CN-115456800-B - Method for restoring disease course through insurance claim document

CN115456800BCN 115456800 BCN115456800 BCN 115456800BCN-115456800-B

Abstract

The invention relates to the technical field of medical management and provides a method for restoring disease course by insurance claim documents, which comprises the steps of cleaning and standardizing claim record data according to the insurance claim documents of insurers, taking claim records as training samples, establishing a claim document and diagnosis correlation judgment model through machine learning, taking disease groups corresponding to diagnosis names in claim documents as seed events, initializing disease course through seed events, distributing claim records without disease course allocation to the disease course with highest correlation degree in the process of initializing disease course, associating disease courses in different disease groups according to the correlation judgment model of claim documents and diagnosis, combining the disease courses in the same disease group, and iteratively executing the steps until each disease course group is unchanged and then outputting the disease course cost information of a patient. The invention can realize the cost estimation of the whole course of the patient disease and avoid the loss of insurance claims caused by the estimation deviation of the actual treatment cost.

Inventors

CHENG JIAN
Wang Tengkuan
CHEN CHANGXU
Dou zhenglei

Assignees

上海商涌科技有限公司

Dates

Publication Date: 20260512
Application Date: 20220914

Claims (3)

1. A method of restoring disease course through insurance claim documents, the method comprising: s1, cleaning and standardizing claim record data according to a participant disease insurance claim document; s2, taking the cleaned and standardized claim records as training samples, and establishing a claim document and diagnosis correlation judgment model through machine learning, wherein the method comprises the steps of taking the same invoice number and the claim document under the same diagnosis as the training samples, taking disease classification corresponding to the diagnosis information as a text subject, and taking the items related to the claim document as the contents in the text; S3, grouping diseases corresponding to diagnosis names in the claim documents as seed events, and initializing a disease course through the seed events, wherein the item and diagnosis information related to the same invoice number and diagnosis thereof are defined as a disease course, and the disease course comprises disease group information, item information and start-stop time information; s4, distributing the claim records without the course to the course with highest correlation degree in the course of initializing the course, wherein the method comprises the steps of sequentially searching the claim records without the course according to the ascending order of time, judging the relation between the claim records or the record groups and the course, processing the claim records without the course according to the relation, and updating the starting time and the duration time of the course by using the information of the claim records after the processing; The method comprises the steps of judging the relation between a claim record or a record group and a course, and processing the claim record which does not contain the course according to the relation, wherein the method comprises the steps of inquiring a course number of the claim record sharing an invoice number with the claim record or the record group, if an inquiry results, distributing the claim record to the course number, if the result is not obtained, searching by taking the claim record as a center through a set time window, judging whether a theme corresponding to the searched record is consistent with the course by utilizing a claim document and diagnosis correlation judgment model, if the theme is inconsistent with the course, distributing the record to the course number of the course, and if the theme is inconsistent with other courses, judging whether the record is consistent with the course; S5, according to the correlation judgment model of the claim document and diagnosis, the disease processes in different disease groups are associated, and the disease processes in the same disease group are combined; s6, iteratively executing the steps S4 and S5 until the disease course groups are unchanged, and outputting the disease course cost information of the patient.
2. The method of claim 1 wherein the course of the disease is restored by an insurance claim document, the method is characterized in that the step S1 comprises the following steps: sorting the claim records in time; Mapping the diagnostic data in the claim records into ICD diagnostic codes, and distributing the ICD diagnostic codes to corresponding disease groups through preset ICD diagnostic coding rules; Mapping the medicines, medical consumables and diagnosis and treatment services in the claim records into a standard medicine catalog, a consumable catalog and a diagnosis and treatment service catalog respectively.
3. The method for recovering from disease course through insurance claim documents according to claim 2, wherein the setting of the time window is to set different search time windows for different projects or disease course groups according to the theoretical latency of disease, disease course information by querying a pre-established knowledge base.

Description

Method for restoring disease course through insurance claim document Technical Field The invention relates to the technical field of medical management, in particular to a method for restoring disease course through insurance claim documents. Background The existing methods of disease grouping are mainly based on single visit disease diagnosis or treatment procedures, and representative methods thereof are DRG (disease diagnosis related group) and APG (outpatient dry on head combined outpatient case group). The disease grouping method is centered on one treatment, does not consider the development and evolution in the disease treatment process, cannot realize the cost estimation of the whole disease course of a patient when evaluating the disease risk and the cost risk of the patient, and easily causes the transfer of the treatment cost to the front/back of hospitalization and the deviation of the actual treatment cost. For example, assuming that a person is hospitalized for treatment due to appendicitis, and is hospitalized for rehabilitation again in another hospital due to wound infection after discharge, the grouping method represented by DRG and APG can only capture the individual cost and risk information of two hospitalizations respectively, and cannot prompt that the wound infection hospitalization after discharge is the sequelae of appendicitis operation, thus generating the estimated deviation of the cost of treatment of this appendicitis and the health state of the participants. In insurance claims, the prior art disease grouping method causes significant economic loss to insurance companies because of the deviation in actual treatment cost estimation calculations for the participants. Disclosure of Invention The invention mainly solves the technical problems that the development and evolution of the disease treatment process are not considered in the prior art, and the cost estimation of the whole disease course of the patient cannot be realized when the disease risk and the cost risk of the patient are evaluated. The invention provides a method for restoring disease course through insurance claim documents, which comprises the following steps: s1, cleaning and standardizing claim record data according to a participant disease insurance claim document; S2, taking the cleaned and standardized claim records as training samples, and establishing a claim bill and diagnosis correlation judgment model through machine learning; s3, grouping diseases corresponding to the diagnosis names in the claim document as seed events, and initializing a disease course through the seed events; S4, distributing the claim records without the disease course to the disease course with highest correlation degree in the process of initializing the disease course; S5, according to the correlation judgment model of the claim document and diagnosis, the disease processes in different disease groups are associated, and the disease processes in the same disease group are combined; s6, iteratively executing the steps S4 and S5 until the disease course groups are unchanged, and outputting the disease course cost information of the patient. Further, the step S1 comprises the steps of sorting the claim records according to time, mapping diagnostic data in the claim records into ICD diagnostic codes, distributing the ICD diagnostic codes to corresponding disease groups through preset ICD diagnostic coding rules, and mapping medicines, medical consumables and diagnosis and treatment services in the claim records into standard medicine catalogues, consumable catalogues and diagnosis and treatment service catalogues respectively. Further, the step S2 comprises the steps of taking the claim document under the same invoice number and the same diagnosis as a training sample, taking disease classification corresponding to diagnosis information as a text subject, and taking items related to the claim document as contents in the text, wherein the items comprise medicines, consumables and medical services, and training an implicit Dirichlet distribution model by utilizing the training sample to form a claim document and diagnosis correlation judgment model. Further, the step S3 comprises defining the item and diagnosis information associated with the same invoice number and diagnosis thereof as a disease course, wherein the disease course comprises disease group information, item information and start-stop time information. Further, the step S3 further comprises the step that the symptom class group in the disease group does not participate in the establishment of the initialization course as a seed event, wherein the symptom class group comprises symptoms, physical signs, clinical and laboratory abnormalities, and factors affecting the health state and contact with the health care institution. Further, the step S4 comprises the steps of sequentially searching the claim records which do not contain the course according to the ascending sequ