JP2018149168A - Score determination device of medical test on vital function, and program - Google Patents

Abstract

It is possible to easily determine a medical examination based on self-report. A medical examination score determination device for a biological function includes data representing a medical examination test result including a medical examination judgment score based on self-report among data relating to a first subject, A weight calculation unit that calculates a weight based on first data including at least one of data representing a physical feature, a weight calculated by the weight calculation unit, and a test for a medical examination among data related to a second subject A score determination unit that determines a determination score of the second subject based on data representing a result and second data including at least one of data representing physical characteristics. [Selection] Figure 1

Description

  The present invention relates to a medical examination score determination apparatus and program for biological functions.

  Conventionally, there is “Mini Mental State Examination” (MMSE; Mini Mental State Examination) as a method for easily determining whether or not a subject has dementia and its degree. In this MMSE, a subject is asked a question, and the answer is scored to determine whether or not the subject has dementia and its degree. However, the analysis of questions and answers in the mini-mental state examination is labor intensive and labor intensive. Therefore, in recent years, there has been a technique for calculating a dementia determination score without using a manual question and answer analysis by utilizing a diagnostic device such as an electroencephalogram measuring device, a nuclear magnetic resonance imaging device, or a CT scan device. It is disclosed (see, for example, Patent Document 1).

Special table 2010-528744

  However, in the prior art as described above, there is a problem that a diagnosis apparatus used for calculating a dementia determination score becomes large, and dementia cannot be easily determined.

  The present invention has been made in view of the above points, and provides a medical test score determination device and program for a biological function that can easily perform a medical test determination based on self-reporting.

  (1) The present invention has been made to solve the above-mentioned problems, and one aspect of the present invention includes a medical examination determination score based on self-report among data relating to a first subject, A weight calculating unit that calculates a weight based on first data including at least one of data representing an examination result of the examination and data representing physical characteristics; the weight calculated by the weight calculating unit; The determination score of the second subject is determined based on second data including at least one of data representing a medical examination result and data representing physical characteristics among data relating to two subjects. And a score determination unit for medical examination of a biological function.

  (2) In addition, according to one aspect of the present invention, in the above-described scoring device for medical examination of biological functions, the judgment score is a scoring for dementia based on self-report.

  (3) Further, according to one aspect of the present invention, in the medical examination score determination apparatus for the biological function, the weight calculation unit calculates the weight by machine learning based on the first data.

  (4) Further, according to one aspect of the present invention, in the medical test score determination device for the biological function, the data indicating the test result of the medical test is the result of the general blood test of the first subject. It is data including at least blood test information to be shown.

  (5) Further, according to one aspect of the present invention, in the medical examination score determination apparatus for the biological function, the data representing the examination result of the medical examination is the first subject by the noninvasive brain activity measurement apparatus. This data includes at least the measurement result of the brain activity.

  (6) Further, according to one embodiment of the present invention, the computer includes data on the first test subject including the determination score of the medical examination based on the self-report, and the data representing the examination result of the medical examination, and the physical characteristics A weight calculation step for calculating a weight based on first data including at least one of the data representing the data, the weight calculated by the weight calculation step, and a medical examination of the data related to the second subject A program for executing a score determination step of determining the determination score of the second subject based on data representing a result and second data including at least one of data representing physical characteristics .

  According to the present invention, it is possible to easily determine a medical examination based on self-report.

It is a figure which shows an example of a function structure of the determination system of embodiment. It is a figure which shows an example of the blood test information which the data supply part of this embodiment supplies. It is a figure which shows an example of the learning process of the determination apparatus of this embodiment. It is a figure which shows an example of the neural network of this embodiment. It is a figure which shows an example of the determination process of the determination apparatus of this embodiment. It is a figure which shows an example of the procedure which calculates the weight of this embodiment.

(Embodiment)
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a diagram illustrating an example of a functional configuration of a determination system 1 according to the present embodiment.
The determination system 1 is based on medical data based on self-report on the subject's biological function based on the body data PD representing the physical characteristics of the subject and the examination result data MT representing the examination result of the medical examination on the biological function. Determine the test score.
The medical examination based on the self-report on the biological function is, for example, a so-called “mini mental state examination” (MMSE) that is generally used for determination of dementia. The mini-mental state test consists of eleven questions with a maximum score of 30 that cover orientation, memory, computation, linguistic ability, graphic ability, etc. The score of the test subject's answer to the question of the mini mental state test is determined to be normal when the score is 24 or higher, moderately degraded when the score is less than 20, and severely degraded when the score is less than 10.

The mini mental state examination is an examination in which an examiner (such as a doctor) and a subject face each other and repeatedly ask and answer questions. Questions and answers in the mini-mental state examination take about 5 to 10 minutes.
The determination system 1 refers to the physical data PD representing the physical characteristics of the subject and the examination result data MT representing the examination result of the medical examination, so that the mini-mental can be performed without performing the tasks such as questioning and answering in person. A determination score corresponding to the state inspection score is determined.
Therefore, the determination system 1 can reduce labor and time due to the questions and answers being performed face-to-face.
Hereinafter, a functional configuration of the determination system 1 that realizes determination of a determination score without performing tasks such as questions and answers by face-to-face will be described with reference to FIG.

[Functional configuration of judgment system 1]
The determination system 1 includes a teacher data supply unit 10, a determination device 20, a determination data supply unit 30, and a presentation unit 40.
The determination device 20 calculates the weight W by a machine learning technique called deep learning. Deep learning is a machine learning technique using a multilayer neural network (a neural network having two or more hidden layers).
The determination device 20 calculates the weight W based on the teacher data supplied from the teacher data supply unit 10. The determination device 20 determines a determination score based on the determination data supplied from the determination data supply unit 30 and the calculated weight W.

The teacher data supply unit 10 supplies teacher data to the determination device 20. The teacher data is data in which a teacher determination score LSC, teacher test result data LMT, and teacher body data LPD are associated with each other. In the teacher data, the output (teacher determination score LSC) with respect to the inputs (teacher test result data LMT and teacher body data LPD) is known in advance. Therefore, the determination apparatus 20 can use the teacher data to calculate a weight W that outputs a correct result with respect to the input. The user of the determination system 1 needs to collect a predetermined number or more of data used as teacher data in advance.
The determination data supply unit 30 supplies determination data to the determination device 20. The determination data is data composed of a set of test result data MT and body data PD.
The type of data included in the teacher test result data LMT corresponds to the type of data included in the test result data MT. The type of data included in the teacher physical data LPD corresponds to the type of data included in the physical data PD. Hereinafter, the test result data MT and the body data PD may be collectively referred to as explanatory variables.
The teacher data supply unit 10 and the determination data supply unit 30 may be human interface devices such as a keyboard, a tablet, and a scanner, or may be information storage devices such as a server.

The teacher determination score LSC is a score of the answer of the subject of MMSE.
The test result data MT is information indicating a test result of a medical test of the subject. The test result data MT includes, for example, blood test information GBT and brain activity measurement result information BA.
Here, the blood test information GBT is information indicating the test result of the subject's general blood test. The general blood test is a blood test in which test items are general, for example, a blood test performed in a general health checkup. An example of the items of the general blood test will be described with reference to FIG. The brain activity measurement result information BA is the result of the measurement result of the brain activity of the subject by the noninvasive brain activity measurement device.
The body data PD is, for example, age information AG that is data indicating the age of the subject.

FIG. 2 is a diagram illustrating an example of blood test information GBT included in the test result data MT supplied by the determination data supply unit 30 of the present embodiment. Blood test information GBT includes test items for general blood tests and test values for each test item. In this example, blood test information GBT includes white blood cell count (WBC), MCV, MCH, platelet count (PLT), total protein (TP), albumin (Alb), uric acid (UA), urea nitrogen (BUN), creatinine. (Crea), sodium (So), and crawl (Cl) are included as inspection items.
In the blood test information GBT, a test item, a reference value for each test item, a test value, and a unit of the reference value and the test value are associated with each other. In the case of a specific example of a subject shown in the figure, the blood test information GBT includes a white blood cell count (WBC) of a standard value of 4000 to 9000 [cells / μL], a test value of 5000 [cells / μL], and an MCV of Information of inspection value 90 [fl]... Is included for reference values 84 to 99 [fl].

The brain activity measurement result information BA will be described.
A non-invasive brain activity measuring device (not shown) measures the brain activity of a subject by measuring the blood flow of the brain using a non-invasive detection means such as a near infrared ray. This non-invasive brain activity measuring apparatus measures, for example, measurement target items such as hemoglobin concentration (Hb), oxygenated hemoglobin concentration (HbO2), deoxygenated hemoglobin concentration (HbDO2), and oxygen saturation (SO2). In this case, the non-invasive brain activity measuring apparatus measures the above-described measurement target items for the left brain side and the right brain side, respectively. In this example, the determination data supply unit 30 includes the left brain side hemoglobin concentration L_Hb, the right brain side hemoglobin concentration R_Hb, the left brain side oxygenated hemoglobin concentration L_HbO2, the right brain side oxygenated hemoglobin concentration R_HbO2, the left brain side deoxygenated hemoglobin concentration L_HbDO2, and the right brain. The side deoxygenated hemoglobin concentration R_HbDO2, the left brain side oxygen saturation L_SO2, and the right brain side oxygen saturation R_SO2 are supplied to the determination device 20 as a part of the test result data MT.

  Returning to FIG. 1, the description of the functional configuration of the determination system 1 will be continued. The determination apparatus 20 includes a teacher determination score acquisition unit 210, a teacher examination result data acquisition unit 220, a teacher body data acquisition unit 230, a weight calculation unit 240, an examination result acquisition unit 250, a body data acquisition unit 260, A score determination unit 270.

The teacher determination score acquisition unit 210 acquires the teacher determination score LSC supplied by the teacher data supply unit 10. The teacher determination score acquisition unit 210 supplies the acquired teacher determination score LSC to the weight calculation unit 240.
The teacher examination result data acquisition unit 220 acquires the teacher examination result data LMT supplied by the teacher data supply unit 10. The teacher examination result data acquisition unit 220 supplies the acquired teacher examination result data LMT to the weight calculation unit 240.
The teacher body data acquisition unit 230 acquires the teacher body data LPD supplied from the teacher data supply unit 10. The teacher body data acquisition unit 230 supplies the acquired teacher body data LPD to the weight calculation unit 240.

  The weight calculation unit 240 includes a teacher determination score LSC acquired by the teacher determination score acquisition unit 210, teacher test result data LMT acquired by the teacher test result data acquisition unit 220, and teacher body data acquired by the teacher body data acquisition unit 230. The weight W is calculated based on the LPD. The weight calculation unit 240 outputs the calculated weight W to the score determination unit 270.

The inspection result acquisition unit 250 acquires inspection result data MT supplied from the determination data supply unit 30. The inspection result acquisition unit 250 supplies the acquired inspection result data MT to the score determination unit 270.
The body data acquisition unit 260 acquires the body data PD supplied from the determination data supply unit 30. The body data acquisition unit 260 supplies the acquired body data PD to the score determination unit 270.

  The score determination unit 270 determines a determination score based on the test result data MT acquired by the test result acquisition unit 250, the body data PD acquired by the body data acquisition unit 260, and the weight W calculated by the weight calculation unit 240. . The score determination unit 270 outputs the determination result SC of the determination score to the presentation unit 40.

The presentation unit 40 is, for example, a display or a printer, and presents the determination result SC supplied from the score determination unit 270 of the determination device 20 by a display unit such as display or printing.
The presentation unit 40 may be a storage device such as a network server. In this case, the presentation unit 40 stores the determination result SC supplied from the score determination unit 270, and supplies the stored determination result SC to another device.

[Operation of Determination Device 20]
Next, an example of the learning process of the determination apparatus 20 will be described with reference to FIG.
FIG. 3 is a diagram illustrating an example of a learning process of the determination device 20 according to the present embodiment.

(Step S <b> 10) The teacher determination score acquisition unit 210 acquires the teacher determination score LSC from the teacher data supply unit 10.
(Step S20) The teacher examination result data acquisition unit 220 acquires the teacher examination result data LMT from the teacher data supply unit 10.
(Step S30) The teacher body data acquisition unit 230 acquires the teacher body data LPD from the teacher data supply unit 10.
(Step S40) The weight calculation unit 240 calculates the weight W based on the teacher determination score LSC acquired in Step S10, the teacher examination result data LMT acquired in Step S20, and the teacher body data LPD acquired in Step S30. To do.
In deep learning, the degree of coupling between units in the network and the bias term are adjusted so that the desired prediction can be obtained. The inter-unit coupling degree is a weight W. Hereinafter, the inter-unit coupling degree and the bias term may be collectively referred to as a weight W. Also, calculating the weight W may be referred to as learning.
When using the machine learning technique, there is a concern that the prediction may deteriorate due to the use of extra explanatory variables. Here, the explanatory variables in the present embodiment are data included in the test result data MT and the body data PD, respectively. One advantage of using deep learning in the present embodiment is that the weight W is automatically selected by learning even if the explanatory variables increase.

Here, the neural network NN used in this embodiment will be described with reference to FIG.
FIG. 4 is a diagram illustrating an example of the neural network NN of the present embodiment.
The neural network NN includes an input layer IL, hidden layers HL1, HL2, HL3, and HL4, and an output layer OL. The input layer IL, the hidden layers HL1, HL2, HL3, and HL4, and the output layer OL are each composed of units arranged in a line.
The neural network NN is a neural network called a feedforward network. A feedforward network is a neural network in which information propagates in one direction from an input layer to an output layer. In this embodiment, a feedforward network in which the number of hidden layers is four and the number of units in each hidden layer is two is used.

The input layer IL is composed of units I-i (i = 1, 2,..., M: M is the number of explanatory variables). Each of the units I-i (i = 1, 2,..., M: M is the number of explanatory variables) corresponds to each explanatory variable.
The hidden layer HL1 includes a unit H1-1 and a unit H1-2. The hidden layer HL2 includes a unit H2-1 and a unit H2-2. The hidden layer HL3 includes a unit H3-1 and a unit H3-2. The hidden layer HL4 includes a unit H4-1 and a unit H4-2.
The output layer OL is composed of a unit O-1 and a unit O-2. The number of units in the output layer OL being 2 corresponds to the determination device 20 performing determination for classifying the subject into two classes. Here, the two classes are composed of a class having an MMSE answer score of 24 points or more (normal) and a class having a score of less than 24 points.
The units constituting the hidden layers HL1, HL2, HL3, and HL4 and the output layer OL are coupled to the outputs from a plurality of units constituting the layer on the input layer IL side adjacent to the respective layers constituting the unit. Multiply by degree and add bias term as input. Here, the inter-unit coupling degree (weight W) and the bias term are different for each set of the output unit and the input unit. When each unit constituting the hidden layers HL1, HL2, HL3, and HL4 and the output layer OL receives an input, a value obtained by substituting the value of the input into an activation function (described later) is output.
Details of the procedure by which the weight calculation unit 240 calculates the weight W will be described later.

Next, an example of the determination process of the determination device 20 will be described with reference to FIG.
FIG. 5 is a diagram illustrating an example of a determination process of the determination device 20 according to the present embodiment.

(Step S50) The inspection result acquisition unit 250 acquires the inspection result data MT from the determination data supply unit 30.
(Step S60) The body data acquisition unit 260 acquires the body data PD from the determination data supply unit 30.
(Step S70) The score determination unit 270 determines a determination score based on the test result data MT acquired in step S50, the body data PD acquired in step S60, and the weight W calculated in step S40 of FIG. The determination result SC indicates whether the score of the subject's MMSE answer is 24 points or more, or the score of the subject's MMSE answer is less than 24 points.
(Step S80) The score determination unit 270 outputs the determination result SC to the presentation unit 40 and ends the series of processes.

[Decision accuracy of determination device 20]
The determination accuracy of the determination device 20 will be described.
When verifying the determination accuracy of the determination apparatus 20, a sufficient number of determination data may not be available. Therefore, a part of the teacher data is extracted and used as determination data, and the remaining data is used for calculating the weight W to verify the determination accuracy. In this embodiment, a technique called Leave-One-Out cross-validation is used. The number of teacher data is N (N is a natural number).
The determination device 20 extracts one piece of data from the teacher data and uses it as determination data, and calculates the weight W using the remaining N−1 pieces of data. The determination device 20 determines a determination score based on the determination data and the calculated weight W. The determination device 20 determines whether the determination result SC is correct by comparing the determination result SC of the determination score with the teacher determination score LSC corresponding to the teacher data used as the determination data.
The determination apparatus 20 repeats this process N times until all of the teacher data is used as determination data.
According to the present embodiment, the test results of 98 subjects over 40 years old, the results of the general blood test, the measurement results of the subject's brain activity by the non-invasive brain activity measuring device, the age, and the MMSE answer score are used as teacher data. When used, it has been verified that the determination device 20 outputs a correct determination result SC for 96 pieces of determination data. That is, in the above case, it is verified that the determination accuracy of the determination device 20 is 98%.

As described above, the determination device 20 according to the present embodiment is based on the weight W calculated by the weight calculation unit 240 and the determination data including the examination result data MT and the body data PD among the data related to the subject. The determination score of the subject is determined. Here, the determination device 20 calculates the weight W based on the teacher determination score LSC, the teacher examination result data LMT, and the teacher body data LPD among the data regarding the subject.
Therefore, according to the determination apparatus 20 of the present embodiment, it is possible to reduce the labor and time that have conventionally occurred due to the questions and answers being performed face-to-face in a medical examination based on self-report.

[Calculation of weight W]
FIG. 6 is a diagram illustrating an example of a procedure for calculating the weight W according to the present embodiment.
Here, the inter-unit coupling degree and the bias term are collectively referred to as a weight W.

In the procedure for calculating the weight W according to the present embodiment, the weight calculator 240 performs learning of the neural network NN for each hidden layer as an example.
At the start of the procedure for calculating the weight W, the neural network NN includes an input layer, a first hidden layer, and an output layer. The weight calculation unit 240 calculates the weight W for the first hidden layer. Thereafter, the weight calculation unit 240 adds a second hidden layer between the first hidden layer and the output layer to the neural network NN. The weight calculation unit 240 calculates the weight W for the second hidden layer. The weight calculation unit 240 calculates the weight W for each hidden layer by repeating the above procedure. The weight calculation unit 240 ends the calculation of the weight W when the calculation of the weight W for the fourth hidden layer is completed.
Note that the weight calculation unit 240 may perform learning of the neural network NN for all hidden layers.
Hereinafter, a specific procedure for calculating the weight W will be described. Note that this calculation procedure can be executed by a computer.

(Step S400) The weight calculation unit 240 initializes the weight W and the bias B. Here, an element of weight W is an element w _ji , and an element of bias B is an element b _ji . The element w _ji and the element b _ji are used when calculating an input to a unit constituting the (j + 1) th layer adjacent to the jth layer. The weight calculation unit 240 sets a random number distributed between 0 and 1 generated from the normal distribution as the value of w _ji and sets 0 as the value of b _ji .
The weight calculation unit 240 rearranges the N teacher data at random.

(Step S401) The weight calculation unit 240 extracts B unused pieces (B is a natural number) in ascending order from the randomly rearranged teacher data. However, when there is no unused teacher data, the weight calculation unit 240 rearranges the teacher data at random again to obtain unused teacher data. The extracted B teacher data is called a mini batch.

(Step S402) The weight calculation unit 240 applies a dropout method to the neural network NN. Dropout is a technique for making overlearning less likely to occur.
For each of the input layer and the hidden layer, the weight calculation unit 240 invalidates the units constituting each layer with a certain probability p (p is a real number from 0 to 1). Here, invalidating a unit means that the output from the unit is temporarily treated as zero. In the present embodiment, as an example, the value of p is set to 0.5, but the value of p is not limited to this. The probability p may be different for each layer.
In addition, when determining the determination score, the score determination unit 270 multiplies the output of each of the input layer and the hidden layer of the neural network NN according to the invalidation probability p.

(Step S403) The weight calculation unit 240 extracts one teacher data from the extracted mini-batch. The weight calculation unit 240 inputs the teacher test result data LMT and the teacher body data LPD from the extracted teacher data, and inputs x _i (i = 1, 2,..., B: B) to the neural network NN. Is the number of teacher data contained in the mini-batch). Here, x _i is a vector having components of the number of types of data included in the teacher examination result data LMT and the teacher body data LPD.
The weight calculation unit 240 multiplies each input x _i by a weight w _ji (here j = 1) and adds a bias b _ji (here j = 1) to a certain unit in the first hidden layer. Input. The weight calculation unit 240 calculates a value obtained by substituting the input to the unit into the activation function, and sets it as the output h _k (k = 1, 2) from the unit. In the present embodiment, as an example, the Tanh function is used as the activation function. The activation function may be a function other than the Tanh function. The activation function may be, for example, a logistic function or a normalized linear function.
The weight calculation unit 240 multiplies the output h _k (k = 1, 2) from each unit of the first hidden layer by the weight w _ji (here j = 2), and the bias b _ji (here j = 2). ) Is used as an input to each unit to the second hidden layer. Similarly to the first hidden layer, the weight calculation unit 240 calculates the output from each unit of the second hidden layer by substituting the input into the activation function.
The weight calculation unit 240 repeats the above processing up to the output layer, and calculates the output y _k (k = 1, 2) from the output layer.

(Step S404) The weight calculation unit 240 calculates an objective function. The objective function is a function of the weight W, and a smaller value is given as the matching degree between the output of the neural network NN and the teacher determination score LSC included in the teacher data is higher. The higher the degree of coincidence between the output of the neural network NN and the teacher determination score LSC, the more the neural network NN outputs a more correct result with respect to the input.
Here, the objective function is given by equation (1).

In the present embodiment, the number N of teacher data in Expression (1) is equal to the number B of teacher data included in the mini-batch.
The objective function is an average of the loss function l given for each teacher data. Here, the loss function l is given by a square error between the output y and y _i obtained from the teacher data (i = 1, 2,..., N: N is the number of data of the teacher data). Here, y _i is a two-component vector. y _i is given by y _i = (1, 0) when the teacher determination score LSC is 24 points or more. On the other hand, when the teacher determination score LSC is less than 24, y _i = (0, 1) is given.
The loss function l uses a square error as an example, but is not limited thereto. The loss function l may be, for example, an entropy error.

(Step S405) The weight calculation unit 240 updates the weight W using the gradient descent method. The gradient descent method is a method of calculating a parameter so that the parameter is changed in a gradient direction in which an error (objective function) is small. Performing learning using only a part of the teacher data by using a mini-batch as in the present embodiment is called a stochastic gradient descent (SGD).
Specifically, the weight calculation unit 240 updates the weight W according to Expression (2).

(Step S406) The weight calculation unit 240 determines whether or not the objective function has converged by updating the weight W in step S405. If the weight calculation unit 240 determines that the objective function has converged (YES), the weight W calculation process ends. On the other hand, if it is determined that the objective function has not converged (NO), the process of step S401 is performed.
One of the reasons for learning using the stochastic gradient descent method is to prevent the weight W from falling into a local solution before the optimal solution where the objective function converges.

  In the present embodiment, the explanation has been given of cases where the explanatory variables are age information AG, blood test information GBT, and brain activity measurement result information BA, but the explanatory variables are not limited to this. For example, only some of age information AG, blood test information GBT, and brain activity measurement result information BA may be used as explanatory variables. Further, as the explanatory variable, only a part of the test values for each test item included in the blood test information GBT may be used, or a part of the measurement values of the measurement target item included in the brain activity measurement result information BA. You may use only.

For example, as explanatory variables, age information AG, albumin test value, uric acid test value, left brain side hemoglobin concentration L_Hb, right brain side hemoglobin concentration R_Hb, left brain side oxygen saturation L_SO2, and right brain side oxygen saturation Only R_SO2 may be used.
As another example, only age information AG, left brain side hemoglobin concentration L_Hb, right brain side hemoglobin concentration R_Hb, left brain side oxygen saturation L_SO2, and right brain side oxygen saturation R_SO2 may be used as explanatory variables. Good.

Further, as the explanatory variable, other information indicating the physical characteristics of the subject may be added to the body data PD, or other information indicating the physical characteristics of the subject may be used instead of the age information AG. Other information indicating the physical characteristics of the subject may be, for example, information indicating sex, information indicating height, information indicating weight, information indicating obesity, and the like. The information indicating the degree of obesity is, for example, a body mass index (BMI: Body Mass Index).
Further, as the explanatory variable, other information indicating the test result of the medical test of the subject other than the blood test information GBT and the brain activity measurement result information BA may be added to the test result data MT, or the blood test information Instead of the GBT and the brain activity measurement result information BA, other information indicating the test result of the medical examination of the subject may be used.

  Moreover, in this embodiment, although the case of MMSE was demonstrated as an example of the medical test | inspection based on a self-report, medical tests based on a self-report may be other than MMSE. The medical examination based on self-report may be, for example, a medical examination in which results are scored, such as a state-trait anxiety examination (STAI) or a personality examination.

In the present embodiment, the number of hidden layers of the neural network NN is four and the number of units of each hidden layer is two. However, the number of hidden layers and the number of units of each hidden layer are described here. Not exclusively.
In this embodiment, the case where the number of units in the output layer of the neural network NN is 2 has been described. However, the number of units in the output layer is not limited to this. For example, the determination device 20 may have one unit in the output layer. In that case, the determination device 20 determines the determination score according to one value of the output of the output layer. Further, the determination device 20 may classify the determination score more finely by setting the number of units of the output layer of the neural network NN to 3 or more.

[Summary]
According to the determination apparatus 20 according to the embodiment described above, based on the weight W calculated by the weight calculation unit 240 and the determination data including the test result data MT and the body data PD among the data related to the subject, Determine the test score of the subject. Here, the determination device 20 calculates the weight W based on the teacher determination score LSC, the teacher examination result data LMT, and the teacher body data LPD among the data regarding the subject. Therefore, according to the determination apparatus 20, the medical examination based on the self-report can be easily determined.

  Moreover, according to the determination apparatus 20 which concerns on embodiment mentioned above, the determination score is the same as the score of the test | inspection (MMSE) of a dementia based on a self-report. For this reason, according to the determination apparatus 20, the determinant who has determined the dementia state of the subject using the conventional MMSE can easily understand the dementia determination score.

  According to the determination apparatus 20 according to the above-described embodiment, the weight W is calculated by machine learning. For this reason, according to the determination apparatus 20, the inspector is not subject to the preconception about the meaning of the physical data representing the physical characteristics of the subject or the examination result data representing the examination result of the medical examination. The determination of the medical examination that has been performed based on the self-report of the subject can be easily performed without being based on the self-report of the subject.

  According to the determination apparatus 20 according to the above-described embodiment, the data representing the medical test result includes blood test information GBT indicating the result of the subject's general blood test. For this reason, according to the determination apparatus 20, a special test | inspection and an inquiry are unnecessary. Therefore, according to the determination apparatus 20, the determination of the medical examination which was conventionally performed based on the self-report of the subject can be easily performed without being based on the self-report of the subject.

  According to the determination apparatus 20 according to the above-described embodiment, the data representing the examination result of the medical examination includes the brain activity measurement result information BA indicating the measurement result of the brain activity of the subject by the noninvasive brain activity measurement apparatus. For this reason, according to the determination apparatus 20, a special test | inspection and an inquiry are unnecessary. Therefore, according to the determination apparatus 20, the medical examination based on the self-report can be easily determined.

  The embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and appropriate modifications may be made without departing from the spirit of the present invention. it can. You may combine the structure as described in each embodiment mentioned above.

  In addition, each part with which each apparatus in said embodiment is provided may be implement | achieved by exclusive hardware, and may be implement | achieved by memory and a microprocessor.

  Each unit included in each device is configured by a memory and a CPU (central processing unit), and a program for realizing the function of each unit included in each device is loaded into the memory and executed to realize the function. It may be.

In addition, a program for realizing the function of each unit included in each device is recorded on a computer-readable recording medium, and the program recorded on the recording medium is read into a computer system and executed, whereby the control unit You may perform the process by each part with which it is equipped.
Here, the “computer system” includes an OS and hardware such as peripheral devices.

Further, the “computer system” includes a homepage providing environment (or display environment) if a WWW system is used.
The “computer-readable recording medium” refers to a storage device such as a flexible medium, a magneto-optical disk, a portable medium such as a ROM and a CD-ROM, and a hard disk incorporated in a computer system. Furthermore, the “computer-readable recording medium” dynamically holds a program for a short time like a communication line when transmitting a program via a network such as the Internet or a communication line such as a telephone line. In this case, a volatile memory in a computer system serving as a server or a client in that case, and a program that holds a program for a certain period of time are also included. The program may be a program for realizing a part of the functions described above, and may be a program capable of realizing the functions described above in combination with a program already recorded in a computer system.

  DESCRIPTION OF SYMBOLS 1 ... Determination system, 10 ... Teacher data supply part, 20 ... Determination apparatus, 30 ... Determination data supply part, 40 ... Presentation part, 210 ... Teacher determination score acquisition part, 220 ... Teacher test result data acquisition part, 230 ... Teacher body Data acquisition unit, 240 ... weight calculation unit, 250 ... test result acquisition unit, 260 ... body data acquisition unit, 270 ... score determination unit

Claims (6)

Based on the first data including at least one of data representing a medical examination test result and physical characteristics, including a medical examination determination score based on self-report among data relating to the first subject A weight calculation unit for calculating weights,
Based on the weight calculated by the weight calculator, and second data including at least one of data representing a medical examination result and data representing a physical feature among data relating to the second subject. A score determination unit for determining the determination score of the second subject;
A medical examination score determination device for a biological function.   2. The medical test score judging device according to claim 1, wherein the judgment score is a score of a test for dementia based on self-report.   The score determination device for a medical examination for a biological function according to claim 1, wherein the weight calculation unit calculates the weight by machine learning based on the first data.   The biological function according to any one of claims 1 to 3, wherein the data indicating the test result of the medical test is data including at least blood test information indicating a result of the general blood test of the first subject. A medical test scoring system.   The data representing the test result of the medical examination is data including at least a measurement result of the brain activity of the first subject by a noninvasive brain activity measurement device. A medical test scoring system for biological functions. On the computer,
Based on the first data including at least one of data representing a medical examination test result and physical characteristics, including a medical examination determination score based on self-report among data relating to the first subject A weight calculating step for calculating the weight by
Based on the weight calculated in the weight calculating step, and second data including at least one of data representing a medical examination result and data representing physical characteristics among data relating to the second subject. And a score determination step of determining the determination score of the second subject.

Images