Medical insurance fund illegal use mining method based on big data analysis

Abstract

The invention discloses a big data analysis-based medical insurance fund illegal use mining method. The method comprises the following steps of: establishing a unified standard data establishing the unified standard database according to a medical insurance service coding standard and a data standard; establishing a violation risk model knowledge base including a doctor-seeing behavior violation risk model, a medical project violation risk model and an expense settlement violation risk model; data importing: importing data such as medical insurance settlement data, fixed-point medical institution medicine and high-value consumable purchase-sale-stock data, a diagnosis and treatment service facility directory and diagnosis and treatment project income and expenditure into a system, and then verifying the data; big data analysis: screening illegal risk points through big data analysis according to the illegal risk model knowledge base; according to the method, data support is provided for regional medical insurance fund investigation and evidence collection for medical insurance fund illegal project screening scenes, and the industrial problem that medical insurance fund illegal use troubleshooting and evidence collection are difficult is solved.

Description

Technical Field

[0001] This invention relates to the field of big data analytics, and in particular to a method for uncovering the misuse of medical insurance funds based on big data analytics. Background Technology

[0002] The medical insurance fund is the people's "money for medical treatment" and "lifesaving money," and the Party Central Committee and the State Council attach great importance to its security. Based on a unified national medical insurance information platform and two databases, we will conduct intelligent auditing and monitoring of the medical insurance fund, and utilize big data screening and analysis to promote proactive supervision, continuously improve the professionalism and intelligence of fund supervision, and safeguard the fund's security. We must comprehensively apply intelligent monitoring and big data analysis methods for medical insurance to combine online and offline approaches in all aspects of supervision, including clue discovery, investigation and evidence collection, handling of violations, and application of results, forming a closed-loop supervision system throughout the entire process.

[0003] Traditional methods primarily target individual designated medical institutions, providing real-time alerts for medical service activities and conducting preliminary reviews and warnings before medical insurance fund settlement. However, these methods suffer from limitations due to their limited data scope, making it difficult to effectively identify suspected cross-regional, cross-institutional, or group-based activities. Summary of the Invention

[0004] To address the problems existing in the prior art, the purpose of this invention is to provide a method for mining illegal use of medical insurance funds based on big data analysis. This invention targets the screening scenario of illegal and irregular medical insurance fund projects, provides data support for regional medical insurance fund investigation and evidence collection, and solves the industry problems of difficulty in investigating and obtaining evidence of illegal and irregular use of medical insurance funds.

[0005] To achieve the above objectives, the technical solution adopted by this invention is: a method for detecting the misuse of medical insurance funds based on big data analysis, comprising the following steps:

[0006] Step 1: Establish a unified and standardized database: Establish a unified and standardized database based on the medical insurance business coding standards and data specifications;

[0007] Step 2: Establish a knowledge base for violation risk models, including violation risk models for medical treatment behavior, medical project violations, and fee settlement violations.

[0008] Step 3: Data Import: Import medical insurance settlement data, inventory data of drugs and high-value consumables of designated medical institutions, catalog of medical service facilities, income and expenditure of medical services, etc. into the system, and then verify the data;

[0009] Step 4: Big Data Analysis: Based on the knowledge base of the violation risk model, use big data analysis to screen out illegal and non-compliant risk points.

[0010] As a further improvement of the present invention, in step 2, the medical treatment behavior violation risk model is specifically used to identify: ① Spatiotemporal clustering: abnormal behavior of the same insured person seeking medical treatment multiple times in a short period of time or settling accounts at the same medical institution in a centralized manner; ② Spatiotemporal travel: spatiotemporal travel behavior of the insured person settling accounts at medical institutions at a distance in a short period of time; ③ High frequency and low amount: abnormal transaction behavior of frequent small-amount settlements; ④ False hospitalization identification: identifying false hospitalization behavior by comparing bed fees and nursing fees.

[0011] As a further improvement of the present invention, in step 2, the medical project violation risk model is specifically used to identify: ① duplicate charges: multiple charges for the same project; ② charges exceeding the standard: comparing the price standard with the settlement amount to identify charges exceeding the standard; ③ project overcharging: analyzing the overcharging behavior of surgical operations and costs that do not match under the DRG / DIP payment model; ④ consumable abuse: comparing the usage of consumables for the same project in similar medical institutions to identify consumable abuse.

[0012] As a further improvement of the present invention, in step 2, the expense settlement violation risk model is specifically used to identify: ① abnormal fund disbursement: the amount of violation is identified by the difference between the fund disbursement ratio and the policy reimbursement ratio; ② abuse of personal accounts: abnormal settlement behavior of insured persons purchasing more drugs than permitted.

[0013] As a further improvement of the present invention, in step 3, the data verification specifically includes: checking the data integrity, the uniqueness of the patient identification number, and the correlation between the settlement table and the detailed table.

[0014] As a further improvement of the present invention, step 4 specifically includes the following steps:

[0015] Step 4.1, Association Rule Mining: Use the FP-Growth algorithm to screen for abnormal behavior in project portfolios;

[0016] Step 4.2, Time Series Analysis: LSTM analysis is used to analyze the time series characteristics of medical insurance settlement records to identify high-frequency, high-amount abnormal settlement behaviors;

[0017] Step 4.3, Cluster Analysis: After clustering the DRG / DIP grouped data, identify the high-risk behavior of the project.

[0018] This invention first establishes a unified and standardized database according to the nationally unified medical insurance business coding standards and data specifications. Then, it imports medical insurance settlement data, inventory data of drugs and high-value consumables of designated medical institutions, catalogs of medical service facilities, and income and expenditure data of medical services into the system and performs data validity checks to achieve the fusion of multi-source heterogeneous medical data. Next, it establishes a knowledge base for violation risk models based on the intelligent supervision "two databases" rules and knowledge points publicly released by the National Healthcare Security Administration. Finally, through big data analysis, the system automatically judges whether medical insurance reimbursement behavior is compliant, thereby preventing losses to the medical insurance fund.

[0019] The beneficial effects of this invention are:

[0020] This invention provides regulatory authorities with intuitive data support, facilitating an understanding of the overall violation situation, assessing regulatory effectiveness, and providing a scientific basis for formulating and adjusting regulatory strategies. Furthermore, in-depth analysis of violation data can uncover potential risk points, providing direction for regulatory work. Attached Figure Description

[0021] Figure 1 This is a flowchart of an embodiment of the present invention. Detailed Implementation

[0022] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0023] Example 1

[0024] like Figure 1 As shown, a method for detecting misuse of medical insurance funds based on big data analysis includes:

[0025] Step 1: Establish a unified and standardized database, in accordance with the national medical insurance information business coding standards:

[0026]

[0027]

[0028] Step Two: Establish a knowledge base for the violation risk model:

[0029] 1. Medical Treatment Behavior Violation Risk Model:

[0030] Spatiotemporal clustering: Abnormal behavior of the same insured person seeking medical treatment multiple times within a short period of time or settling accounts at the same medical institution;

[0031] Time-space travel: The "time-space travel" behavior of insured persons settling accounts at remote medical institutions within a short period of time;

[0032] High frequency, low amount: Abnormal transaction behavior involving frequent small-amount settlements;

[0033] Fraudulent hospitalization identification: By comparing bed fees and nursing fees, fraudulent hospitalization behavior can be identified.

[0034] 2. Medical Project Violation Risk Model:

[0035] Double charging: Charging for the same service multiple times;

[0036] Overcharging: Identify overcharging by comparing the price standard with the settlement amount;

[0037] Project overcharging: Analyzing the "overcharging" behavior of surgical procedures and costs mismatch under the DRG / DIP payment model;

[0038] Consumable abuse: Identify consumable abuse by comparing the usage of consumables for the same projects in similar medical institutions.

[0039] 3. Fee Settlement Violation Risk Model:

[0040] Abnormal fund disbursement: Identify irregular amounts by comparing fund disbursement ratios with policy reimbursement ratios;

[0041] Abuse of personal accounts: Detect abnormal settlement behavior of insured individuals purchasing excessive amounts of medicine.

[0042] Step 3: Data Import

[0043] Data such as medical insurance settlement data, inventory data of drugs and high-value consumables from designated medical institutions, catalogs of medical service facilities, and income and expenditure data of medical services are imported into the system, and then the data is verified. The main checks are on data integrity, the uniqueness of the patient identification number (settlement document number), and the correlation between the settlement table and the detailed table.

[0044] Step Four: Big Data Analysis

[0045] Based on a knowledge base of violation risk models, big data analysis is used to screen for points of illegal and non-compliant risk. The specific algorithm model is as follows:

[0046] 1. Association rule mining:

[0047] The FP-Growth algorithm is used to screen for abnormal behavior in project portfolios.

[0048] 2. Time series analysis:

[0049] By analyzing the temporal characteristics of medical insurance settlement records using LSTM, we can identify high-frequency, large-amount abnormal settlement behaviors, such as irregular operations like "split hospitalization" and "over-medicalization".

[0050] 3. Cluster analysis:

[0051] After clustering the DRG / DIP grouped data, the "project overspending" behavior was identified.

[0052] The following example is a surprise inspection of a city's medical insurance fund in 2024:

[0053] First, a unified and standardized business code database was established based on the nationally unified medical insurance business coding standard. Then, medical insurance settlement data, high-value consumables inventory data, and treatment item revenue and expenditure data from the 10 hospitals to be inspected from January 1, 2022 to December 31, 2023 were imported into the system and data verification was completed. Next, based on the system's built-in violation risk model knowledge base, the system automatically filtered out the violation data through big data analysis.

[0054] Among the items listed, the "Carcinoembryonic Antigen Measurement," "Carbohydrate Antigen Measurement," "Left Ventricular Function Measurement," "Plasma Lactate Measurement," "Serium Bicarbonate Measurement," and "β2 Microglobulin Measurement" at a certain county-level Traditional Chinese Medicine Hospital met the characteristics of an over-testing model; the "Tracheotomy Care," "Corneal Curvature Measurement," "Manual Fracture Reduction," "Peripheral Nerve Compression Release," "Spinal Canal Decompression," "Pericular Picus Dilation," "Percutaneous Selective Arteriography," "Percutaneous Superselective Arteriography," "Lumbar Disc Herniation Removal," "Laparoscopy," "General Bacterial Smear Examination," and "Fungal Smear Examination" at a certain county-level Traditional Chinese Medicine Hospital met the characteristics of a duplicate charging model; and "C-arm Intraoperative Fluoroscopy" met the characteristics of an over-standard charging model. Through on-site verification of past receipts and personnel, the violations were confirmed. The total medical expenses involved were approximately 1.75 million yuan, resulting in a loss of approximately 1.11 million yuan to the medical insurance fund.

[0055] The embodiments described above are merely illustrative of specific implementations of the present invention, and while the descriptions are detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A method for detecting misuse of medical insurance funds based on big data analysis, characterized in that, Includes the following steps: Step 1: Establish a unified and standardized database: Establish a unified and standardized database based on medical insurance business coding standards and data specifications; Step 2: Establish a knowledge base for violation risk models, including violation risk models for medical treatment behavior, medical project violations, and fee settlement violations. Step 3: Data Import: Import medical insurance settlement data, inventory data of drugs and high-value consumables of designated medical institutions, catalog of medical service facilities, income and expenditure of medical services, etc. into the system, and then verify the data; Step 4: Big Data Analysis: Based on the knowledge base of the violation risk model, use big data analysis to screen out illegal and non-compliant risk points.

2. The method for detecting misuse of medical insurance funds based on big data analysis according to claim 1, characterized in that, In step 2, the medical treatment behavior violation risk model is specifically used to identify: ① Spatiotemporal clustering: abnormal behavior of the same insured person seeking medical treatment multiple times in a short period of time or settling accounts at the same medical institution; ② Spatiotemporal travel: spatiotemporal travel behavior of the insured person settling accounts at a distant medical institution in a short period of time; ③ High frequency and low amount: abnormal transaction behavior of frequent small-amount settlements; ④ False hospitalization identification: identifying false hospitalization behavior by comparing bed fees and nursing fees.

3. The method for detecting misuse of medical insurance funds based on big data analysis according to claim 2, characterized in that, In step 2, the medical project violation risk model is specifically used to identify: ① duplicate charges: multiple charges for the same project; ② charges exceeding the standard: comparing the price standard with the settlement amount to identify charges exceeding the standard; ③ project overcharging: analyzing the mismatch between surgical operations and costs under the DRG / DIP payment model; ④ consumable abuse: comparing the usage of consumables for the same project in similar medical institutions to identify consumable abuse.

4. The method for detecting misuse of medical insurance funds based on big data analysis according to claim 3, characterized in that, In step 2, the expense settlement violation risk model is specifically used to identify: ① abnormal fund disbursement: identifying the amount of violation by comparing the fund disbursement ratio with the policy reimbursement ratio; ② abuse of personal accounts: detecting abnormal settlement behavior of insured persons purchasing more drugs than permitted.

5. The method for detecting misuse of medical insurance funds based on big data analysis according to claim 1, characterized in that, In step 3, the data verification specifically includes: checking the integrity of the data, the uniqueness of the patient identification number, and the correlation between the settlement table and the detailed table.

6. The method for detecting misuse of medical insurance funds based on big data analysis according to claim 4, characterized in that, Step 4 specifically includes the following steps: Step 4.1, Association Rule Mining: Use the FP-Growth algorithm to screen for abnormal behavior in project portfolios; Step 4.2, Time Series Analysis: LSTM analysis is used to analyze the time series characteristics of medical insurance settlement records to identify high-frequency, high-amount abnormal settlement behaviors; Step 4.3, Cluster Analysis: After clustering the DRG / DIP grouped data, identify the high-risk behavior of the project.

Images