Information Processing Apparatus, Information Processing Method, and Information Processing Program

Abstract

To more effectively check the medical fee statement. 【Solution means】An information processing apparatus including: a first calculation unit that inputs, as input data, medical fee statement data to be submitted to a review institution into a medical treatment implementation probability estimation model generated by learning, as teacher data, review results of medical fee statement data including medical treatment act information submitted from a medical institution to the review institution, and calculates a probability of being able to perform a medical treatment act to be reviewed included in the medical fee statement data to be submitted to the review institution; and a notification unit that notifies the medical treatment act to be reviewed when the probability of being able to perform the medical treatment act to be reviewed is equal to or less than a first predetermined value.

Description

Technical Field

[0001] The present invention relates to an information processing apparatus, an information processing method, and an information processing program.

Background Art

[0002] In the above technical field, Patent Document 1 discloses a technique for improving the accuracy of checking the combination of medical treatment acts and disease names in a medical fee statement.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the technique described in the above document, as shown in FIG. 13, it only checks whether the disease name is correctly described in the medical fee statement on the premise that the medical treatment act is correct, and cannot cope with other problems in the medical fee statement.

[0005] An object of the present invention is to provide a technique for solving the above problems.

Means for Solving the Problems

[0006] To achieve the above object, the apparatus according to the present invention A first calculation unit that inputs, as input data, the medical fee statement data to be submitted to the review institution into a medical act execution probability estimation model generated by learning, as teacher data, the review results of the medical fee statement data including medical act information submitted from a medical institution to a review institution, and calculates the probability that a medical act to be reviewed included in the medical fee statement data to be submitted to the review institution can be executed. A notification unit that notifies when the probability that the medical act to be reviewed can be executed is a first predetermined value When it exceeds, it is determined whether the total number of times the patient specified by the medical fee statement data to be submitted to the review institution has performed the medical treatment subject to review exceeds the reasonable number of times, and the appropriate number of times of the medical treatment subject to review is provided. It is an information processing device.

[0007] To achieve the above object, the method according to the present invention A calculation step in which a calculation unit inputs, as input data, the medical fee statement data to be submitted to the review institution into a medical act execution probability estimation model generated by learning, as teacher data, the review results of the medical fee statement data including medical act information submitted from a medical institution to a review institution, and calculates the probability that a medical act to be reviewed included in the medical fee statement data to be submitted to the review institution can be executed. A notification step in which a notification unit notifies when the probability that the medical act to be reviewed can be executed is a first predetermined value When it exceeds, it is determined whether the number of times the patient specified by the medical fee statement data to be submitted to the review institution has performed the medical treatment subject to review exceeds the reasonable number of times, and when it exceeds the reasonable number of times, it is determined that the medical treatment subject to review is excessive is included. It is an information processing method.

[0008] To achieve the above object, the program according to the present invention A calculation step in which a calculation unit inputs, as input data, the medical fee statement data to be submitted to the review institution into a medical act execution probability estimation model generated by learning, as teacher data, the review results of the medical fee statement data including medical act information submitted from a medical institution to a review institution, and calculates the probability that a medical act to be reviewed included in the medical fee statement data to be submitted to the review institution can be executed. A notification step in which a notification unit notifies when the probability that the medical act to be reviewed can be executed is a first predetermined value When it exceeds, it is determined whether the number of times the patient specified by the medical fee statement data to be submitted to the review institution has performed the medical treatment subject to review exceeds the reasonable number of times, and when it exceeds the reasonable number of times, it is determined that the medical treatment subject to review is excessive A notification step for notifying; It is an information processing program that causes a computer to execute.

Advantages of the Invention

[0009] According to the present invention, it is possible to more effectively check the medical fee statement.

Brief Description of the Drawings

[0010]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Mode for Carrying Out the Invention

[0011] Hereinafter, with reference to the drawings, embodiments of the present invention will be exemplarily described in detail. However, the components described in the following embodiments are merely examples, and are not intended to limit the technical scope of the present invention thereto.

[0012] [First Embodiment] The information processing apparatus 100 as the first embodiment of the present invention will be described with reference to FIG. 1. The information processing apparatus 100 is an apparatus for checking a medical fee statement (so-called receipt).

[0013] As shown in FIG. 1, the information processing apparatus 100 includes a medical treatment act execution probability calculation unit 101 that calculates the probability of being able to execute a medical treatment act subject to review, and a notification unit 102 that notifies a medical treatment act subject to review with a problem.

[0014] The medical treatment act execution probability calculation unit 101 includes a medical treatment act execution probability estimation model 111 generated by learning, as teacher data, the review result 110 of medical fee statement data submitted from a medical institution to a review institution. By inputting the medical fee statement data 120 to be submitted to the review institution as input data into the medical treatment act execution probability estimation model 111, the probability 112 that the medical treatment act subject to review included in the medical fee statement data 120 is an executable medical treatment act is calculated.

[0015] The notification unit 102 notifies the medical treatment act 130 subject to review when the probability 112 of being able to execute the medical treatment act subject to review is equal to or less than a predetermined value.

[0016] According to the above configuration, effective checks can be performed on the medical treatment acts described in the medical fee statement.

[0017] [Second Embodiment] Next, before explaining the information processing apparatus according to the second embodiment of the present invention, the actual situation of the review process of the medical fee statement (hereinafter referred to as the receipt) will be explained. FIGS. 2 and 3 are diagrams showing examples of receipts. The receipt 200 in FIG. 2 is for the time of hospitalization, and the receipt 300 in FIG. 3 is for outpatient treatment. The medical fee statements 200 and 300 describe the disease names 201 and 301 and the medical treatment acts 202 and 302.

[0018] 1. Review Process and Actual Situation of Rating When the review payment agency reviews the items of the receipt and determines that they are inappropriate according to the rules of insurance medical treatment, etc., "rating" refers to not recognizing the claim itself and reducing points for the claim items. The rated content is notified to the medical institution by a "point increase / decrease notice (notification)" as shown in FIG. 4. At that time, the reasons for point increase / decrease (reasons for correction (such as amount reduction / point reduction) for each item) 401 are shown, and payment is made based on the adjusted amount. "Rating" cannot be reclaimed. On the other hand, when the review payment agency reviews and finds deficiencies or unclear points, etc., and it is difficult to correct, a "return breakdown form" as shown in FIG. 5 is used to "return" the receipt itself for content confirmation. "Return" can be reclaimed. The specific reasons for the deficiencies are described in the reason column 501 of the return breakdown form.

[0019] In the column of the reasons 401 of the point increase / decrease notice, the following four classifications are mainly described as medical reasons. A: Those not recognized as medically appropriate Regarding the indications of drugs Regarding the indications of medical treatment acts other than drugs Those where the disease name is missing or incorrect B: Those recognized as medically excessive or duplicate Regarding the excessive administration of drugs Regarding the excessive frequency and duplication of medical treatments other than drugs C: Those not recognized as appropriate for medical reasons other than A·B Regarding contraindications and off-label use D: Those where the medical treatment does not meet the calculation requirements shown in the notice For example, for those diagnosed with malignant tumors (cancer), the tumor markers are determined to be calculated using the "Treatment Management Fee for Malignant Tumor Specific Substances". If this rule is not followed, it will be a D assessment.

[0020] The review of the assessment and return is carried out by an objective review process for judging the fulfillment of the calculation requirements, in addition to a subjective review process for judging the validity of the calculation by the review committee or review members.

[0021] For example, there are the following cases where it is judged to be inappropriate, that is, an inappropriate assessment, by this subjective review for judging the validity of the calculation.

[0022] (Case 1) The relevance, necessity, and validity of the disease name and medical treatment were judged inappropriate and assessed. Claim item: Urinary sediment (microscopic examination) Result: A assessment. Despite performing the examination (claim item) to check for the presence of urinary tract infection, the medical treatment receipt did not have a disease name related to urinary tract infection. Therefore, it was judged that there was no need for the examination, and the claim for the examination was not approved and points were deducted. Consideration: Among numerous disease names, the group of disease names (recognized disease names) suitable for examining urinary tract infection recognized in the review is not specified by the review institution. This case is a difference in subjective judgment between clinicians and review members, and it is impossible to distinguish absolute right or wrong from a medical and clinical perspective and to infer the judgment recognized by the review members.

[0023] (Case 2) The frequency of medical treatment and medication was judged excessive and assessed. Claim item: CRP Result: Grade B assessment. For patients with cholecystitis, the CRP test (claim item) for evaluating the inflammatory response was performed 4 times within 1 week. The applicable disease is cholecystitis without any problem. However, based on the content of the medical fee schedule (medical treatment content), the 4 tests were judged to be excessive, and the claims for 3 out of the 4 tests were not approved and points were deducted. Consideration: Not limited to this case, the number of times medical treatment is recognized and the number of times medication is recognized are not specified by the review institution. This case is a difference in subjective judgment between clinicians and review committee members, and it is impossible to distinguish absolute right or wrong from a medical and clinical perspective, or to infer the judgment recognized by the review committee.

[0024] (Case 3) Judged inappropriate and assessed due to medical reasons. Claim item: Central venous catheter insertion method Result: Grade C assessment. For patients who underwent central venous catheter insertion, the central venous catheter insertion method (1,400 points) was calculated and claimed. However, the catheter material claimed at the same time was a peripheral indwelling catheter. Therefore, the appropriate catheter insertion method to be claimed was judged to be the peripheral indwelling central venous catheter insertion method (700 points) that matches the used material instead of the central venous catheter insertion method (1,400 points), and the difference was deducted. Consideration: In cases like this case where the medical treatment (procedure) and the used material do not match, the review institution does not specify what the material that matches the procedure or the procedure that matches the material is. This case is a difference in subjective judgment between clinicians and review committee members, and it is impossible to distinguish absolute right or wrong from a medical and clinical perspective, or to infer the judgment recognized by the review committee.

[0025] (Case 4) Assessed based on the judgment that it does not meet the calculation requirements for notification. Claim item: Intraoperative vascular imaging addition Result: D assessment. For patients who underwent intracranial hematoma removal surgery (a type of neurosurgical procedure), indocyanine green (a type of contrast agent) was used to confirm blood vessels and tumors, etc., and this addition was claimed. However, it was determined that there was no need for blood vessel delineation in terms of the content of this surgery, and the claim for this addition was not recognized, resulting in a full assessment. Consideration: Not limited to this case, in terms of the validity and necessity of medical treatment, the basis, cases, opinions, and guidelines for judging its validity are not explicitly stated by the review agency. This case is a difference in subjective judgments between clinicians and review committee members, and it is impossible to distinguish absolute right and wrong from a medical and clinical perspective, or to infer the judgment recognized by the review committee.

[0026] As described above, in the review process, when judging its validity, necessity, relevance, etc., there are no common judgment bases and criteria for all review agencies and all review committee members across the country, and it is left to the subjective judgment of the review committee or review committee members. For this reason, even for receipts with the same content, there are cases where they are recognized and cases where they are assessed by review committee members or the review agencies they are in charge of. As a result, in medical institutions, assessments (rejections) and returns occur constantly.

[0027] Therefore, in this embodiment, based on the results of the review agency, receipt data (UKE files, format 1 files, EF files, etc.), medical data, etc., are analyzed, analyzed, and learned, and receipts that may be subject to assessment and return by the review agency or review committee members are extracted, and a function is provided to recommend corrections to the description content of the receipts.

[0028] More specifically, the system according to this embodiment classifies a large number of receipts based on their attributes (including disease names), and statistically learns the medical treatment behaviors that can be performed and the number of times for each classification. Furthermore, the relationship between the medical treatment behaviors and the number of times described in the receipt and the medical treatment behaviors and the number of times that are likely to be performed as a result of learning, and the reason for the assessment return of the receipt is further learned, and the reason for the assessment return with a high probability is notified.

[0029] <Schematic Configuration of Information Processing Apparatus> FIG. 6 is a diagram showing a schematic configuration of an information processing apparatus 600 according to the present embodiment. The information processing apparatus 600 includes a medical practice execution probability calculation unit 601 that calculates the probability of being able to execute a medical practice subject to review, a proper number estimation unit 602 that estimates the proper number of times of the medical practice subject to review, a reason estimation unit 603 that estimates the reason for the return of the assessment, and a notification unit 604 that performs various notifications.

[0030] The medical practice execution probability calculation unit 601 is provided with a medical practice execution probability estimation model 611 generated by learning, as teacher data, receipt data 610, etc. and its review result data 620, and further open data 630, which are submitted from a medical institution to a review institution. The medical practice execution probability estimation model 611 inputs receipt data 640 (usually, a group of receipt files for one month) to be submitted to the review institution as input data, and calculates the probability 612 of being able to execute the medical practice subject to review included in each receipt file.

[0031] The medical practice execution probability estimation model 611 estimates the medical practices that can be performed from the disease name, for example, by a decision tree-based algorithm. The algorithm for generating the medical practice execution probability estimation model 611 in the present invention is not limited to this. The medical practice execution probability estimation model 611 is generated by learning, for example, the gender, age, disease name, medical practice, and medication history of the patient. However, the teacher data to be learned in generating the medical practice execution probability estimation model 611 is not limited to this. Any data that is an element for determining the medical practices that can be performed for the disease name, such as UKE files, EF files, format 1 files, and medical papers as past receipt data (both those that have been assessed and those that have not), may be learned to generate the model.

[0032] The medical treatment implementation probability estimation model 611 determines the implementation probability of each medical treatment for each disease diagnosis. That is, by inputting the disease diagnosis, the feasible probability can be calculated for each of multiple medical treatments. As a result, it can be determined whether each medical treatment included in the receipt data is an acceptable medical treatment. Here, those determined to be feasible with a probability of 50% or more are treated as feasible (appropriate) medical treatments. Also, the information (features) that contributed when determining feasibility are extracted simultaneously. Features are patient information described in the UKE file, style 1 file, and EF file, specifically age, gender, main disease, medications, etc. For example, when it is determined that medical treatment A can be implemented with a probability of 70%, it is determined how much of the 70% each individual feature such as feature 1 at 40%, feature 2 at 30%, feature 3 at 0%... affects (contribution degree). Then, features 613 with a contribution degree above 0% are extracted. When making improvement proposals for receipts to hospital staff using this system, the features with a high contribution degree are also notified.

[0033] Note that in the medical treatment implementation probability calculation unit 601, separately from the learning model, absolute conditions such as "To calculate medical treatment A, it is necessary to calculate any of disease diagnoses 1, 2, and 3" are filtered and mastered. And in the notification unit 604, receipts that do not meet this absolute condition are notified. Also, when constructing the medical treatment implementation probability estimation model 611, receipts that do not meet the above absolute conditions are excluded from the learning target.

[0034] When the probability 612 of implementing the medical treatment under review is less than or equal to the first predetermined value, the notification unit 604 notifies the medical treatment under review and outputs a message 650 notifying that the receipt will be audited. At that time, it also notifies that the above feature 613 is a problem as the audit reason.

[0035] To avoid an audit, the notification unit 604 may derive the disease diagnoses recognized in the past and notify a message such as "Please consider whether this disease diagnosis can be assigned to the patient."

[0036] When the probability 612 of being able to perform the medical treatment subject to review is greater than the first predetermined value, the appropriate number estimation unit 602 estimates the appropriate number of medical treatments to be performed for the patient's injury or illness using the appropriate number estimation model 621. The appropriate number estimation unit 602 inputs the review target receipt data 640 to be submitted to the review institution as input data into the appropriate number estimation model 621, and estimates the appropriate number 622 of the medical treatment subject to review. The appropriate number estimation model 621 is constructed by a decision tree-based algorithm. That is, the appropriate number estimation model 621 determines the probability for each number in the data given by learning, and sets the number with the highest probability as the appropriate number 622. At this time, the information (features) that contributed when determining the appropriate number are extracted simultaneously.

[0037] For example, when the probability that it is appropriate to perform medical treatment A 10 times is determined to be 70%, it is determined what percentage of the 70% each individual feature affects, such as feature 1 being 40%, feature 2 being 30%, feature 3 being 0%, and so on. Then, features 623 above 0% are extracted. When making a proposal for improving the receipt to hospital staff using this system, the appropriate number 622 and the features 623 with high contribution degrees are notified together.

[0038] The notification unit 604 first determines whether the medical treatment described in the review target receipt data 640 is a medical treatment for which the total number should be calculated by tracing past receipts. In the case of a medical treatment that does not need to be traced, simply compare the appropriate number 622 with the number described in the review target receipt data 640, and if there is a difference, notify with a message. Of course, it is possible to notify when there are too many, and it is also possible to notify when there are too few.

[0039] On the other hand, when it is determined that the medical treatment is one for which the total number should be calculated by tracing past receipts, the in-hospital receipt database 645 is checked, and the total number of the same medical treatment performed on the same patient in the past is calculated.

[0040] For the patient specified by the review target receipt data 640, it is determined whether the total number of times the review target medical treatment has been performed in the past plus the number of times included in the receipt data 640 exceeds a reasonable number. The notification unit 604 determines whether the number of medical treatments in the receipt data 640 is excessive using the total number of medical treatments that may be performed and the number of medical treatments already performed.

[0041] If it exceeds the reasonable number, the notification unit 604 notifies that the number of medical treatments is excessive and that it will be rated or returned, and outputs a message 650 notifying the review target medical treatment and its appropriate number. Further, a message 650 recommending that the reason for the number be described may be output.

[0042] For example, the notification unit 604 may notify a message 650 such as "There is a possibility that it may be judged as excessive or duplicate. Please comment on the reason for implementation in the receipt as a detailed description of symptoms" in order to avoid rating. Further, the notification unit 604 may notify an example (already learned) of the description of the detailed symptoms. The notification unit 604 may notify, as the message 650, the content to be corrected by learning the receipts of the same patient over several months. The notification unit 604 may notify, as the message 650, the content to be corrected by reviewing the inpatient and outpatient receipts of the same patient.

[0043] In the medical treatment implementation probability calculation unit 601, for the medical treatment described in the receipt data 640 but judged as not implementable, the cause estimation unit 603 estimates the cause using the cause estimation model 631.

[0044] The cause estimation model 631 is a machine learning model for estimating the reason for rating, which is generated by learning, as teacher data, the review results of the medical treatment fee statement data including medical treatment information submitted from a medical institution to a review institution. This cause estimation model 631 determines the probability of an individual cause step by step through a flowchart with multiple conditional statements, and finally determines the cause. The cause judged to have the highest probability is treated as the final cause.

[0045] The cause presumption unit 603 inputs, as input data, medical treatment acts that are described in the receipt data 640 but are determined by the medical treatment act execution probability calculation unit 601 as not executable, into the cause presumption model 631. Then, it calculates the probability that the assessment cause of the receipt data 640 is A (the probability that the disease name included in the receipt data 640 is incorrect).

[0046] For example, for a certain medical treatment act, if it is determined by a flowchart that the probability of cause A is 70% and the probability of cause D is 30%, it is presumed to be cause A. When it is determined to be cause A, the notification unit 604 creates a recommendation from past performance.

[0047] On the other hand, regarding those for which it is determined that the number of times in the receipt data 640 is excessive in comparison between the appropriate number of times estimated by the appropriate number of times estimation unit 602 and the actual performance value, the cause presumption unit 603 presumes that it corresponds to cause B or C. Then, it creates the text of the message 650 from the past assessment history. For example, if there is text such as "Regarding medical treatment act A, when the disease name is X, it has been observed N times" in a past assessment, it determines the similarity between the receipt that received such an assessment and the receipt to be checked, and uses the same text as the recommendation message. Regarding the similarity of receipts, it is determined based on the similarity of patient attributes and medical treatment acts.

[0048] When the probability that it is cause A (the probability that the disease name is incorrect) is equal to or greater than the second predetermined value, the notification unit 604 notifies a message indicating that the disease name should be changed.

[0049] For those for which it is determined that the probability of cause A is equal to or greater than the second predetermined value, search the past appraisal return records for review result data 620 in which the main injury is the same, other injuries are the same, medical treatment actions are the same, and the cause is A. If matching review result data 620 exists, presume cause A and report it. On the other hand, if no matching review result data 620 exists, presume cause D and report it. If there is past review result data 620 that matches the above, the notification unit 604 outputs the comment included in the appraisal return data as message 650. If it matches review result data 620 with no comment, output as message 650 that there is a past record of appraisal return. When it is presumed that it corresponds to cause B or C, compare the past record and the appropriate number of times to determine the filter & recommendation content.

[0050] For example, extract those for which the number estimated by the appropriate number of times estimation unit 602 < the number of times of execution of the receipt data 640. Next, refer to the past review result data 620 and search for review result data 620 in which the main injury is the same, other injuries are the same, the medical treatment action and the number of times are the same, and the cause is B. If no matching review result data 620 exists, it is excluded from the appraisal return target. If there is matching review result data 620, create message 650 using the comment described in the matching review result data 620 and the corrected number of times included in the review result data 620.

[0051] <Outline of processing of information processing device> FIG. 7 is a diagram showing an outline of the processing of the information processing device 600 according to the present embodiment. As described above, the data used by the information processing device 600 is electronic receipt data 610, etc., the appraisal return result 620 of the review institution, and open data 630 that serves as a material for medical judgment.

[0052] The receipt data 610, etc. includes a UKE file (electronic receipt data), an EF file (one of the DPC data (production receipt data) submitted by a hospital to the Ministry of Health, Labour and Welfare, which includes medical treatment details information and procedure details information), and a Form 1 file (one of the DPC data submitted by a hospital to the Ministry of Health, Labour and Welfare, which includes patient hospitalization information and medical record information). As shown in FIGS. 12 to 16, the EF file includes information showing patient information in more detail.

[0053] The review institution's assessment return results 620 include a point increase / decrease notice, a return breakdown statement, a point increase / decrease and return notice, an error / re-review result notice, a national health insurance error adjustment result notice, a public expense burden medical error adjustment result notice, a late-stage elderly error adjustment result notice, a public expense burden medical error adjustment result notice, etc. The open data 630 includes data such as academic papers and research presentations.

[0054] In generating the learning models 611, 621, and 631, the electronic receipt data 610, etc., the review institution's assessment return results 620, and the open data 630 that serves as the material for medical judgment are used as explanatory variables (teacher data). Then, learning models 611, 621, and 631 are generated with the "practicable medical treatment acts 701", "assessment reasons 702", and "number of times of implementation 703" extracted from the electronic receipt data 610, etc. as target variables.

[0055] For judging the validity of medical treatment acts, the input data input into the learning models 611, 621, and 631 in each medical institution is the receipt data 640 to be reviewed (before being sent to the review institution).

[0056] On the other hand, the items, contents, etc. output by the information processing device 600 are as follows. (1) In the receipt to be reviewed, the names of medical treatment acts, medications, etc. that may be subject to assessment and return (2) Reasons for assessment and return (3) Suggestions or recommendations for detailed symptoms, comments, disease names, medical treatment acts, medications, and number of times that are highly likely not to be subject to assessment and return

[0057] In the information processing apparatus 600, using the learning model 611, the patient to be analyzed is classified according to the disease name, medical history, gender, etc., and a group of medical treatment actions that can be performed and a group of medical treatment actions that cannot be performed are determined for each classified patient.

[0058] Then, for the patient included in the receipt data 640 to be reviewed, it is determined whether the medical treatment action included in the receipt data 640 can be performed.

[0059] If the medical treatment action included in the receipt data 640 is a medical treatment action that should not be approved by the review institution according to the analysis (a medical treatment action that cannot be performed), it is inferred that the disease name is inappropriate (A rating) or D rating.

[0060] Next, when it comes to a "medical treatment action that can be performed" but it corresponds to "the calculated number is not appropriate", it is inferred that it is a B rating or C rating. In other words, although the implementation itself is not a problem, for a medical treatment action with a different appropriate number, it is inferred that it is either an excessive or duplicate reason for the B / C rating.

[0061] The determination master installed in the conventional receipt checker product includes, for example, the disease names necessary for calculating the described medical treatment actions. However, the vendor does not deliver with all combinations of medical treatment actions and disease names set, but only sets and delivers a representative part of the combinations, and after introduction, the hospital staff is required to appropriately add and update them as needed. For this reason, from the hospital's perspective, they have to investigate, add, and update by themselves, which is a heavy practical burden and they cannot make full use of it. As a result, there are scattered cases where it only stays at the basic item check such as whether there is no omission in the registration of basic information. Therefore, hospital staff are doing a futile labor of printing all receipts on paper and repeating analog checks even though there is a receipt checker.

[0062] This embodiment solves the problem that it is difficult to set, update, and optimize such a master. Also, in the conventional system, the subjective ratings of the review committee that cannot be predicted or countered can also be estimated in this embodiment, significantly improving the accuracy of the master.

[0063] (Hardware Configuration) FIG. 8 is a diagram showing the hardware configuration of the information processing apparatus 600. The information processing apparatus 600 is, for example, a server provided on the Internet.

[0064] The information processing apparatus 600 is configured as a computer including hardware such as a processor 801, a ROM 802, a communication control unit 803, a RAM 804, a storage 805, and a bus connecting these. Each of these hardware components operates by electric power supplied from a power source (not shown). In the following description, the term "apparatus" can be read as a circuit, device, unit, module, etc. The hardware configuration of the information processing apparatus 600 is not limited to these, and it may include an input device (for example, keys, microphones, switches, buttons, sensors, etc.) or an output device (for example, displays, speakers, LED lamps, etc.) or both. Of course, it may include hardware such as a touch screen in which an input device and an output device are integrated.

[0065] The information processing apparatus 600 may be configured to include one or more of each of the hardware components shown in FIG. 8, or may be configured without including some of the apparatuses. Also, a plurality of apparatuses provided separately may be communicatively connected to realize the information processing apparatus 600 as a system as a whole.

[0066] The processor 801 controls the entire information processing apparatus 600 by executing, for example, an operating system and various program modules. The processor 801 may be constituted by a central processing unit (CPU: Central Processing Unit) including an interface with peripheral devices, a control device, an arithmetic device, registers, etc. Also, for example, a baseband signal processing unit and a call processing unit may be realized by the processor 801. The processor 801 may be mounted by one or more chips.

[0067] As the processor 801, in addition to a CPU (Central Processing Unit), a DSP (Digital Signal Processor) or a GPU (Graphics Processing Unit) may be used. Alternatively, instead of the processor 301, an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), an FPGA (Field Programmable Gate Array), or the like may be used.

[0068] The ROM 802, the RAM 804, and the storage 805 are all computer-readable recording media, and can store a program (program code), a software module, or the like that is executable to implement the method according to this embodiment.

[0069] The ROM 802 is a read-only memory and stores various parameters and the like in addition to the boot program that the processor 801 should execute first. Examples of the ROM 802 include an EPROM (Erasable Programmable ROM) and an EEPROM (Electrically Erasable Programmable ROM). The RAM 804 is a random access memory, has an execution area for various applications, and is also called a register, a cache, a main memory (main storage device), or the like.

[0070] The storage 805 may be composed of at least one of, for example, a flash memory, an SSD (Solid State Drive), a hard disk, an optical disk, a magneto-optical disk, a magnetic strip, or the like. The storage 805 may sometimes be called an auxiliary storage device. In the case of a smartphone, a storage medium called an eMMC (embedded Multi Media Card) or a UFS (Universal Flash Storage) may be used.

[0071] Generally, the processor 801 expands program code, software modules, data, etc. from the ROM 802, storage 805, etc. to the RAM 804 and executes various processes, but the present invention is not limited thereto.

[0072] The communication control unit 803 is hardware (a transmission / reception device) for performing communication between computers via at least one of a wired network and a wireless network, and is also referred to as, for example, a network device, a network controller, a network card, a communication module, etc.

[0073] Each of the above-mentioned hardware is connected by a bus for communicating information. The bus may be configured using a single bus, or may be configured using different buses for each device.

[0074] Each functional block 601 to 604 of the information processing apparatus 600 is realized by a corresponding program module being read into the RAM 804 and executed by the processor 801, and controlling communication by the communication control unit 803 or controlling at least one of reading and writing of data in the storage 805. The various program modules may be executed sequentially by one processor 801, or may be executed in parallel by two or more processors 801.

[0075] Here, the storage 805 stores electronic receipt data 610, etc., the audit return result 620 of the audit institution, and open data 630 that is material for medical judgment.

[0076] The hazard learning module 840 generates and updates a medical practice execution probability estimation model 611, an appropriate number estimation model 621, and a cause estimation model 631 by learning the electronic receipt data 610, etc., the audit return result 620 of the audit institution, and the open data 630.

[0077] The notification module 841 generates a message based on the estimation results using the medical treatment implementation probability estimation model 611, the appropriate number estimation model 621, and the cause estimation model 631, and notifies the user of the medical institution.

[0078] Figure 9 is a flowchart showing the flow of the process according to this embodiment. First, the learning model generation phase will be described. In step S901, the receipt data 610, the appraisal return result 620, and various medical-related data 630 are input as teacher data. Specifically, at least the UKE file, the EF file, and the format 1 file are input as explanatory variables. In step S903, using the input teacher data, a medical treatment implementation probability estimation model 611, an appropriate number estimation model 621, and a cause estimation model 631 are constructed.

[0079] Next, when moving to the learning model utilization phase, in step S911, using the generated medical treatment implementation probability estimation model 611, the implementation probability of the medical treatment included in the unreviewed receipt data is estimated. As the input data, the unreviewed receipt data is used. When inputting receipt data in a lump sum such as for one month, in step S911, the receipt data of the same patient is extracted, and from the attributes of the patient, it is determined whether each medical treatment included in each receipt data is an implementable medical treatment (S912). That is, a receipt check process for each patient is performed.

[0080] If it is implementable, the process proceeds to step S913, and using the appropriate number estimation model 621, the appropriate number of times of that medical treatment is estimated. If it is determined in step S914 that the number of times is not appropriate, the process proceeds to step S915, and using the cause estimation model 631, it is determined which of the causes A to D it corresponds to. If it is determined in step S914 that the number of times is appropriate, the process moves to the determination of the next medical treatment included in the receipt, and the processes of S911 to S915 are repeated until the determination of all the medical treatments included in the receipt to be reviewed is completed.

[0081] When the determination is completed for all medical procedures included in the receipt subject to review, the process proceeds to step S916, where the presumed reasons and recommendations for eliminating those reasons are notified. The recommendations are generated using past review result data 620 and an LLM (Large Language Model), etc.

[0082] Figure 10 is a flowchart showing the flow of the process for determining the validity of the number of medical procedures. First, in step S1001, it is determined whether the medical procedure described in the receipt data 640 subject to review is a medical procedure for which the total number should be calculated by tracing back past receipts. In the case of a medical procedure for which there is no need to trace back, the process proceeds to step S1002, where simply the valid number 622 is compared with the number described in the receipt data 640 subject to review. If it exceeds the valid number or is less than a predetermined value by a certain amount or more, the process proceeds to step S1003, and a notification is made in a message. Of course, it is possible to notify not only when it is too much but also when it is too little.

[0083] If it is determined in step S1001 that the medical procedure is one for which the total number should be calculated by tracing back past receipts, the process proceeds to step S1004, where the in-hospital receipt database 645 is examined, and in step S1005, the total number of the same medical procedure performed on the same patient in the past is calculated.

[0084] Then, the process proceeds to step S1006, where it is determined whether, for the patient specified in the receipt data 640 subject to review, adding the number included in the receipt data 640 to the total number of times the medical procedure subject to review has been performed in the past exceeds the valid number. If it exceeds the valid number, the process proceeds to step S1003, and a notification is made that the number is not valid.

[0085] Figure 11 is a diagram showing an example of the message 650 output by the notification unit 604. It includes receipt number 1101, problematic medical procedure 1102, reason 1103, probability 1104, increase / decrease points 1105, presumed detailed reason 1106, and recommendation 1107. The recommendation 1107 includes matters for consideration and their content.

[0086] Regarding the increase / decrease points 1105, the estimated detailed reasons 1106, and the recommendations 1107 in the message 650, for each reason, compare with the past appraisal return results for each reason, and if there is a match, generate the text from that case. If not, notify with fixed text according to the reason.

[0087] As described above, according to this embodiment, it is possible to effectively check the receipt.

[0088] Note that, for example, learning models may be generated separately for each gender, age, and presence or absence of hospitalization of the patient. For example, a medical practice implementation probability estimation model for men and a medical practice implementation probability estimation model for women may be generated separately. Naturally, in this case, the teacher data will also be prepared and learned separately for male data and female data. By preparing models according to the attributes of the patient in this way, more accurate estimation becomes possible.

[0089] [Other Embodiments] As described above, the present invention has been described with reference to the embodiments, but the present invention is not limited to the above embodiments. Various modifications that can be understood by those skilled in the art within the technical scope of the present invention can be made to the configuration of the present invention. Also, systems or devices in which the separate features included in each embodiment are combined in any manner are also included in the technical scope of the present invention.

[0090] Furthermore, the present invention may be applied to a system composed of a plurality of devices or to a single device. Additionally, the present invention is applicable even when an information processing program that realizes the functions of the embodiments is supplied to a system or device and executed by a built-in processor. To implement the functions of the present invention on a computer, a program installed on the computer, a medium storing the program, a server for downloading the program, and a processor for executing the program are all included in the technical scope of the present invention. In particular, at least a non-transitory computer readable medium storing a program that causes a computer to execute the processing steps included in the above-described embodiments is included in the technical scope of the present invention.

Claims

1. A medical treatment act execution probability estimation model for estimating the probability of being able to execute each medical treatment act, which is generated by learning the examination result of the medical fee statement data including medical treatment act information submitted from a medical institution to an examination institution as teacher data. As input data, the medical fee statement data to be submitted to the examination institution is input, and a first calculation unit that calculates the probability of being able to execute the medical treatment act to be examined included in the medical fee statement data to be submitted to the examination institution, When the probability of being able to execute the medical treatment act to be examined exceeds a first predetermined value, it is determined whether the number of times the patient specified by the medical fee statement data to be submitted to the examination institution has executed the medical treatment act to be examined exceeds a reasonable number of times. When it exceeds the reasonable number of times, a notification unit that notifies that the medical treatment act to be examined is excessive, An information processing apparatus comprising:

2. A reasonable number of times estimation model for estimating the probability for each number of medical treatment acts, which is generated by learning the examination result of the medical fee statement data including medical treatment act information submitted from a medical institution to an examination institution as teacher data. As input data, the medical fee statement data to be submitted to the examination institution is input, and further comprising a reasonable number of times estimation unit that estimates the reasonable number of times of the medical treatment act to be examined The information processing apparatus according to claim 1.

3. In an estimation model for estimating the probability that the disease name is incorrect, which is generated by learning the examination result of the medical fee statement data including medical treatment act information submitted from a medical institution to an examination institution as teacher data, as input data, the medical treatment act to be examined whose calculated probability of being able to execute by the first calculation unit is equal to or less than a first predetermined value is input, and further comprising a second calculation unit that calculates the probability that the disease name included in the medical fee statement data to be submitted to the examination institution is incorrect, The information processing apparatus according to claim 1, wherein the notification unit notifies a message indicating that the disease name should be changed when the probability that the disease name is incorrect is equal to or more than a second predetermined value.

4. The information processing apparatus according to claim 1, wherein when the total number of times the medical treatment act to be examined has been executed exceeds the reasonable number of times, the notification unit notifies a message recommending that the reason be described.

5. The information processing apparatus according to claim 1, wherein the medical treatment act execution probability estimation model is generated by learning the gender, age, disease name, medical treatment act, and medication history of a patient.

6. In a medical treatment implementation probability estimation model generated by learning, as teacher data, the review results of medical treatment claim statement data including medical treatment act information submitted from a medical institution to a review institution, an input step of inputting, as input data, the medical treatment claim statement data to be submitted to the review institution, and a calculation step of a calculation unit calculating the probability of being able to implement a medical treatment act to be reviewed included in the medical treatment claim statement data to be submitted to the review institution; a notification step of, when the probability of being able to implement the medical treatment act to be reviewed exceeds a first predetermined value, determining whether the number of times the patient specified by the medical treatment claim statement data to be submitted to the review institution has implemented the medical treatment act to be reviewed exceeds a reasonable number of times, and when the number of times exceeds the reasonable number of times, a notification unit notifying that the medical treatment act to be reviewed is excessive; An information processing method including the above.

7. In a medical treatment implementation probability estimation model generated by learning, as teacher data, the review results of medical treatment claim statement data including medical treatment act information submitted from a medical institution to a review institution, an input step of inputting, as input data, the medical treatment claim statement data to be submitted to the review institution, and a calculation step of calculating the probability of being able to implement a medical treatment act to be reviewed included in the medical treatment claim statement data to be submitted to the review institution; a notification step of, when the probability of being able to implement the medical treatment act to be reviewed exceeds a first predetermined value, determining whether the number of times the patient specified by the medical treatment claim statement data to be submitted to the review institution has implemented the medical treatment act to be reviewed exceeds a reasonable number of times, and when the number of times exceeds the reasonable number of times, notifying that the medical treatment act to be reviewed is excessive; An information processing program for causing a computer to execute the above.

Images