JP2018198016A - Biological information evaluation system - Google Patents

Abstract

To provide a biological information evaluation system capable of easily and surely finding specificity of an index contained in biological information.SOLUTION: A disease risk evaluation model creation section 1 includes a biological information storage section 11, a disease risk evaluation model creation section 13, and a disease risk calculation section 14. The disease risk evaluation model creation section 13 inputs a part or all of a plurality of indexes of biological information as a feature quantity to a hierarchical type time memory algorithm being a learning device, and defines an evaluation model created by the hierarchical type time memory algorithm as a disease risk evaluation model.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a biological information evaluation system that calculates a disease risk from changes in biological information.

  Conventionally, as this kind of biological information evaluation system, as shown in Patent Document 1 below, RNA transcripts of one or more selected gene markers or their expression products in a gastrointestinal sample obtained from a patient Devices are known that determine the signs of cell proliferation by determining the expression level of.

Japanese Patent Laying-Open No. 2015-165811

  Here, in the conventional biological information evaluation system, the expression level is determined from a specific gene marker. From the biological information such as a huge amount of gene information, what kind of index is included in the biological information in the first place. There is a tremendous amount of clinical trials needed.

  Therefore, an object of the present invention is to provide a biological information evaluation system that can easily and reliably find the specificity of an index included in biological information.

The biological information evaluation system of the first invention is a biological information evaluation system for calculating a disease risk from a change in biological information,
A disease in which a plurality of the biological information having a plurality of indices are input, and a disease risk evaluation model for calculating a disease risk when the same type of biological information is input is created and stored from the plurality of input biological information A risk assessment model creation section;
When the biological information on which the disease risk is to be calculated is specified, the disease risk of the biological information is determined by the disease risk evaluation model stored in the disease risk evaluation model creation unit based on the plurality of indicators of the biological information. A disease risk calculator that calculates and outputs
The disease risk evaluation model creation unit inputs a part or all of the plurality of indices of the biological information as a feature quantity to a hierarchical time memory algorithm that is a learning device, and an evaluation model generated by the hierarchical time memory algorithm The disease risk evaluation model is used.

  According to the biological information evaluation system of the first invention, some or all of a plurality of indices of biological information are input as feature quantities to a hierarchical time memory algorithm that is a learning device, and the evaluation generated by the hierarchical time memory algorithm is generated. The model is a disease risk assessment model. Here, since the temporal change is reflected in the hierarchical time memory algorithm, it is possible to easily and reliably find a specific appearance pattern and occurrence establishment by taking into account the time change from a large amount of biological information.

  Thus, according to the biological information evaluation system of the first invention, the specificity of the index included in the biological information can be easily and reliably found.

  The disease risk includes a constitution (hormone balance), stress, personality, life span, etc. in addition to the risk of developing a specific disease.

The biological information evaluation system of the second invention is the first invention,
The biological information is a cytokine value of saliva, and the plurality of indices are a plurality of cytokine markers of the saliva,
The disease risk evaluation model creation unit is configured to input a plurality of saliva cytokine values having a plurality of cytokine markers, and to create and store a cancer risk evaluation model for calculating cancer risk from the input cytokine markers. And

  According to the biological information evaluation system of the second invention, saliva having a plurality of cytokine markers in a hierarchical time memory algorithm by using the cytokine value of saliva as biological information and using a plurality of cytokine markers of the saliva as a plurality of indices. A plurality of cytokine values are input, and a cancer risk evaluation model for calculating cancer risk can be created from the input cytokine markers.

  Thus, according to the biological information evaluation system of the second invention, the specificity of a specific cytokine marker can be actually and easily found with respect to the cytokine value of saliva.

The biological information evaluation system of the third invention is the first or second invention,
The disease risk evaluation model creation unit also inputs a part or all of the plurality of indices of the biological information as a feature amount to a learning device other than the hierarchical time memory algorithm, and the hierarchical time memory algorithm and other The disease risk assessment model is created by integrating the primary assessment models created by the learning devices.

  According to the biological information evaluation system of the third invention, part or all of a plurality of indices of biological information are also input as feature quantities to a learning device other than the hierarchical time memory algorithm, and the hierarchical time memory algorithm and other By integrating the primary evaluation models created by each learner, a disease risk evaluation model exceeding the generalization ability of each learner can be created.

  Thus, according to the biological information evaluation system of the third invention, the specificity of the index included in the biological information can be found easily and more reliably.

The system block diagram which shows the outline | summary of the biometric information evaluation system of this embodiment. The flowchart which shows the processing content of the biometric information evaluation system of FIG. Explanatory drawing which shows the processing content of the biometric information evaluation system of FIG. The system block diagram which shows the outline | summary of the biometric information management system in which the biometric information evaluation system of FIG. 1 is integrated. The flowchart which shows the processing content in the biometric information management system of FIG. The flowchart which shows the processing content in the biometric information management system of FIG. The flowchart which shows the processing content in the biometric information management system of FIG. The flowchart which shows the processing content in the biometric information management system of FIG.

  As shown in FIG. 1, the biological information evaluation system 1 of this embodiment is a system that calculates a disease risk from changes in biological information, and is configured by a computer system.

  Although detailed illustration is omitted, the computer system is, for example, a system including one computer or a plurality of computers that can communicate with each other. A computer constituting the computer system is equipped with an arithmetic processing unit including a CPU, a RAM, a ROM, an interface circuit, and the like.

  In addition, the computer system includes an input operation unit configured by a keyboard, a mouse, a touch panel, etc., an information output unit configured by a display, a printing machine, etc. A wireless communication device, and a storage device including a hard disk and the like are included.

  In addition, when the biometric information price evaluation system 1 is configured by a plurality of computers, these computers may be distributed at a plurality of remote locations.

  The input operation unit or the information output unit may include a personal computer or a mobile terminal (smart phone, tablet terminal, etc.) used by the administrator or user of the biological information price evaluation system 1 as needed.

  In the present embodiment, the biological information to be evaluated by the biological information price evaluation system 1 is, for example, a saliva cytokine value obtained by collecting saliva. The biometric information price evaluation system 1 includes a biometric information storage unit 11 and personal data as functions realized by a hardware configuration of the computer (the arithmetic processing unit) and software installed in the computer. A storage unit 12, a disease risk evaluation model creation unit 13, and a disease risk calculation unit 14 are provided.

  The biological information storage unit 11 is a means for storing and holding biological information, and stores, for example, time-series data of salivary cytokine values associated with an ID that does not specify an individual.

  The personal data storage unit 12 is a means for storing and holding information on a subject related to biometric information. For example, data such as sex information, age information, and smoking information associated with an ID that does not specify an individual is stored. Retained.

  The disease risk evaluation model creation unit 13 creates a disease risk evaluation model for calculating the disease risk from the input biological information.

  Specifically, the disease risk evaluation model creation unit 13 inputs a part or all of a plurality of indices of biological information as feature quantities to a hierarchical time memory algorithm that is a learning device, and the hierarchical time memory algorithm generates The evaluated model is used as a disease risk evaluation model.

  For example, if saliva cytokine values are used as biological information and a plurality of cytokine markers of saliva are used as a plurality of indices, a plurality of saliva cytokine values having a plurality of cytokine markers are input and input to the hierarchical time memory algorithm. A cancer risk evaluation model for calculating cancer risk can be created from the cytokine marker.

  The disease risk calculation unit 14 uses the disease risk evaluation model stored in the disease risk evaluation model creation unit 13 based on a plurality of indices of the biological information when the biological information on which the disease risk is to be calculated is designated. Then, the disease risk of the biological information is calculated and output.

  Next, with reference to FIG. 2 and FIG. 3, the content of the disease risk evaluation model creation processing by the disease risk evaluation model creation unit 13 will be described.

  As shown in FIG. 2, the disease risk evaluation model creation unit 13 first acquires biological information (STEP 11 / FIG. 2).

  For example, when the cytokine value of saliva is used as biological information, time-dependent data of 30 types of cytokine values measured from saliva is acquired.

  Next, the disease risk evaluation model creation unit 13 specifies a part or all of the biometric information acquired in STEP 11 that can be a characteristic from a plurality of indexes of the biometric information (STEP 12 / FIG. 2).

  For example, when the cytokine value of saliva is used as biological information, among the 30 types of cytokine values obtained as measurement values, 18 characteristic types as shown on the lower side of FIG. 3 are specified as cytokine markers.

  Next, the disease risk evaluation model creation unit 13 creates a state transition diagram of a pattern based on the time-series data of the index specified in STEP 12 (STEP 13 / FIG. 2).

  For example, when the salivary cytokine value is used as the biological information, as shown in the upper side of FIG. 3, as the first layer, four cytokine data are grouped into one group, and the appearance patterns of these four groups are A second layer based on the appearance probability is formed. Then, a third layer is formed based on the appearance pattern of the second layer and its probability. Only such a transition pattern is created.

  Next, the disease risk evaluation model creation unit 13 performs grouping so that those having a high transition probability are grouped as combinations of biometric information indexes to be included in the state transition diagram created in STEP 13 (STEP 14 / FIG. 2).

  For example, when the salivary cytokine value is used as biological information, a combination having a high transition probability from the time-series data of 18 kinds of cytokine markers is set as grouping.

  Next, the disease risk evaluation model creation unit 13 assigns the grouping determined in STEP 14 to the state transition diagram of STEP 13 and calculates a conditional probability (STEP 15 / FIG. 2).

  For example, when the salivary cytokine value is used as the biological information, the state transition diagram includes a group of cytokine markers determined in STEP 14 (a combination in which the transition probability increases from the time-series data of 18 types of cytokine markers). Assign to one layer. Then, the appearance pattern and the appearance probability of the grouping (combination) are calculated for the first hierarchy.

  Next, the disease risk evaluation model creation unit 13 determines whether or not the conditional probability calculation processing in STEP 15 has been completed for all nodes and hierarchies (STEP 16 / FIG. 2).

  If all the nodes and hierarchies have not been completed (NO in STEP 16 / FIG. 2), a series of processes from STEP 13 are repeatedly executed until the end of all the nodes and hierarchies, and the calculated appearance pattern is obtained. And memorize the appearance probability.

  On the other hand, when all the nodes and hierarchies have been completed (YES in STEP 16 / FIG. 2), the disease risk evaluation model creating unit 13 ends the series of processes.

  The above is the content of the process for creating a disease risk evaluation model by the disease risk evaluation model creating unit 13, and the disease risk evaluation model uses a hierarchical time memory algorithm in which changes with time are reflected, so that a large amount of biological information can be obtained. In consideration of the time change, it is possible to easily and reliably find the appearance pattern and appearance establishment of a specific index. That is, the specificity of the index included in the biological information can be found easily and reliably, and a disease risk evaluation model based on this can be created.

  In the present embodiment, the case where the hierarchical time memory algorithm is used as the learning device has been described. However, the present invention is not limited to this, and a learning device other than the hierarchical time memory algorithm may be combined. For example, a disease risk assessment model that exceeds the generalization ability of each learner may be created by integrating the primary assessment models created by the hierarchical temporal memory algorithm and deep learning, which is another learner. Good.

  Furthermore, in addition to deep learning, other learners to be combined may be combined with a random forest or an elastic net.

  Moreover, although this embodiment demonstrated the case where a disease risk evaluation model is produced using the biometric information memorize | stored in the biometric information memory | storage part 11, it is not limited to this, A personal data memory | storage part is added to this. The appearance pattern and the appearance probability may be calculated by combining data such as sex information, age information, and smoking information stored in FIG.

  The disease risk evaluation model created in this way is used, for example, in the creation of cancer risk evaluation results (STEP 232 / FIG. 7) in the biological information management system shown in FIG.

  The biometric information management system shown in FIG. 4 is a system that stores and holds the biometric information of the user, and includes a user terminal 100 provided on the user side, and a management system main body 200 connected to the user terminal 100 via a network. And a distributed ledger system 300 that is connected to the management system main body 200 over a network and stores and holds the biological information of the user.

  Here, the biological information is various information such as various physiological and anatomical information about the living body, and in addition to information specifying an individual like DNA information, an individual is not specified like heart rate. It is a concept that includes both information.

  The user terminal 100 is an information terminal such as a smartphone or tablet of a user of a salivary cancer marker test kit that is a biometric information measurement kit, and includes a network communication unit 110, a work memory 111, a program memory 112, and the like. A display input unit 13, a photographing unit 114, and a key data storage unit 115.

  The network communication unit 110 is a communication processing unit that performs connection with a wide area communication network such as the Internet.

  The work memory 111 temporarily stores and holds data for various processes in the user terminal 100.

  The program memory 112 stores the program itself in various processes in the user terminal 100. In the present embodiment, the program memory 112 includes a Web application unit 112a that stores a Web application program and an API (Application Program Interface) server unit 112b. .

  The display / input unit 113 is a display unit as an interface capable of display and touch input.

  The imaging unit 114 is an imaging unit that includes a CCD camera, a CMOS sensor, or the like.

  The key data storage unit 115 is a storage unit that stores and holds a decryption key to be described later.

  The management system main body 200 includes a network communication unit 220, a work memory 221, a program memory 222, and a storage unit 225.

  The network communication unit 220 functions as a communication processing unit that connects to the user terminal 100 with a wide-area communication network such as the Internet, and also performs a local network connection to the distributed ledger system 300. Function as.

  The work memory 221 temporarily stores and holds data for various processes in the management system main body 200.

  The program memory 222 stores the program itself in various processes in the management system main body 200. In the present embodiment, the program memory 222 includes a Web server unit 223 and an API (Application Program Interface) server unit 224.

  The web server unit 223 is an application server that performs processing according to the web application unit 112a of the user terminal 100. In this embodiment, the saliva return unit 223a, the evaluation result browsing unit 223b, and the analysis result input unit 223c. And a saliva ID generation unit 223d and an administrator function unit 223e.

  The API server unit 224 includes a saliva matching unit 224a, a biological information writing unit 224b, and an evaluation result acquisition unit 224c.

  In this embodiment, the storage unit 225 is configured as a saliva metadata storage unit that stores and holds all data related to saliva data that is biometric information.

  The distributed ledger system 300 is, for example, a block chain database. In this embodiment, the distributed ledger system 300 includes two databases DB1 and DB2 in which the same saliva data is stored, which are encrypted with different keys.

  The above is the configuration of the biological information management system of this embodiment. In the above configuration, the user terminal 100, the management system main body 200, and the distributed ledger system 300 of the biological information management system are, for example, a CPU (Central Processing Unit), a ROM (Read Only Memory), and a RAM (Random Access). The above-described control processing is performed by storing a program that executes processing by the processing units 110 to 113, 115, 220 to 225, DB1, and DB2 in a memory and executing the program. It functions as an arithmetic unit (sequencer) for executing.

  Next, processing contents of the biological information management system configured as described above will be described with reference to FIGS.

  First, the key pair generation process will be described with reference to FIG.

  In FIG. 8, by reading a QR code (registered trademark) or serial number having a function as a unique identification ID attached to the salivary cancer marker test kit in advance, the user terminal 100 can access the management system main body 200. When the user terminal 100 can use the web application with the management system 200, the key pair of the encryption key and the decryption key is generated by the user's access to the top screen (STEP 111 / FIG. 8). (STEP 112 / FIG. 8).

  Here, the key pair of the encryption key and the decryption key is associated with the unique identification ID assigned to each salivary cancer marker test kit.

  Then, it is determined whether or not a key pair has been created (STEP 113 / FIG. 8). If a key pair cannot be generated (STEP 113 / NO in FIG. 8), an error is output.

  On the other hand, when the key pair has been generated (STEP 113 / YES in FIG. 8), the encryption key is transmitted to the management system main body 200 and stored in the management system main body 200 (STEP 211 / FIG. 8), and the decryption key. Is stored in the user terminal 100 (STEP 115).

  Next, the registration process of saliva ID metadata will be described with reference to FIG.

  First, when a user collects saliva using a salivary cancer marker test kit and reads, for example, a QR code (registered trademark) of saliva ID attached to a collection tube via the user terminal 100 (STEP 121 / FIG. 6) It is determined whether the QR code (registered trademark) has been read correctly (STEP 122 / FIG. 6).

  If the QR code (registered trademark) is not read correctly (NO in STEP 122 / FIG. 6), an error is output (STEP 123 / FIG. 6).

  On the other hand, when the QR code (registered trademark) has been read correctly (YES in STEP 122 / FIG. 6), an inquiry item input screen is displayed (STEP 124 / FIG. 6), and the answered inquiry item is the main body of the management system. 200 to register the metadata of the target saliva ID (STEP 221 / FIG. 6).

  The QR code (registered trademark) of saliva ID attached to the collection tube here is basically the same as the QR code (registered trademark) or serial number having a function as a unique identification ID attached to the salivary cancer marker test kit. For example, when there are a plurality of collection tubes for one salivary cancer marker test kit, different ones may be used.

  When the unique identification ID and the saliva ID are the same, the above-described association with the encryption key and the decryption key is maintained as it is. On the other hand, when there are a plurality of saliva IDs with respect to the unique identification ID, the unique identification ID and the plurality of saliva IDs are associated in advance, and the association between the encryption key and the decryption key is performed via the unique identification ID. Also associated with saliva ID. In this case, it is preferable that the unique identification ID is appropriately replaced with the identification ID of the user terminal 100. This is because when a user purchases a salivary cancer marker test kit a plurality of times, it is necessary to associate the saliva ID of the collection tube bundled with each user with the user.

  Next, with reference to FIG. 7, registration of cytokine data of saliva ID and creation processing of a cancer risk evaluation result will be described.

  When a user previously mails a collection tube from which saliva has been collected using a salivary cancer marker test kit to a testing organization, the testing organization analyzes cytokines in the saliva.

  Here, the analysis result of the cytokine is provided to the management system main body 200, and the management system main body 200 acquires the analysis result of the cytokine specified by the saliva ID (STEP 231 / FIG. 7).

  Then, the management system main body 200 specifies an encryption key based on the unique identification ID of the acquired cytokine data of the target saliva ID, encrypts the cytokine data with the encryption key, and then registers it in the distributed ledger system 300. And stored in the distributed ledger system 300 (STEP 331 / FIG. 7).

  Furthermore, the management system body 200 creates a cancer risk evaluation result based on the acquired cytokine data of the target saliva ID (and time series data including the already acquired cytokine data, if any). The obtained cancer risk evaluation result is encrypted with an encryption key and stored in the management system body 200 (STEP 232 / FIG. 7).

  The creation of the cancer risk evaluation result is executed by the disease risk calculation unit 14 of the biological information evaluation system 1 of the present embodiment.

  Next, with reference to FIG. 8, a process for providing a cancer risk evaluation result to a user will be described.

  First, when the user operates the user terminal 100 to access the evaluation result display screen (STEP 141 / FIG. 8), the management system main body 200 uses the saliva ID (unique identification ID) based on the cytokine data of the target saliva ID. And obtained from the distributed ledger system 300 (STEP 241 / FIG. 8).

  Next, the management system main body 200 acquires the cancer risk evaluation result of the target saliva ID from the management system main body 200 based on the saliva ID (unique identification ID) (STEP 242 / FIG. 8).

  Here, the cytokine data and the cancer risk evaluation result of the target saliva ID acquired in STEP 241 and STEP 242 are in a state encrypted by the encryption key.

  Therefore, it is determined whether or not a decryption key that forms a key pair with the encryption key has been acquired by transmission from the user terminal 100 (STEP 142 / FIG. 8) (STEP 243 / FIG. 8).

  If the decryption key cannot be obtained (NO in STEP 243 / FIG. 8), an error is output to the user terminal 100 (STEP 143 / FIG. 8). If the decryption key can be obtained (YES in STEP 243). / FIG. 8), the cytokine data and the cancer risk evaluation result are decrypted with the decryption key, and then transmitted to the user terminal 100 for display (STEP 144 / FIG. 8).

  The above is the configuration and processing contents of the biological information management system of the present embodiment. According to such a biological information management system, when storing and storing cytokine data, which is biological information, from the user terminal 100 to the management system body 200 in advance. By the access, an encryption key and a decryption key are generated, the encryption key is stored in the management system main body 200, and the decryption key is stored in the user terminal 100.

  Therefore, the decryption key only exists in the user terminal 100, and the user who is the subject is not directly specified by the decryption key.

  Further, the cytokine data, which is the user's biometric information, is encrypted based on the encryption key of the management system body 200, and the encrypted cytokine data is stored and held in the distributed ledger system 300. Furthermore, since the encrypted cytokine data read from the distributed ledger system 300 is decrypted with the decryption key acquired from the user terminal 100, access to the biometric information is restricted by the encryption key and the decryption key.

  Further, the encryption key and the decryption key are linked to the unique identification ID (saliva ID) of the salivary cancer marker test kit that is the measurement kit, whereby the cytokine data that is the measurement result is expressed by the encryption key based on the unique identification ID. Encryption can be performed, and only the user terminal having the decryption key associated with the unique identification ID can access the biometric information.

  Thus, according to the biometric information management system of this embodiment, access to biometric information can be restricted while preventing a user from being specified.

  In this embodiment, the case where a salivary cancer marker test kit, which is a biometric information measurement kit, is used as the biometric information management system is described. However, the present invention is not limited to this, and other test kits (colorectal cancer test) are used. Kit) or biological information measuring equipment (such as a blood pressure monitor or activity meter) other than the test kit.

  In this case, the biometric information (data value or measurement value used as an index in the examination) and (if necessary) the evaluation result based on these are encrypted from the management key 200 and distributed to the ledger system. The biometric information and the evaluation result read from the distributed ledger system 300 are decrypted by the decryption key of the user terminal 100 and provided to the user terminal 100 from the management system main body 200 while being stored and saved in the 300.

DESCRIPTION OF SYMBOLS 1 ... Biometric information price evaluation system, 11 ... Biometric information storage part, 12 ... Personal data storage part, 13 ... Disease risk evaluation model preparation part, 14 ... Disease risk calculation part, 100 ... User terminal, 200 ... Management system main body, 300: Distributed ledger system.

Claims (3)

A biological information evaluation system for calculating disease risk from changes in biological information,
A disease in which a plurality of the biological information having a plurality of indices are input, and a disease risk evaluation model for calculating a disease risk when the same type of biological information is input is created and stored from the plurality of input biological information A risk assessment model creation section;
When the biological information on which the disease risk is to be calculated is specified, the disease risk of the biological information is determined by the disease risk evaluation model stored in the disease risk evaluation model creation unit based on the plurality of indicators of the biological information. A disease risk calculator that calculates and outputs
The disease risk evaluation model creation unit inputs a part or all of the plurality of indices of the biological information as a feature quantity to a hierarchical time memory algorithm that is a learning device, and an evaluation model generated by the hierarchical time memory algorithm A biological information evaluation system, characterized in that the disease risk evaluation model is used. The biological information evaluation system according to claim 1,
The biological information is a cytokine value of saliva, and the plurality of indices are a plurality of cytokine markers of the saliva,
The disease risk evaluation model creation unit is configured to input a plurality of saliva cytokine values having a plurality of cytokine markers, and to create and store a cancer risk evaluation model for calculating cancer risk from the input cytokine markers. A biological information evaluation system. The biological information evaluation system according to claim 1 or 2,
The disease risk evaluation model creation unit also inputs a part or all of the plurality of indices of the biological information as a feature amount to a learning device other than the hierarchical time memory algorithm, and the hierarchical time memory algorithm and other A biological information evaluation system, wherein the disease risk evaluation model is created by integrating the primary evaluation models created by the learning devices.