JP2006031132A - Inference system, and inference program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an inference system capable of making practical use when a DPC code and a DRG number are determined. <P>SOLUTION: The inference system is provided with a working memory 1 storing working memory elements, a knowledge base 2 storing a rule with a condition section and a conclusion section, and an inference engine 3. At least one of diagnosis data, diagnosis activity data, intermediate result data, the DPC code, and the DRG number is stored in the working memory 1 as working memory elements. At least one of the diagnosis data, the diagnosis activity data, and the intermediate result data is stored in the condition section. At least one of the diagnosis data, the diagnosis activity data, the intermediate result data, the DPC code, and the DRG number is stored in the conclusion section. The inference engine infers the DPC code and the DRG number by using the working memory elements and the rule. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

The present invention relates to an inference system, and more particularly, to an inference system that infers a DPC (Diagnostic Procedure Combination) code or a DRG (Diagnostic Related Group) number.
Here, the DPC code refers to an evaluation index for comparing the use of medical resources for each patient, which is a combination of diagnosis and medical practice by a doctor, and the DRG number is, for example, an ICD (International Classification of). According to Diseases), it is classified into MDC (Major Diagnostic Category), then classified according to the presence or absence of surgical treatment, and further classified according to the clinical process, and finally, age, name of secondary disease, complications, and discharge form (Refer to Non-Patent Document 1).

  In recent years, in the media, medical expenses have increased rapidly, the average length of hospital stay compared to Europe and the United States, undifferentiated functions, lack of quality of service, facility environment and hospital cooperation, regional disparities in medical expenses, and disparities between facilities However, it has been pointed out as a problem in Japanese hospital medicine. However, in Japan, there is no information that objectively evaluates or proves these problems. Therefore, evaluation indexes such as DPC codes and DRG numbers have been attracting attention as data that can be used for such information for several years (see Non-Patent Document 1).

  However, since there are a great many types of disease names and medical treatments, there is a considerable amount of human / physical resources for manually determining the DPC code and DRG number for each patient. Cost will be required.

  Therefore, conventionally, the following DRG number determination system has been proposed as a data processing system for determining a DRG number by forward reasoning (see Patent Document 1). That is, it is a DRG number determination system that performs processing according to each disease name and a code number assigned to a treatment / surgical name (the fifth and sixth lines of paragraph [0020] in Patent Document 1 and the fifth line of [0021]). Eyes-9th line).

  In this DRG number determination system, the inference function 80 uses the “obtained main disease name code, secondary disease name code, treatment / operation name code, patient information code, severity code, and rules stored in the rule base 82. Inference is performed to determine the DRG number ”(line 5 to line 9 of paragraph [0021] of Patent Document 1). Specifically, “the inference function 80 first extracts the corresponding rule from the rule base 82 according to the code number of the main disease name. Subsequently, the sub-path name code, the treatment / operation name code, and the The patient information code and the severity code are collated, and the DRG number is determined as a conclusion part ”(10th to 15th lines of paragraph [0021] of Patent Document 1). “Rule base 82 contains rules defined for each main disease name code number” (line 9 to line 10 of paragraph [0021] of Patent Document 1).

Edited by Shinya Matsuda, “Medical and diagnostic group classification in the 21st century: Practice and possibilities of DPC code”, Jiho, September 2003 JP 2002-132936 A

  However, as described above, in recent years, disease names, medical treatments, and the like have been further subdivided, and the types thereof have become extremely large. Along with this, the number of rules used to determine the DRG number has reached tens of thousands. Therefore, in the conventional DRG number determination system in which the main disease name, secondary disease name, treatment / operation name, patient information, and severity described in these tens of thousands of rules are all described in codes, Is extremely difficult to decipher. In addition, when the rule is displayed on the screen and the inference process is shown to the user, it becomes difficult for the user to understand the rule.

  In the conventional DRG number determination system, the rules are only extracted according to the code number of the main disease name. Therefore, in the above conventional DRG number determination system, since all instantiations (combinations of rules and main disease names, secondary disease names, etc.) involved in determining the DRG number cannot be identified, the determination of the DRG number is inaccurate. It will become something.

  In addition, the conventional DRG number determination system described above is one that determines a DRG number unilaterally without taking the user's opinion, and thus lacks flexibility and is not flexible.

  Furthermore, as described above, there are not only DRG numbers but also DPC codes as indices for evaluating long average hospital stays, regional disparities in medical expenses, inter-facility disparities, and the like compared to Europe and the United States. Therefore, since the conventional DRG number determination system can determine only the DRG number, it cannot be used at all when only the DPC code is used and the DRG number is not used.

  Therefore, the conventional DRG number determination system described above is an inadequate system that is extremely impractical and merely asserts that “the DRG number is simply determined by a computer”.

  Therefore, an object of the present invention is to provide an inference system that can be put into practical use when determining a DPC code and a DRG number.

  According to the present invention, the above problem is solved by the following means.

  A first invention is an inference system comprising a working memory in which working memory elements are stored, a knowledge base in which rules having a condition part and a conclusion part are stored, and an inference engine. , Diagnostic data, medical practice data, intermediate result data, DPC code, and DRG number are stored as the working memory element, and the condition part includes diagnostic data, medical practice data, and intermediate result data. At least one of them is stored, and at least one of diagnosis data, medical practice data, intermediate result data, DPC code, and DRG number is stored in the conclusion section, and the inference engine is stored in the working memory Working memory required to read working memory elements The reading means, the rule reading means for reading the rules stored in the knowledge base, the working memory element read by the working memory element reading means and the condition part of the rule read by the rule reading means are collated and matched A condition matching unit that stores the set of the working memory element and the rule as an instantiation in the conflict set, and an end unit that ends the process when one instantiation is not stored in the conflict set by the condition matching unit; Conflict resolution means for selecting one instantiation satisfying a predetermined condition from the plurality of instantiations stored in the competition set when a plurality of instantiations are stored in the competition set by the condition matching means; , By the condition checking means If there is only one instantiation stored in the conflict set, the conclusion part of the rule in the instantiation stored in this conflict set is stored in the working memory, while the condition matching means stores An execution means for storing a conclusion part of a rule in the instantiation selected by the conflict resolution means as a working memory element in the working memory when there are a plurality of stored instantiations, This is an inference system.

  In the second invention, the working memory element reading means reads all working memory elements stored in the working memory, and the rule reading means reads all rules stored in the knowledge base, The condition collating unit collates all working memory elements read by the working memory element reading unit and condition parts of all rules read by the rule reading unit, and sets a set of matching working memory elements and rules. The inference system according to the first invention is characterized in that it is stored in a competition set as an instantiation.

  According to a third invention, in the first invention or the second invention, the rule stored in the knowledge base includes a rule having a code as a condition part and a name as a conclusion part. It is such an inference system.

  According to a fourth aspect of the present invention, the instantiation satisfying the predetermined condition selected by the conflict resolution means is an instantiation based on data input via a keyboard, a microphone, or a touch panel. An inference system according to any one of the inventions to the third invention.

  According to a fifth aspect of the present invention, there is provided a working memory in which at least one of diagnostic data, medical practice data, intermediate result data, DPC code, and DRG number is stored as a working memory element, diagnostic data, medical practice data, and intermediate results A condition part storing at least one of the data and a rule having a conclusion part storing at least one of diagnosis data, medical practice data, intermediate result data, DPC code, and DRG number are stored In a reasoning system comprising a knowledge base and a reasoning engine, the reasoning engine reads a working memory element stored in the working memory. The reasoning engine stores in the knowledge base. Reading rules that are read The inference engine compares the working memory element read in the working memory element reading step with the condition part of the rule read in the rule reading step, and instantiates a pair of the matching working memory element and rule. A condition matching step for storing in the competition set as a condition completion step, an end step in which the inference engine terminates processing when no instantiation is stored in the competition set by the condition matching step, and When a plurality of instantiations are stored in the conflict set by the matching step, a conflict resolution step of selecting one instantiation that satisfies a predetermined condition from the plurality of instantiations stored in the conflict set; Inference If Jin has one instantiation stored in the conflict set by the condition matching step, Jin stores the conclusion part of the rule in the instantiation stored in the conflict set in the working memory, When there are a plurality of instantiations stored in the conflict set by the condition matching step, an execution step of storing a conclusion part of a rule in the instantiation selected by the conflict resolution means as a working memory element in the working memory; This is an inference program characterized by having

  In the sixth invention, the working memory element reading step reads all the working memory elements stored in the working memory, and the rule reading step reads all the rules stored in the knowledge base, In the condition matching step, all working memory elements read in the working memory element reading step are compared with the condition parts of all rules read in the rule reading step, and a set of matching working memory elements and rules is obtained. An inference program according to a fifth aspect of the invention is characterized in that the inference program is stored in a competition set as an instantiation.

  The seventh invention is characterized in that the instantiation satisfying the predetermined condition selected in the conflict resolution step is an instantiation based on data input via a keyboard, a microphone, or a touch panel. An inference program according to the invention or the sixth invention.

  According to the present invention, as will be described below, it is possible to provide an inference system that can be put into practical use that exceeds the assertion that “the DRG number is simply determined by a computer”.

  First, the present invention performs processing after converting a main disease name code, a secondary disease name code, and the like into a main disease name, a secondary disease name, and the like. Therefore, according to the present invention, most of the processing is performed not by the code but by the main disease name or the secondary disease name. Therefore, most of the rules can be described by the name such as the primary disease name or the secondary disease name instead of the code. Therefore, according to the present invention, it becomes very easy for a person to decipher the rule at the time of maintenance, and also when the rule is displayed on the screen and the inference process is shown to the user, the user can easily understand the rule. it can.

  Further, the present invention uses not only the name of the main disease but also the name of the secondary disease, the name of the treatment / operation, the patient information, the severity, etc., against the rules and uses them for generating the instantiation. Therefore, according to the present invention, since all instantiations involved in the determination of the DPC code and DRG number can be identified, the determination of the DPC code and DRG number can be made accurate.

  Moreover, in this invention, a user's opinion can be taken in. Therefore, according to the present invention, it is possible to provide an inference system that is flexible and has a little flexibility in determining a DPC code and a DRG number.

  Furthermore, the present invention can be used not only for determining a DRG number but also for determining a DPC code. Therefore, the present invention can be used in a wide range as compared with the conventional DRG number determination system that can be used only for determining the DRG number.

  The best mode for carrying out the present invention will be described below in detail with reference to the accompanying drawings.

FIG. 2 is a diagram showing an inference system according to the embodiment of the present invention.
As shown in FIG. 2, the inference system according to the embodiment of the present invention includes a working memory 1 in which working memory elements are stored, a knowledge base 2 in which rules having a condition part and a conclusion part are stored, and an inference And an engine 3.
Here, the working memory element in the embodiment of the present invention refers to diagnostic data, medical practice data, intermediate result data, DPC code, DRG number, and the like, as shown in FIG. Is stored in the working memory.
Further, as shown in FIG. 4, the rules are stored in the knowledge base in the form of “if condition part then conclusion part”, and the condition part connects one or more condition elements with “and” or “or”. It has become. In the conclusion part, one or more conclusion elements are stored. Here, the condition element refers to diagnostic data, medical practice data, intermediate result data, and the like, and the conclusion element refers to diagnostic data, medical practice data, intermediate result data, DPC code, DRG number, and the like. In addition to the names of main disease, secondary disease, treatment / operation, patient information, severity, etc., the diagnosis data includes the main disease name code, secondary disease name code, treatment / operation name code, patient information code, and severity. A code such as a code is also included, and the medical practice data includes a medical practice code in addition to the medical practice. The intermediate result data includes diagnostic data and medical practice data.
In the inference system according to the embodiment of the present invention, the inference engine 3 receives input from the user via the application program 4 or the keyboard 9, and the inference engine uses the inference result (DPC code or DRG number) as an application. Return to user via program 4 or keyboard 9. Therefore, the inference system according to the embodiment of the present invention is independent from the user via the application program 4 and the keyboard 9. Therefore, according to the inference system according to the embodiment of the present invention, it is possible to correct and add rules and to infer DPC codes and DRG numbers even while the application program 4 is being executed.
For example, an ordering system or an electronic medical record can be used as the application program 4, but it goes without saying that the embodiment of the present invention does not limit the types of application programs.

  FIG. 1 is a flowchart showing the operation of the inference system according to the embodiment of the present invention. Hereinafter, the operation of the inference system 3 according to the embodiment of the present invention will be described with reference to FIG.

First, the inference engine 3 receives a main disease name code, a secondary disease name code, a treatment / operation name code, a patient information code, and a severity code from a user through the application program 4 or the keyboard 9. These input codes are stored in the working memory 1 (step 1).
The input main disease name code, secondary disease name code, treatment / operation name code, patient information code, severity code, etc. are entered in step 7 to be described later, the main disease name, secondary disease name, treatment / operation name, patient information, After being converted into a name such as severity, it is stored in the working memory 1 as a working memory element, and thereafter handled as a name rather than a code. Therefore, most of the rules in the inference system according to the embodiment of the present invention can be written in Japanese, so that humans can easily read and maintain them. Therefore, according to the inference system according to the embodiment of the present invention, the condition part and the conclusion part in the rule can be easily added / modified / erased / confirmed. Even if it is done, it can respond quickly.

  Next, the inference engine 3 reads all the working memory elements stored in the working memory 1 (step 2).

  Next, the inference engine 3 reads all the rules stored in the knowledge base 2 (step 3).

Next, the inference engine 3 compares all the working memory elements read out in step 2 with the condition parts of all the rules read out in step 3, and competes as a set of matching working memory elements and rules as instantiations. Store in the set (step 4). However, even when a matching working memory element and rule pair exists, if this pair is an instantiation executed once in step 7 described later, this pair is not included in the competition set. Absent. This is to prevent repetition of the same process. Note that the “contention set” in the present invention refers to a logical memory in which instantiations having an attribute of contention generated on a physical memory (6, 7, or 8) are stored.
The collation in step 4 is not limited to the main disease name and main disease name code, but also the main disease name, secondary disease name, treatment / operation name, patient information, severity, etc., primary disease name code, secondary disease name code, treatment / Codes such as an operation name code, a patient information code, and a severity code are also performed. Therefore, all possibilities in inferring all DPC codes and DRG numbers can be identified. In the inference of the DPC code and the DRG number, there are cases where a plurality of unexpected results are inferred from the same data, but inference according to the embodiment of the present invention in which all working memory elements and all rules are collated. According to the system, all possible DPC codes and DRG numbers can be inferred without omission by searching all options included in the inference.
In addition, when inferring DPC codes and DRG numbers, there are multiple information regarding age, birth weight, severity information (JCS), and information related to surgery and anesthesia (first and second surgery, unilateral and bilateral, general and local anesthesia). However, according to the inference system according to the embodiment of the present invention that treats not only the main disease name and the main disease name code but all working memory elements equally, it is often necessary to make a judgment by combining these items. The partial special conditions can be evaluated equally.

  Next, when there is no instantiation in the competition set, the inference engine 3 ends the process (step 5).

Next, when a plurality of instantiations exist in the conflict set, the inference engine 3 selects one instantiation that satisfies a predetermined condition from the plurality of instantiations stored in the conflict set (step 6). Note that the present invention does not limit the conditions here, but, for example, causes the inference engine to make a selection on the condition of one of the following (1) to (3) or a combination thereof: be able to.
(1) Select an instantiation that includes the working memory element created most recently in time.
(2) An instantiation of a rule having a large number of condition elements is selected (according to this, a rule satisfying the strictest condition can be given priority).
(3) If the processing is performed in the order of (1) and (2) above but cannot be selected, it is arbitrarily selected (according to this, it is possible to avoid a situation in which selection is not possible).

  Next, when there is one instantiation stored in the conflict set, the inference engine 3 stores the conclusion part of the rule in the instantiation stored in this conflict set in the working memory 1, while When there are a plurality of instantiations stored in the set, the conclusion part of the rule in the instantiation selected in step 6 is stored in the working memory 1 as a working memory element (step 7). In the present specification, data stored in the working memory 1 from the middle of processing is referred to as intermediate result data.

  The inference system 3 repeatedly executes the above steps 2 to 7 until the DPC code and the DRG number are inferred (see FIGS. 5, 6, and 7). Then, when the DPC code or the DRG number is executed, the instantiation existing in the competition set becomes zero, and thus the process ends in step 5. In the present invention, the process can be forcibly terminated when it takes a long time to infer the DPC code or DRG number, or the process can be forcibly terminated by an instruction from the user. You can also.

When the inference engine 3 terminates the process in step 5 and the DPC code or DRG number is stored in the working memory 1, the inference program 3 uses the DPC code or DRG number as an inference result, the application program 4, the monitor 11, etc. Output to.

  In the inference system according to the embodiment of the present invention, the following steps can be added. That is, this is a step of displaying the inference process to the user at the time of execution or termination of inference. Although the present invention does not limit the method of realizing this step, for example, the inference process is displayed to the user by automatically outputting a rule chain to the screen according to a certain rule or outputting it by voice. can do. According to the inference system according to the embodiment of the present invention to which such steps are added, there is an effect that the user can accept and accept the conclusion derived by the system, and also from the aspect of debugging the system. It has the effect of helping to verify expert knowledge.

  In addition, the inference system according to the embodiment of the present invention may include a simulation step that can try various combinations such as replacement of the main disease name and the secondary disease name with respect to one patient's data. According to the inference system according to the embodiment of the present invention to which this simulation step is added, not only a fixed pattern but also various combinations such as replacement of a primary disease name and a secondary disease name are performed on one patient's data. Can be provided to the user. The present invention does not limit the method of realizing this simulation step. For example, a method of performing the selection in step 6 described above based on input data from the user via the keyboard 9 is conceivable.

As described above, according to the reasoning system according to the embodiment of the present invention, “plain expression” and “quick response to future rule changes” in that the rules are mostly expressed in daily expressions in Japanese. , Effects such as “independent of program” can be obtained.
Also, according to the inference system according to the embodiment of the present invention, all possible DPC codes can be determined without omission.
Furthermore, according to the inference system according to the embodiment of the present invention, a partial special condition can be evaluated in the same way as a main disease name or the like.
Further, according to the inference system according to the embodiment of the present invention, it is possible to provide a user with a description of a simulation or an inference process.

Note that the inference system according to the embodiment of the present invention can be realized by, for example, a computer shown in FIG. Specifically, the inference engine 3 can be realized by the CPU 5, and the working memory 1 and the knowledge base 2 can be realized by the ROM 7, the HD 8, or the like. Further, if a computer that realizes the inference system according to the embodiment of the present invention is connected to a network such as the WAN 15 or the LAN 15 via the NIC 10, the present invention is started from the clients 13 and 14 on the network as shown in FIG. The DPC code and the DRG number can be inferred using the computer 12 that implements the inference system according to the embodiment. The present invention includes a program that causes the inference system according to the embodiment of the present invention to execute the above steps. The program can be recorded on a computer-readable recording medium and distributed, but can also be distributed via a wired or wireless network 15.
In the embodiment of the present invention, the input from the user is input via the keyboard 9, but in the present invention, the input from the user is input via a touch panel, a microphone, or the like. You can also

  The present invention can be used in all fields that require a DPC code or a DRG number.

It is a flowchart which shows operation | movement of the inference system which concerns on embodiment of this invention. It is a figure which shows the inference system which concerns on embodiment of this invention. It is a figure for demonstrating the working memory in embodiment of this invention. It is a figure for demonstrating the knowledge base in embodiment of this invention. It is FIG. (1) explaining a mode that the rule is performed in embodiment of this invention. It is FIG. (2) explaining a mode that the rule is performed in embodiment of this invention. It is FIG. (3) explaining a mode that the rule is performed in embodiment of this invention. It is a figure which shows the computer which implement | achieves the inference system which concerns on embodiment of this invention. It is a figure which shows the network with which the inference system which concerns on embodiment of this invention was applied.

Explanation of symbols

1 Working Memory 2 Knowledge Base 3 Inference Engine 4 Application Program 5 CPU (Central Processing Unit)
6 RAM (Random Access Memory)
7 ROM (Read Only Memory)
8 HD (Hard Disk)
9 Keyboard 10 NIC (Network Interface Card)
DESCRIPTION OF SYMBOLS 11 Monitor 12 Computer which implement | achieves the inference system which concerns on embodiment of this invention 13 Client 14 Client 15 Networks, such as WAN (Wide Area Network) and LAN (Local Area Network)

Claims (7)

In an inference system comprising a working memory in which working memory elements are stored, a knowledge base in which rules having a condition part and a conclusion part are stored, and an inference engine,
In the working memory, at least one of diagnostic data, medical practice data, intermediate result data, DPC code, and DRG number is stored as the working memory element,
The condition part stores at least one of diagnostic data, medical practice data, and intermediate result data,
The conclusion part stores at least one of diagnostic data, medical practice data, intermediate result data, DPC code, and DRG number,
The inference engine is
Working memory element reading means for reading a working memory element stored in the working memory;
Rule reading means for reading out the rules stored in the knowledge base;
Condition collation that collates the working memory element read by the working memory element reading means and the condition part of the rule read by the rule reading means, and stores the matched working memory element and rule pair as an instantiation in the competitive set Means,
Ending means for terminating the processing when one instantiation is not stored in the competition set by the condition matching means;
Conflict resolution means for selecting one instantiation satisfying a predetermined condition from the plurality of instantiations stored in the competition set when a plurality of instantiations are stored in the competition set by the condition matching means; ,
When there is one instantiation stored in the conflict set by the condition matching means, the conclusion part of the rule in the instantiation stored in this conflict set is stored in the working memory, while the condition matching means When there are a plurality of instantiations stored in the conflict set by the execution means for storing the conclusion part of the rule in the instantiation selected by the conflict resolution means as a working memory element in the working memory;
Comprising
Inference system characterized by that. The working memory element reading means reads all working memory elements stored in the working memory,
The rule reading means reads all rules stored in the knowledge base,
The condition collating unit collates all working memory elements read by the working memory element reading unit and condition parts of all rules read by the rule reading unit, and sets a set of matching working memory elements and rules. Store it in the conflict set as an instantiation,
The inference system according to claim 1, wherein:   3. The inference system according to claim 1, wherein the rule stored in the knowledge base includes a rule having a code as a condition part and a name as a conclusion part.   The instantiation satisfying the predetermined condition selected by the conflict solution means is an instantiation based on data input via a keyboard, a microphone, or a touch panel. The inference system according to item 1. A working memory in which at least one of diagnostic data, medical practice data, intermediate result data, DPC code, and DRG number is stored as a working memory element;
A condition part storing at least one of diagnostic data, medical practice data, and intermediate result data, and at least one of diagnostic data, medical practice data, intermediate result data, DPC code, and DRG number are stored A knowledge base in which a rule having a conclusion part is stored;
An inference engine,
Inference system with
A working memory element reading step by which the inference engine reads a working memory element stored in the working memory;
A rule reading step for the inference engine to read a rule stored in the knowledge base;
The inference engine collates the working memory element read in the working memory element reading step with the condition part of the rule read in the rule reading step, and sets the matched working memory element and rule as an instantiation. A condition matching step stored in
An end step for the inference engine to terminate processing when no instantiation is stored in the conflict set by the condition matching step;
When a plurality of instantiations are stored in a conflict set by the condition matching step, the inference engine selects one instantiation satisfying a predetermined condition from the plurality of instantiations stored in the conflict set. Conflict resolution steps to
When the inference engine stores one instantiation stored in the conflict set by the condition matching step, the inference engine stores the conclusion part of the rule in the instantiation stored in the conflict set in the working memory, When there are a plurality of instantiations stored in the conflict set by the condition matching step, an execution step of storing a conclusion part of a rule in the instantiation selected by the conflict resolution means as a working memory element in the working memory When,
An inference program characterized by running The working memory element reading step reads all working memory elements stored in the working memory;
The rule reading step reads all the rules stored in the knowledge base,
In the condition matching step, all working memory elements read in the working memory element reading step are compared with the condition parts of all rules read in the rule reading step, and a set of matching working memory elements and rules is obtained. Store it in the conflict set as an instantiation,
The inference program according to claim 5, wherein: 7. The instantiation satisfying a predetermined condition selected in the conflict resolution step is an instantiation based on data input via a keyboard, a microphone, or a touch panel. Reasoning program.

Images