1 will be described. In this embodiment, the classification system 1 provides a service for supporting the user's health (hereinafter referred to as a health support service) based on the pulse diagnosis concept of oriental medicine. In the health support service, the user's pulse pattern (pulse type) is determined based on the concept of pulse diagnosis, and the user is provided with the result of the determination and advice on health management based on the determination result.

[Configuration of classification system 1]
A configuration of the classification system 1 will be described. FIG. 1 is a block diagram showing an example configuration of a classification system 1 according to an embodiment. The classification system 1 includes, for example, a measurement terminal 10 and a classification server 20 . The measurement terminal 10 and the classification server 20 are communicably connected via a communication network NW.

The measuring terminal 10 is a computer. For example, the measurement terminal 10 is a wearable terminal, smart phone, personal computer, mobile phone, tablet terminal, or the like. Application software that provides health support services (hereinafter referred to as a health support application) is installed in the measurement terminal 10, for example. The measurement terminal 10 measures the user's pulse wave via the health support application, and transmits information (pulse wave data) indicating the measured pulse wave to the classification server 20 .

For example, as shown in FIG. 1, the measurement terminal 10 is communicably connected to the pulse wave sensor 100 via wireless communication such as Bluetooth (registered trademark) or USB (Universal Serial Bus).

Pulse wave sensor 100 measures the user's pulse. The pulse wave sensor 100 may be any sensor as long as it can measure at least a pulse, but for example, it is an optical photoelectric pulse measurement device. The pulse wave sensor 100 is attached to a measurement position for measuring the pulse, and irradiates the measurement position with light. The measurement position here is the wrist, more specifically, the pulsating site of the radial artery touched inside the radial styloid process of the left and right forearms. The pulse wave sensor 100 receives light transmitted through the measurement position or light reflected at the measurement position. The pulse wave sensor 100 utilizes the property that a part of the irradiated light is absorbed according to the blood flow rate of the blood at the measurement position, and measures the pulse based on the relationship between the light amount of the received light and the blood flow rate. The pulse wave sensor 100 transmits information indicating the measured pulse to the measurement terminal 10 as pulse wave data. Measurement terminal 10 receives pulse wave data from pulse wave sensor 100 . Thereby, the measuring terminal 10 measures the user's pulse wave.

Measurement terminal 10 may incorporate pulse wave sensor 100 and acquire pulse wave data measured by pulse wave sensor 100 via pulse wave sensor 100 to measure the user's pulse wave.

Classification server 20 is a computer. For example, the classification server 20 is a cloud device, a server device, a personal computer, or the like. Classification server 20 provides, for example, health support services. For example, the classification server 20 receives the user's pulse wave data from the measurement terminal 10 and determines the pulse pattern in the received pulse wave data. The classification server 20 transmits the determination result of determining the pulse phenomenon to the measurement terminal 10 .

The classification server 20 determines a pulse pattern (type of pulse wave data) based on the classification result of the pulse wave data, and transmits the result to the measurement terminal 10 . The example of this figure shows a display example of the measurement terminal 10 when the user's pulse phenomenon is determined to be a "thrush".

[Configuration of measurement terminal 10]
A configuration of the measuring terminal 10 will be described. FIG. 2 is a block diagram showing an example configuration of the measuring terminal 10 according to the embodiment. The measurement terminal 10 includes, for example, a communication unit 11, a storage unit 12, a control unit 13, a display unit 14, and an input unit 15.

The communication unit 11 communicates with the classification server 20 via the communication network NW. The communication unit 11 also receives pulse wave data measured by the pulse wave sensor 100 .

The storage unit 12 is a storage medium such as a HDD (Hard Disk Drive), flash memory, EEPROM (Electrically Erasable Programmable Read Only Memory), RAM (Random Access read/write Memory), ROM (Read Only Memory), or a combination thereof. Consists of The storage unit 12 stores programs for executing various processes of the measuring terminal 10 and temporary data used when performing various processes.

The control unit 13 is implemented by causing a CPU (Central Processing Unit) provided as hardware in the measuring terminal 10 to execute a program. The control unit 13 controls the measuring terminal 10 as a whole. The control unit 13 controls the display unit 14 and the input unit 15 according to, for example, a program related to a health support application.

The control unit 13 executes processing according to the program of the health support application. For example, when the icon of the health support app is tapped, the control unit 13 activates the health support app. As a result, login to the health support application is executed. When the login is successful, for example, an activation image of the health support application is displayed on the display unit 14 .

The control unit 13 includes an acquisition unit 130, a measurement support unit 131, and a device control unit 132, for example. Acquisition unit 130 acquires various types of information. Acquisition unit 130 acquires pulse wave data measured by pulse wave sensor 100 via communication unit 11 . Also, for example, the acquisition unit 130 acquires information input by the user's application operation via the input unit 15 .

The measurement assistance unit 131 assists the user in measuring the pulse. In general, there are various diagnostic methods for pulse diagnosis. It is practiced to examine the pulse of each of the measurement locations. Also, in the six-part stereotactic pulse diagnosis, the pulse is examined in each of the three measurement states of floating, middle, and submerged. Floatation is to feel the pulse by lightly placing a finger on the measuring point. Submersion is to examine the pulse by immersing the finger with force. In addition, nakadori refers to examining the pulse by placing a finger on the measurement position with a force that is just between the flotation and sinking. Then, the pulse pattern is judged synthetically based on the respective pulses of floating, middle and sinking in each of the sun, seki, and shaku of both arms.

In this embodiment, the measurement support unit 131 supports the user so that the measurement is performed according to the measurement position and measurement state. For example, the measurement support unit 131 causes the display unit 14 to display a guidance image.

Specifically, when measuring the float at the measurement position of the left hand "dimension", for example, the measurement support unit 131 displays on the display unit 14 an image showing the measurement position with the measurement position of the left hand marked. do. Then, the measurement support unit 131 "measures the floatation of the left hand 'sun'. Refer to the image and wear the pulse wave sensor 100 so that it lightly touches the position of "size" on the left hand." is displayed. Measurement support unit 131 displays a measurement start message such as "Start measurement" when a stable waveform is acquired by pulse wave sensor 100, and displays the measurement start message for a predetermined time (for example, 1 minute), display a measurement completion message such as "Measurement completed".

Alternatively, when measuring the sinking at the measurement position of the right hand “shaku”, for example, the measurement support unit 131 displays on the display unit 14 an image showing the measurement position marked at the right hand shaku position. Then, the measurement support unit 131 "measures the sinking of the 'shaku' of the right hand. Referring to the image, please wear the pulse wave sensor 100 so that it touches the position of the "scale" on the right hand, and press the pulse wave sensor 100 from above". Measurement support unit 131 displays a measurement start message such as "Start measurement" when a stable waveform is acquired by pulse wave sensor 100, and displays the measurement start message for a predetermined time (for example, 1 minute), display a measurement completion message such as "Measurement completed".

The measurement support unit 131 transmits the pulse wave data thus acquired to the measurement terminal 10 together with its attribute information. The attribute information here is information related to the pulse wave data, and includes measurement positions (indicating distinctions such as dimension, seki, shaku, etc.), measurement states, etc. It is information indicating The attribute information may also include information indicating measurement time, measurement date and time, and the like.

It should be noted that, in pulse diagnosis, measurements at some measurement positions and some measurement states may be omitted. For example, only the left hand pulse may be measured and the right hand omitted. Even if the measurement position and measurement state are omitted in this way, the pulse phenomenon can be determined, although the number of pulse phenomena that can be determined may decrease. Based on this point of view, in this embodiment, the measurement position and measurement state may be selected according to the user's wishes. In this case, the pulse pattern is determined based on the measured pulse wave data and its attribute information.

Also, the measurement support unit 131 may notify the user of the measurement timing. The measurement support unit 131 notifies the user of the measurement timing by, for example, sounding an alarm or displaying a message such as "Let's take measurements." The measurement support unit 131 measures the pulse at different times of the day, for example, at rest immediately after the user wakes up, during the day, before going to bed, and so on.

The device control section 132 controls the measuring terminal 10 in a centralized manner. For example, device control section 132 outputs the pulse wave data received by communication section 11 to acquisition section 130 . The device control unit 132 outputs the pulse wave data output by the measurement support unit 131 and its attribute information to the communication unit 11 , thereby transmitting the pulse wave data to the classification server 20 .

The display unit 14 includes a display device such as a liquid crystal display, and displays an image under the control of the control unit 13 . The input unit 15 includes an input device such as a mouse, a keyboard, and a touch panel, and outputs information input by a user's application operation to the control unit 13 .

[Configuration of Classification Server 20]
A configuration of the classification server 20 will be described. FIG. 3 is a block diagram showing an example configuration of the classification server 20 according to the embodiment. The classification server 20 includes a communication unit 21, a storage unit 22, and a control unit 23, for example.

The communication unit 21 communicates with the measurement terminal 10 via the communication network NW.

The storage unit 22 is configured by a storage medium such as HDD, flash memory, EEPROM, RAM, ROM, or a combination thereof. The storage unit 22 stores programs for executing various processes of the classification server 20 and temporary data used when performing various processes.

The control unit 23 is implemented by causing a CPU provided as hardware in the classification server 20 to execute a program. The control unit 23 comprehensively controls the classification server 20 .

The control unit 23 executes processing according to the provision of the health support service. The control unit 23 includes an acquisition unit 230, a classification unit 231, a determination unit 232, and a device control unit 233, for example. Acquisition section 230 acquires the pulse wave data transmitted from measurement terminal 10 via communication section 21 and outputs the acquired pulse wave data to classification section 231 .

The classification unit 231 classifies the pulse wave data. The classification unit 231 classifies the pulse wave data based on the pulse rate, regularity, pressure, and waveform. The pulse rate here is the number of times the pulse wave data pulsates in a predetermined unit time (for example, 1 minute). Regularity is the rhythm of the pulse and the degree of variation in the time required for one pulse (called pulse time). The pressure is whether or not the pulse wave data differs between floating, medial and submerged. A waveform is a time-series change in pulse intensity. The classification unit 231 stores the classification result in the pulse wave information 220 . A specific method by which the classification unit 231 classifies the pulse wave data will be described later in detail.

The determination unit 232 determines the pulse pattern based on the classification result of the classification unit 231 . The determination unit 232 stores the determined pulse pattern in the pulse wave information 220 . A specific method by which the determination unit 232 determines the pulse phenomenon will be described later in detail.

The device control unit 233 controls the classification server 20 in an integrated manner. For example, device control section 233 outputs the pulse wave data received by communication section 21 to acquisition section 230 . The device control unit 233 transmits the determination result (pulse pattern) determined by the determination unit 232 and health management advice based on the determination result to the measurement terminal 10 .

[Configuration of pulse wave information 220]
A configuration of the pulse wave information 220 will be described. FIG. 4 is a diagram showing an example of pulse wave information 220 according to the embodiment. The pulse wave information 220 stores, for example, information corresponding to each item such as pulse ID, measurement date and time, measurement position, measurement state, user ID, classification result, and pulse pattern. The pulse ID is identification information that uniquely identifies pulse wave data. The date and time of measurement is information indicating the date and time when the pulse wave data identified by the pulse ID was measured.

The measurement position is information indicating the measurement position (such as size, seki, shaku) at which the pulse wave data identified by the pulse ID was measured. The measurement state is information indicating the measurement state (floating, mediating, sinking, etc.) in which the pulse wave data identified by the pulse ID was measured. The user ID is information identifying the user whose pulse wave data identified by the pulse ID was measured.

The classification result is information indicating the classification result by the classification unit 231 . The classification results are displayed in association with items such as pulse rate, regularity, pressure, waveform, and the like. The pulse pattern is information indicating a pulse pattern as a result of determination by the determination unit 232 .

[Method of Classifying Pulse Wave Data by Classifying Unit 231]
A method for classifying the pulse wave data by the classification unit 231 will be described with reference to FIGS. 5 to 9. FIG. 5 to 9 are diagrams for explaining pulse phenomena in pulse diagnosis.

FIG. 5 shows an example of classification by pulse rate. As shown in FIG. 5, in the pulse diagnosis, the pulse is classified into four pulse symptoms of "slow pulse", "moderate pulse", "sparse pulse", and "fast pulse" based on the pulse rate. For example, a pulse with a pulse rate of 60 [beats/minute] or less is classified as "slow pulse". A pulse with a pulse rate of about 65 [beats/minute] is classified as a "moderate pulse". Pulses with a pulse rate of 90 [beats/minute] or more are classified as "several pulses". A pulse with a pulse rate of 110 [beats/minute] or more is classified as a "pulse". A pulse with a pulse rate of about 65 to less than 90 [beats/minute] is classified as a "medium pulse". The midpoint is a reference pulse, and the state of health of the user is determined based on the time of the midpoint.

The classification unit 231 calculates the pulse rate from the pulse wave data (see FIG. 9). For example, the classification unit 231 sets the number of pulse wave data peaks (point P2 in FIG. 9) included in a predetermined unit time (for example, one minute) as the pulse rate. Then, the classification unit 231 classifies the pulse wave data according to the classification shown in FIG. 5 based on the calculated pulse rate. The classification performed by the classification unit 231 here is an example of the “first classification”.

FIG. 6 shows an example of classification by regularity. As shown in FIG. 6, in the pulse diagnosis, the pulse is classified into three pulse phenomena, ie, "prompted pulse", "congestive pulse", and "alternate pulse", based on regularity. For example, a pulse that has a pulse rate of 90 [beats/minute] or more and is irregularly lacking is classified as a “fast pulse”. Irregularly missing pulses with a pulse rate of 60 [beats/minute] or less are classified as "connections". In addition, a pulse that is regularly missing regardless of the pulse rate is classified as a "shiromyaku".

The classification unit 231 calculates variations in pulse time as an index of regularity from the pulse wave data. For example, the classification unit 231 calculates the pulse frequency characteristics by frequency-analyzing the pulse wave data. The classification unit 231 calculates the S/N ratio of the pulse frequency characteristic. Here, S (signal) is the magnitude of the main component of the pulse rhythm, specifically the strength of the frequency with the largest frequency component in the frequency characteristics. Here, N (noise) is the magnitude of components other than the main component in the rhythm of the pulse, specifically the strength of the frequency with the smallest frequency component in the frequency characteristics. Then, the classification unit 231 calculates variations in pulse duration based on the calculated S/N ratio. For example, when the S/N ratio is equal to or greater than a predetermined threshold, the classification unit 231 determines that there is little variation, that is, there is no omission. On the other hand, when the S/N ratio is less than the predetermined threshold, the classification unit 231 determines that the variation is large, that is, there is a missing part.

When the classification unit 231 determines that the variations in the pulse wave data are large, the classification unit 231 determines whether the omission is irregular or regular. For example, the classification unit 231 calculates the S2/N ratio based on the frequency characteristics of pulse wave data. Here, S2 (second signal) is the intensity of the frequency with the second largest frequency component. For example, when the S2/N ratio is equal to or greater than a predetermined threshold, the classification unit 231 determines that the omission is regular. On the other hand, when the S2/N ratio is less than the predetermined threshold, the classification unit 231 determines that the omission is irregular. Then, the classification unit 231 classifies the pulse wave data according to the classification shown in FIG. The classification performed by the classification unit 231 here is an example of the “second classification”.

FIG. 7 shows an example of classification by pressure. As shown in FIG. 7, in the pulse diagnosis, "floating pulse", "sinking pulse", "abnormal pulse", "abscence pulse", It is classified into five pulse phenomena of "real pulse". For example, a vein having the highest strength in flotation and a lower strength in middle and sedimentation is classified as a “floating vein”.

The classification unit 231 acquires the pulse wave data measured in each of the floating, intermediate and submerged measurement states from the attribute information of the pulse wave data. The classification unit 231 compares the intensity of the acquired pulse wave data, for example, the amplitude in the pulse wave data (the difference between the intensity VM and VS in FIG. 9). The classification unit 231 performs classification according to the divisions shown in FIG. 7 based on the comparison result. The classification performed by the classification unit 231 here is an example of the “third classification”.

8 and 9 show examples of classification by waveform. As shown in FIG. 8, in the pulse diagnosis, based on the thickness, tension, length, blood flow state, etc. of the pulse, "vein", "great", "large", "cord" It is classified into nine pulse phenomena. FIG. 9 schematically shows pulse wave data. The horizontal axis in FIG. 9 indicates time, and the vertical axis indicates pulse strength. For example, in the pulse wave data, a time-series change consisting of a point P1 corresponding to the starting point of the pulse, a point P2 corresponding to the peak of the pulse, and a point P3 corresponding to the ending point of the pulse is repeatedly shown.

The thickness of the pulse shown in FIG. 8 is the magnitude of the amplitude in the pulse wave data. The tension is the steepness of the rise in the pulse wave data (rate of change in strength from point P1 to point P2 in FIG. 9). The length here is the difference in pulse wave data depending on the measurement position. The blood flow state is a time-series change in rise and fall (from point P2 to point P3 in FIG. 9) in the pulse wave data. The blood flow state shown in FIG. 8 is determined whether or not there is a second rise (from point P4 to point P5 in the broken-line waveform data in FIG. 9) in the process of falling from point P2 to point P3. When there is a rise, it is a time-series change.

Classification unit 231 extracts the waveform of pulse wave data. For example, the classification unit 231 extracts the waveform of the pulse wave data by normalizing the pulse wave data by the time corresponding to the pulse cycle (from time TS to TE in FIG. 9). Alternatively, the classification unit 231 extracts the waveform of the pulse wave data by normalizing the pulse wave data by the time corresponding to the amplitude of the pulse. By normalizing the pulse wave data, waveform classification can be simplified.

For example, the classification unit 231 classifies according to the classification of the thickness of the pulse in FIG. 7 based on the magnitude of the amplitude. The classification unit 231 performs classification according to the division of pulse tension in FIG. 7 based on the steepness of the rise in the normalized pulse wave data. The classification unit 231 acquires pulse wave data corresponding to each measurement position such as sun, seki, shaku, etc., and compares the waveforms of the acquired pulse wave data to determine the length of the pulse according to the division of the pulse length in FIG. classification. The classification unit 231 classifies the pulse according to the classification of the blood flow state in FIG. 7 based on the time-series changes such as the change rate of the rise and fall in the normalized pulse wave data. The classification performed by the classification unit 231 here is an example of the "fourth classification".

[Method of judging the pulse phenomenon by the judging unit 232]
A method for determining the pulse phenomenon by the determination unit 232 will be described. The determination unit 232 determines the pulse pattern based on the classification result of the classification unit 231 . If one pulse wave data is classified into one or a plurality of pulse phenomena as a result of classification by the classification unit 231 for each of the pulse rate, regularity, pressure, and waveform, the determination unit 232 determines that the one or more Let the pulse pattern be the pulse pattern of the pulse wave data. Alternatively, when the pulse pattern is determined in a complex manner based on the classification results of the pulse rate, regularity, pressure, and waveform, the determination unit 232 determines the pulse pattern according to the classification.

FIG. 10 shows an example of compound classification. As shown in FIG. 10, in pulse diagnosis, for example, a pulse classified into several pulses based on the pulse rate, and a pulse classified into smooth and short pulses based on the waveform is classified as "arterial". The determination unit 232 determines the pulse pattern according to the classification shown in FIG.

[Flow of Processing Performed by Classification Server 20]
A flow of processing performed by the classification server 20 will be described with reference to FIG. First, the classification server 20 acquires pulse wave data (step S10). The classification server 20 classifies the obtained pulse wave data based on the pulse rate (step S11). The classification server 20 classifies the obtained pulse wave data based on the regularity (step S12). The classification server 20 classifies the acquired pulse wave data based on the pressure (step S13). The classification server 20 classifies the acquired pulse wave data based on the waveform (step S14). Then, the classification server 20 determines the pulse pattern of the pulse wave data based on the classification results of the classification performed in steps S11 to S14 (step S15).

As described above, the classification system 1 according to the embodiment includes the acquisition unit 230, the classification unit 231, and the determination unit 232. Acquisition unit 230 acquires pulse wave data measured in measurement states corresponding to each of floating, medial, and submerged. The classification unit 231 classifies the pulse wave data acquired by the acquisition unit 230 . The classification unit 231 performs a first classification for classifying the pulse wave data based on the representative value of the pulse rate. The classification unit 231 performs a second classification that classifies the pulse wave data based on the degree of pulse rate variation. The classification unit 231 performs a third classification for classifying the pulse wave data based on the difference in the pulse wave data in each state of floating, medial, and submerged. The classification unit 231 performs a fourth classification for classifying the pulse wave data based on the feature amount of the waveform. The determination unit 232 determines a pulse pattern (type of pulse wave data) based on the classification result of the classification unit 231 . As a result, the classification system 1 according to the embodiment can classify pulses based on the concept of pulse diagnosis in oriental medicine using a general-purpose pulse measurement device used in wearable devices such as smart watches. can.

[Modification 1 of Embodiment]
Modification 1 of the embodiment will be described. This modified example differs from the above-described embodiment in that classification is performed by waveform using a trained model. A trained model is a model that has learned the correspondence between waveforms and pulse phenomena. By performing waveform-based classification using the trained model, it is possible to omit waveform-related signal processing and reduce the processing load on the classification server 20 .

FIG. 12 is a block diagram showing an example configuration of the classification system 1 according to Modification 1 of the embodiment. In this modified example, the classification system 1 further includes a learning server 30 . The learning server 30 is a computer. For example, the learning server 30 is a cloud device, a server device, a personal computer, or the like. The learning server 30 generates a trained model. The learning server 30 receives pulse wave data to be classified from the classification server 20 . The learning server 30 uses the learned model to estimate a pulse phenomenon to which the pulse wave data to be classified corresponds, and notifies the classification server 20 of the estimation result.

FIG. 13 is a block diagram showing an example of the configuration of the learning server 30 according to Modification 1 of the embodiment. The learning server 30 includes, for example, a communication unit 31, a storage unit 32, and a control unit 33.

The communication unit 31 communicates with the classification server 20 via the communication network NW. The storage unit 32 is configured by a storage medium such as HDD, flash memory, EEPROM, RAM, ROM, or a combination thereof. The storage unit 32 stores programs for executing various processes of the learning server 30 and temporary data used when performing various processes.

The control unit 33 is implemented by causing a CPU provided as hardware in the learning server 30 to execute a program. The control unit 33 comprehensively controls the learning server 30 . The control unit 33 includes an acquisition unit 330, a learning unit 331, an estimation unit 332, and a device control unit 333, for example.

Acquisition unit 330 acquires various types of information. Acquisition unit 330 acquires a learning data set for learning via communication unit 31 . A learning data set for learning is data used to generate a trained model. Acquisition unit 330 also acquires, via communication unit 31 , pulse wave data to be classified, which is transmitted from classification server 20 .

The learning unit 331 generates a trained model. The learning unit 331 acquires a learning data set for learning via the acquisition unit 330 . A learning data set for learning is a set of information in which pulse wave data for learning (input data) is associated with pulse wave data for learning (output data). For example, a learning data set for learning is created in advance by setting the pulse pattern in the pulse wave data for learning using the results determined by a specialist who can perform pulse diagnosis.

The learning unit 331 generates a learned model by performing machine learning using a learning data set on a learning model such as a CNN (Convolutional Neural Network). The machine learning method here is, for example, supervised learning.

The learning unit 331 inputs the input data of the data set for learning to the learning model. The learning unit 331 adjusts the parameters (weight W, bias component b, etc.) of the learning model so that the data output from the learning model when input data is input approaches the output data of the learning data set. Thereby, the learning unit 331 causes the learning model to learn the correspondence relationship between the pulse wave data and the pulse phenomenon based on the classification by the waveform.

The learning unit 331 repeatedly learns the learning model until a predetermined termination condition is satisfied. The end condition here is, for example, a condition such that the error between the data output from the learning model and the (output data) of the data set for learning is within a predetermined range. The learning unit 331 sets the learning model trained until a predetermined end condition is satisfied as a learned model, and stores information indicating the learned model in the storage unit 32 as a learned model 320 .

Note that the learning model is not limited to CNN. Techniques such as decision trees, hierarchical Bayes, and SVM (Support Vector Machine) may be used as learning models, for example.

The estimation unit 332 performs estimation using the trained model 320 . Estimating unit 332 acquires pulse wave data to be classified via acquiring unit 330 . The estimation unit 332 inputs the acquired pulse wave data to the learned model 320 . As a result, the trained model 320 outputs an estimation result of estimating a pulse phenomenon when the input pulse wave data is classified based on the waveform. The estimation unit 332 acquires the estimation result output from the trained model 320 . Thereby, the estimation unit 332 performs estimation using the trained model 320 .

The device control unit 333 comprehensively controls the learning server 30 . For example, device control section 333 outputs the pulse wave data received by communication section 31 to acquisition section 330 . The device control section 333 transmits the estimation result estimated by the estimation section 332 to the classification server 20 .

FIG. 14 is a sequence diagram showing the flow of processing performed by the classification system 1 according to Modification 1 of the embodiment. Steps S20 to S23 and S28 in FIG. 14 are the same as steps S10 to S13 and S15 in FIG. 11, so description thereof will be omitted.

The classification server 20 transmits the pulse wave data to the learning server 30 and requests the learning server 30 to classify based on the waveform (step S24). The learning server 30 acquires the pulse wave data transmitted from the classification server 20 (step S25), and uses the learned model to estimate the classification result (pulse pattern) of the waveform-based classification (step S26). The learning server 30 transmits the estimation result to the classification server 20 . As a result, the classification server 20 uses the estimation result notified from the learning server 30 as the classification result obtained by classifying the pulse wave data based on the waveform.

Note that although the learned model 320 is stored in the learning server 30 in the above description, the present invention is not limited to this. The classification server 20 may store the learned model 320 . In this case, the classification server 20 can estimate the pulse using the learned model 320 stored in its own device without accessing the learning server 30 .

As described above, the classification system 1 according to Modification 1 of the embodiment further includes the learning unit 331 . A learning unit 331 generates a trained model 320 . The learned model 320 is a model that predicts the classification result of waveform classification in pulse wave data. The learning unit 331 generates a learned model by causing the learning model to learn the correspondence relationship between the pulse wave data for learning and the classification result of classification by waveform in the pulse wave data for learning. The classification unit 231 uses the trained model 320 to classify the pulse wave data according to the waveform. As a result, the classification system 1 according to Modification 1 of the embodiment can classify by waveform using the learned model, and the classification server 20 does not need to perform signal processing related to the waveform. Therefore, it is possible to reduce the processing load.

In the above description, the case where the pulse wave data is input to the trained model and the waveform of the pulse wave data is classified is exemplified, but the present invention is not limited to this. By inputting the feature amount extracted from the pulse wave data to the trained model, the pulse wave data may be classified according to the waveform. The feature amount in this case is, for example, data obtained by normalizing the pulse wave data, information indicating the maximum value, the minimum value, the maximum value and the minimum value of the inclination, etc. in the pulse wave data. In this case, the trained model is a model that has learned the correspondence relationship between the feature quantity extracted from the pulse wave data for learning and the pulse pattern in the pulse wave data for learning. In this way, by using the feature amount extracted from the pulse wave data, it is possible to reduce the amount of data to be input to the trained model compared to the pulse wave data itself.

In addition, a program for realizing all or part of the functions of the classification system 1, the measurement terminal 10, the classification server 20, and the learning server 30 described above is recorded on a computer-readable recording medium, and recorded on this recording medium. The processing may be performed by loading the program into a computer system and executing it. Here, "loading and executing the program recorded on the recording medium into the computer system" includes installing the program in the computer system. The "computer system" here includes hardware such as an OS and peripheral devices. A "computer system" may also include a plurality of computer devices connected via a network including communication lines such as the Internet, WAN, LAN, and dedicated lines. The term "computer-readable recording medium" refers to portable media such as flexible discs, magneto-optical discs, ROMs and CD-ROMs, and storage devices such as hard discs incorporated in computer systems. Thus, the recording medium storing the program may be a non-transitory recording medium such as a CD-ROM. Recording media also include internal or external recording media accessible from the distribution server for distributing the program. The program code stored in the recording medium of the distribution server may be different from the program code in a format executable by the terminal device. That is, as long as it can be downloaded from the distribution server and installed in a form that can be executed on the terminal device, the form stored in the distribution server does not matter. It should be noted that the program may be divided into a plurality of programs, and the divided programs may be downloaded at different timings and then integrated in the terminal device, or the distribution servers that distribute the divided programs may be different. Furthermore, a "computer-readable recording medium" is a volatile memory (RAM) inside a computer system that acts as a server or client when the program is transmitted via a network, and retains the program for a certain period of time. It shall also include things. Further, the program may be for realizing part of the functions described above. Furthermore, it may be a so-called difference file (difference program) that can realize the above-described functions in combination with a program already recorded in the computer system.

1 Classification system 10 Measurement terminal 131 Measurement support unit 20 Classification server 230 Acquisition unit 231 Classification unit 232 Judgment unit 233 Apparatus control unit 30 Learning server 320 Trained model

Claims (6)

an acquisition unit that acquires pulse wave data measured in measurement states corresponding to each of flotation, mediation, and sinking;
In the pulse wave data acquired by the acquisition unit, a first classification that classifies the pulse wave data based on the pulse rate, a second classification that classifies the pulse wave data based on the degree of variation in pulse time, and a floating , a third classification that classifies the pulse wave data based on the difference in the pulse wave data in each of the measurement states of medial and sedimentation, and a fourth classification that classifies the pulse wave data based on the feature amount of the waveform. a classifier that performs
a determining unit that determines the type of the pulse wave data based on the classification result classified by the classifying unit;
A classification system with The acquisition unit acquires pulse wave data measured at measurement positions corresponding to each of the sun, seki, and shaku,
The classification unit, in the fourth classification, classifies the pulse wave data based on the feature amount of the waveform in the pulse wave data measured at the site corresponding to each of the dimension, seki, and shaku.
The classification system of claim 1. The classification unit, in the fourth classification, classifies the pulse wave data based on the waveform feature amount in the normalized pulse wave data,
3. A classification system according to claim 1 or claim 2. By making the learning model learn the correspondence relationship between the feature amount extracted from the pulse waveform in the pulse wave data for learning and the classification result of the fourth classification in the pulse wave data for learning, the pulse in the pulse wave data A learning unit that generates a trained model that predicts the classification result of the fourth classification from the feature amount of the waveform of
further comprising
The classification unit performs the fourth classification using the trained model.
A classification system according to any one of claims 1 to 3. A computer-implemented classification method comprising:
The acquisition unit acquires pulse wave data measured in measurement states corresponding to each of floating, medial, and sedimentation,
A classification unit classifies the pulse wave data acquired by the acquisition unit into a first classification based on the pulse rate and a second classification based on the degree of variation in pulse time. A third classification that classifies the pulse wave data based on the difference in the pulse wave data in each of the measurement states of classification, flotation, mediation, and sinking. Do each of the 4 classifications,
A determining unit determines the type of the pulse wave data based on the classification result classified by the classifying unit;
Classification method. to the computer,
Acquire pulse wave data measured in measurement states corresponding to each of flotation, mediation and sinking,
In the acquired pulse wave data, a first classification that classifies the pulse wave data based on the pulse rate, a second classification that classifies the pulse wave data based on the degree of variation in pulse time, flotation, and mediation and a third classification for classifying the pulse wave data based on the difference in the pulse wave data in the respective measurement states of sedimentation, and a fourth classification for classifying the pulse wave data based on the feature amount of the waveform,
determining the type of the pulse wave data based on the classified result;
program.

Images