JPH0295332A - Health diagnosis support device - Google Patents

Abstract

PURPOSE:To improve the maintenability of a rule, to reduce a cost in a system development and to reduce the work of a doctor by providing primary and secondary health diagnosis deciding parts and a health advice deciding part, simultaneously, adapting a fuzzy inference mechanism to respective deciding parts and executing the primary and secondary decisions and health advice decision. CONSTITUTION:The primary health diagnosis deciding part to infer a disease name and the probability by a fuzzy inference mechanism and to decide a treatment method according to the degree is provided. Further, the secondary health diagnosis deciding part to infer the treatment method for each disease and the probability by the fuzzy inference mechanism and to select the treatment method with the highest probability based on secondary decision rule composed of secondary health diagnosis data, primary health diagnosis data or past data in making a re-examinee into an object is provided. A health advice deciding part to infer health advice contents according to the degree of each disease by the fuzzy inference mechanism and to comment the advice contents based on health advice rules composed of self-conscious symptom, intervewing diagnosis data, etc., is provided.

Description

[Detailed Description of the Invention] (Field of the Invention) The present invention mainly relates to a support device suitable as a support device for group health examinations, and in particular applies fuzzy inference to the 11P logic mechanism of each judgment section, and temporarily Medical examination judgment,
The present invention relates to a medical examination support device that is capable of performing secondary medical examination judgments and health guidance judgments.

(Summary of the Invention) The present invention provides a primary health checkup determination section and a health guidance determination section as a support device for health checkup, and
By applying a fuzzy inference mechanism to each judgment part and making primary judgments, secondary judgments, and health guidance judgments, we aim to reduce maintenance of rules, reduce system development costs, and reduce physician work. It is something.

(Prior art and its problems) Conventionally, for example, database systems that simply store and search health examination data have been used as support devices for group health examinations.
Alternatively, there are expert systems for diagnosis support equipped with an inference mechanism using binary logic.

However, such conventional diagnostic support systems only rely on human output and storage of data, or make inferences based on binary logic, so they are unable to support judgments, or even if they can make judgments, the rules for the judgments are enormous. In addition to problems such as the difficulty of maintaining and updating rules, conventional support systems did not have anything to do with determining health guidance.

(Purpose of the Invention) This invention was made by focusing on such conventional problems, and its purpose is to apply fuzzy inference to diagnosis support, and to use fuzzy inference to determine primary health checkup,
By performing secondary medical examination judgment and health guidance judgment,
The purpose is to solve the above problems.

(Structure and Effects of the Invention) In order to achieve the above-mentioned purpose, this invention uses a fuzzy method to identify disease names and their vagueness based on primary judgment rules consisting of primary health examination data and past data. A primary health checkup judgment unit that makes inferences using an inference mechanism and determines treatment methods according to the degree of inference, and a secondary checkup judgment unit that uses secondary health checkup data, primary health checkup data, or past data for reexamination candidates. Based on
A primary health checkup judgment unit that uses a fuzzy inference mechanism to infer the treatment method and its probability for each disease and selects the treatment method with the highest probability, and a health guidance rule consisting of subjective symptom data, interview data, etc. The present invention is characterized by comprising a health guidance determining section that infers the content of health guidance according to the degree of each disease using a fuzzy inference mechanism and comments on the content of the guidance.

As described above, the present invention is equipped with a primary health checkup determining section, a primary health checkup determining section, and a health guidance determining section as a support device, each of which is inferred by a fuzzy inference mechanism consisting of predetermined rules, and Since the inference outputs are organically combined, it is possible to easily support decisions.

Furthermore, since the device of the present invention uses a fuzzy inference mechanism and is configured to perform predetermined judgment processing and health guidance etc. by fuzzy inference, it is more knowledgeable ( (rules) changes can be easily accommodated, maintainability is significantly improved, and the device of the present invention can be implemented with fewer rules, so it is particularly useful when targeting elderly people, office workers, or local residents. When developing a new system, such as for group health checkups, the development time can be significantly shortened, so development costs can be reduced.

Furthermore, according to the device of the present invention, in addition to medical examination judgments, it is also possible to provide information necessary for health guidance at the same time, which can reduce the amount of work that doctors have to do, and to avoid data containing ambiguity that is unique to medical data. In addition to being able to deal with many situations, it also allows a large number of judgments to be made based on a certain standard, so the judgment results have the advantage of being highly reliable.

(Description of Embodiments) Next, an example in which the device of the present application is applied to a support device for occupational group health examination will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the hardware configuration of the device of the present invention. The device of the present invention has a built-in fuzzy computer, fuzzy memory, CPU, ROMSRAM, etc. - Via interface 2, FD3,
Human output devices such as a display device 4, a printer 5, and an OCR 6 are connected.

What is shown in FIG. 2 is a block diagram showing the software configuration of the device of the present invention, and the device of the present invention is mainly composed of the following functional blocks.

(1) The medical examination data input/output processing unit 10 uses an optical reader (OCR) and a floppy disk (FD).
, input and check personal information (ID) and medical examination data using the keyboard, mouse, etc., as well as display and modify the data.

(2) The knowledge acquisition support processing unit 11 sets rules for each file, and in the present device, 500 rules can be set for each file for a maximum of 10 files, and the degree of affiliation can be set for each file. It is configured so that up to 5 types of functions (membership functions) can be set for each variable, and 255 types of commonly used functions can be set.

(3) The relational database management unit 12 is composed of a knowledge base 12a and a health checkup database 12b, and the database management unit 12 centrally manages manually-generated IDs, checkup pigs, and determination results. In addition, it has a file maintenance function that allows for easy updating of health checkup data items, and it is configured so that the health checkup data of a specified year can be saved on a backup tape so that past data can be referenced.

(4) The medical examination determination support processing unit 13 has the following functions.

(a) Examine all patients to see if they have any suspicions of illness based on certain judgment rules based on primary health examination data and past data, and if there is any suspicion, use fuzzy inference to name the disease and its certainty. The mechanism performs fuzzy inference and determines the treatment method depending on the degree of inference. The final decision regarding subjects who require re-examination will be made after obtaining confirmation or judgment from an industrial physician. Those who require a reexamination will be issued with a reexamination instruction sheet that specifies the reexamination items, and those who are not eligible for a reexamination will be issued with an individual notification sheet with health guidance comments written in accordance with treatment methods.

(b) Targeting reexamination candidates, based on secondary judgment rules consisting of secondary health examination data, primary health examination data, and past data,
Inference is made in the same way as the primary judgment. At this point, final decisions regarding subjects who require initiation of treatment or detailed examinations will be made after obtaining confirmation or judgment from an industrial physician. For those who require treatment or require detailed examinations, we will issue a letter of introduction to a medical institution that clearly states the diagnosis and judgment results, and for other patients, we will issue a personal notification form with health guidance comments written on it. .

(C) - In addition to the notification slips, instruction slips, and letters of introduction that are output in paragraphs (a) and (b), reports to the Labor Standards Inspection Office, various documents for medical checkup administration and workplace managers - It is possible to output various lists for administrative processing by wise men and industrial doctors.

(5) Statistical processing unit 14, inference process explanation processing unit 15, and screen (image) control processing unit 16 (a) The statistical processing unit 14 displays statistical information based on affiliation, year ω, gender, etc.

(b) The inference process explanation processing unit 15 outputs details of the executed rules in order to facilitate validation of the set knowledge.

(C) The screen (image) control processing unit 16 has a multi-window function, so that in addition to the instruction input screen, a system abnormality occurrence log information screen, a determination processing status screen, etc. can be viewed simultaneously.

Additionally, a password is required for operation to protect data and privacy.

Therefore, in the device of the present invention, the above-mentioned input/output devices are connected to the workstation 1 via the man-machine interface, and the workstation 1
Based on the above-mentioned pig base, various treatments are applied to support medical examinations.

In addition, the device of this application is intended to support regular health checkups in in-house health checkup rooms, which are mainly made up of industrial physicians, public health nurses, and 11th office personnel, and when constructing the database, personal identification items are Data can be manually transferred from a computer to a floppy disk (FD), medical examination results and questionnaires to OCR documents, and clinical test results to a FD from a medical institution. Furthermore, the data for the past three years can be made into a database, and the data for the legal guarantee period can be stored on a cassette tape so that it can be retrieved at any time.

Furthermore, if the number of people eligible for health checkups increases, this can be easily handled by adding storage capacity devices.

Next, to explain an example of the knowledge base 12a in the device of the present invention, the knowledge base of the device of the present invention is based on the affiliation degree function, rules, range of human output, final output comments, etc. when determining the output value of each person. configured.

The membership degree function consists of a function name (fuzzy label) and a membership value, and the apparatus of the present application uses the following fuzzy labels.

(1) For clinical test items excluding qualitative data, a maximum of five types can be used for the antecedent part for each item.

TR8 (quite small) TPS (slightly small) MM (
Normal) TPB (Slightly large) TRB
(Quite large) (2) Used in the antecedent part for clinical test items of qualitative data.

MM (-) TPS (+- or less) PB (+-)
RB (+) TPB (10 or more) TRB (+10 or more) VB (+++) (3) Used in the antecedent part for symptom questioning.

LTI (Not applicable, Sometimes applicable, Applicable) (4
) Used in the consequent part for the degree of disease or intermediate hypothesis.

CLA (Typical) DEF (Deterministic) PRO (
(Probable) PO3 (Possible) SUS (Suspicious) (5) Used in the antecedent and consequent parts for image diagnosis findings, medical examination findings, and intermediate hypotheses.

YES (no) No (no)
) is shown in FIG. 3, and an example of the membership function in (4) is shown in FIG. 4.

Furthermore, the rules are configured so that a maximum of 10 and 2 fuzzy antecedent propositions and two fuzzy consequent propositions can be described, respectively, as shown below.

1F Person data item low fuzzy label THEN
Output item=fuzzy labelFurthermore, the fuzzy label representing the degree of membership function allows negation and logical sum to be set in the same proposition.

The rule file is composed of three files: primary determination, secondary determination, and health guidance determination, and which rule file is to be executed is switched by designation.

An example of the primary judgment rule in the liver function area is shown below.

(1) IF GOT=TPB THEN Abnormal liver function=DEF (2) IF GPT=TPB THEN Abnormal liver function=DEF (3) IF GGT=TPB THEN Abnormal liver function - DEF (4) IF (GOT=<GPT) =YESTHE
N Abnormal liver function = DEF (5) IF Interview (easily tiring) = LTI THEN
Abnormal liver function = PRO (6) IF GOT = MM & GPT = MM &
GGT=MM &(GOT=<GPT)=N
O& Previous GOT=TPB THEN Abnormal liver function=PRO An example of the secondary judgment rule in the area of inflammation in the body is shown below.

(1) IF CRP=MM & RA=MM
&TP=MM THEN Normality in the body=YE 5(2
)IF Internal body normal = YES & WBC = TPB
& CPK man = MM THEN Internal inflammation (progress observation) = DEF (3) IF Internal body normal = YES & WBC = T
PB & CPK female-MM THEN Internal inflammation (progress observation) = DEF (4) IF Internal body normal = YES & WBC = M
M & CPK male - MM & Examination findings of enlarged lymph nodes = YES THEN Inflammation in the body (progress observation) = PRO (5) IF Normal in the body = YES & WBC = M
M & CPK Female = MM & Examination findings of enlarged lymph nodes, Y E S T H E N Inflammation in the body (progress observation) = PRO An example of health guidance judgment rules regarding alcohol consumption is shown below.

(1) IF Drinking 4-5 days/week - YES THEN Drinking amount (reducing alcohol) = YES (2) IF Drinking every day = YES THEN Drinking amount (reducing alcohol) = YES (3) IF 2 or more glasses of alcohol / Times = YES THEN Drinking amount (reduce alcohol) = YES (4) IF (liver damage (alcoholic) = lifestyle guidance) = Y
ES THEN Judgment area 2 Lifestyle guidance = YES (5) IF (Blood pressure - Lifestyle guidance) = YES THEN
Judgment area 2 Lifestyle guidance = YES (6) IF Drinking 1-3 days/week = YES & Judgment area 2 Lifestyle guidance - YES THEN Drinking amount (reduce alcohol) = YES What is shown in Figure 5 is the inference of the present support device It is a configuration diagram showing the process, and as is clear from the diagram, clinical examination data, image diagnosis finding data, medical examination finding data, subjective symptom data, interview data, medical history data,
A primary judgment is made by inferring the disease name and its certainty from past data, etc., and based on the results of this inference, the certainty of each disease is converted using each standard, and treatment methods are determined according to the degree of certainty. . The decided measures include:
A decision is made to re-examine, continue treatment, follow-up observation, provide lifestyle guidance, leave it alone, or find no abnormality.In addition, in the case of re-examination, the industrial physician will instruct the subject to undergo a secondary medical examination.

Further, based on the above data, a secondary judgment is made by inferring the treatment method for each disease and its probability, and the treatment method with the highest probability is selected from the result of the inference. The selected treatment methods include close examination, start of treatment, continuation of treatment, follow-up observation, lifestyle guidance, neglect, and no abnormalities.If close examination and start of treatment is selected, a letter of introduction for detailed examination will be sent to the person eligible for health checkup. hand over.

Furthermore, health guidance is determined by inferring the content of health guidance according to the degree of each disease according to the subjective symptom data and interview data, and according to the results (1) of the primary and secondary determinations. The content of the guidance is then passed to the person subject to the health checkup as a guidance comment.

In addition, in the support device in Example 1, the knowledge has a hierarchical structure in which intermediate hypotheses such as judgment finding classification 1] and lifestyle guidance items are set and the final judgment is made, and it is not conversational but human-powered. Since it is necessary to drive from the data obtained, inference based on Forward Chaining fuzzy production rules is adopted. Ch
Since the ai ining method increases ambiguity as the rules are traced, this device is configured to produce results in two or three layers.

In addition, since the input is clinical examination or interview information, and the output is the degree of disease, etc., direct values or aural truth values can be used as a medium for human output.

The inference mechanism is an independent subsystem as shown in FIG. 6, uses the fact base and knowledge base as human power, and outputs the certainty of the target and the inference process.

The form of reasoning is Fuz, which is an extension of modus ponens, which is one of the 1L theories.
Using zy modus ponens, it can be expressed as:

IF A1=a1 and (A=a →B=b
) THEN B=bl Here, the fuzzy sets in the universal sets U and V are respectively a and b, and their elements are u and v.

Further, the symbol → represents an implication.

The above equation shows that for an ambiguous relationship ([A=a-+B=b] =R) between objects U and V that attend knowledge, U
When information a1 regarding V is obtained, information regarding V is derived from al and R as a conclusion.

The fuzzy set in the universal set U is fla:U-[
0, 1]", and μa (u) indicates the degree to which an element U belongs to a. The fuzzy set a is expressed as follows.

represents the conclusion of / is a separator.

In fuzzy reasoning, it is necessary to provide a method for converting the fuzzy conditional statement A=a→B=b into a fuzzy relation (R). In addition, conclusions are obtained by synthesizing factual data and fuzzy relationships.

A number of methods have been generalized for conversion and composition. This device performs inference using the following method with less divergence of solutions.

8 μR (u, v)] / (v) Here, bl is the conclusion and al is factual data.

That is, the inference section calculates as follows.

Assume that multiple propositions are set in the antecedent part of a rule.

IF A=a & B=b THEN C=
c At this time, if A=al and B=bl are given in the fact base, the conclusion C=cl is given by the following formula.

-Ca 1 (a→c) ] n [b 1
(b → c) Here, ・ is Max-Min operation,
x-+y is .mu.x (u).mu. y (v) The flow of the inference is shown in FIG. 7, but since the above-mentioned fuzzy inference mechanism and flow are conventionally known, detailed explanation thereof will be omitted.

FIG. 8 shows an example of the primary judgment rule for inferring the degree of obesity based on rules (1) and (2). In this example, the inference result is MIN (57.1.0) = 0.
．． I got 57.

That is, MIN (0,57,1,0) =0. 57
. . . Inferred output value However, in the device of the present application, it is checked to which of the conversion criteria of the relevant determination area (obesity in this case) this inferred output value belongs, and the result is determined as the final primary determination result. In the case of the above inference output value, when the conversion standard is the table shown in FIG. 9, the primary judgment result is as follows.

In addition, those shown in FIGS. 10(a) and (b) are II'
This shows an example of the secondary judgment rule for inferring the degree of the disease based on the secondary rules (1) to (6) regarding the four diseases, and its inference results, and the inference output value is 0. 6
．． Suppose we obtain 0.4 and 0.3.

The apparatus of the present application then checks whether each inference output value for the same disease area is equal to or greater than the reference value. If the reference value amount is 2, that action is taken as the final output. Furthermore, if two or more conditions are satisfied, the one with the highest priority is selected.

In the case of the inference output value shown in FIG. 10(b), when the reference value is shown in the table below, the secondary determination result is as follows.

In addition, in the above embodiment, an example in which the support device of the present application is applied to group health checkups in the workplace has been explained, but the support device of the present invention can also be applied to group health checkups for the elderly and the community, as well as other health checkups for individuals. It goes without saying that the fuzzy inference mechanism is not limited to the above embodiments, and can also be applied to diagnosis.

As is clear from the description of the embodiments above, the support device of the present invention is equipped with a primary health checkup determination section, a primary health checkup determination section, and a health guidance determination section, and each section performs inference using a fuzzy inference mechanism consisting of predetermined rules. Because the system is configured to allow easy decision support, it also improves the maintainability of rules, reduces development costs, and reduces doctor work.
This has effects such as improved reliability.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the hardware configuration of the device of the present application, FIG. 2 is a block diagram showing the software configuration, and FIG.
FIG. 4 is an explanatory diagram showing an example of a membership function in the device of the present invention, and FIG. 5 is a configuration diagram showing the inference process of the device of the present invention.
FIG. 6 is a block diagram of the inference mechanism used in the device of the present application;
Figure 7 is a flowchart showing the flow of inference, Figure 8 is an explanatory diagram showing an example of the primary judgment rule for inferring the degree of obesity based on rules (1) and (2), and Figure 9 is an illustration of the conversion criteria for obesity. An explanatory diagram showing an example, FIG. 10(a). (b) is an explanatory diagram showing inference of the degree of pancreatic disease based on the secondary rules (1) to (6) related to the disease and an example of the secondary determination rule, and an explanatory diagram showing the inference result. 1... Workstation 2... Man Red Machine Φ Interface 3, 4. 5.
6...Input/output device 0...Medical examination data input/output processing section 1...Knowledge acquisition support processing section 2...Relational database management section 3...Medical examination judgment support processing section

Claims (1)

[Claims] 1. Based on the primary judgment rule consisting of primary medical examination data and past data, the disease name and its certainty are inferred by a fuzzy inference mechanism for all patients, and treatment is taken according to the degree of inference. The primary health checkup determination section determines the method, and the secondary checkup rule for reexamination candidates is based on secondary health checkup data, primary health checkup data, or past data.
A primary health checkup judgment unit that uses a fuzzy inference mechanism to infer the treatment method and its probability for each disease and selects the treatment method with the highest probability, and a health guidance rule consisting of subjective symptom data, interview data, etc. 1. A health checkup support device comprising: a health guidance determination section that infers the content of health guidance according to the degree of each disease using a fuzzy inference mechanism and comments on the content of the guidance.