A financial data verification method and system

By using the Apriori algorithm to calculate support and enhancement in medical financial data, combined with timing segmentation and audit value coefficient, medical financial data verification is optimized, the problem of inefficiency in the existing technology is solved, and more efficient and accurate are achieved financial data verification.

CN118737470BActive Publication Date: 2024-11-05JINAN JINGJIN ELECTRONIC TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411214695.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-02
Publication Date
2024-11-05
Estimated Expiration
2044-09-02

AI Technical Summary

Technical Problem

The existing Apriori algorithm has low detection efficiency due to a large amount of patient data in medical financial data, making it difficult to effectively identify the phenomenon of charging for multiple items, resulting in high verification and low efficiency.

Method used

By obtaining financial verification data in HRP and HIS systems, the Apriori algorithm is used to calculate the support, confidence and improvement of the item set, and the patient's treatment item set is segmented in time, strong association rules are obtained, and the audit value coefficient and Cutoff coefficient, optimize audit strategies for efficiency and accuracy.

Benefits of technology

It improves the accuracy and efficiency of medical financial data verification, and dynamically adjusts the audit strategy, reduces unnecessary audit workload, optimizes financial processes, and improves the hospital's financial management capabilities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118737470B_ABST
    Figure CN118737470B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of data processing, and more specifically, to a method and system for verifying financial data, the method comprising: obtaining symptoms, symptom numbers and treatment items in the financial verification data of a medical system, obtaining a set of treatment items corresponding to each person for the symptom, and using an Apriori algorithm to obtain the support, confidence and lift of the item set; dividing each patient into multiple segments in time sequence, obtaining a set of strong association rules corresponding to each segment, and classifying them, merging the strong association rules between similar items as comprehensive association rules, and calculating the audit value coefficient of each patient, sorting the audit value coefficient of each patient, obtaining the audit value coefficient curve of the audited patients to analyze the audit cutoff coefficient; in response to the audit cutoff coefficient being less than a preset threshold, no audit is performed in the classification or a random sampling audit is adopted. The present invention intelligently adjusts the audit strategy through the audit cutoff coefficient to improve the overall efficiency of the audit process.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of data processing, and more specifically, to a financial data verification method and system. Background Art

[0002] In the medical industry, especially in the field of hospital finance, the application of financial data verification technology is becoming increasingly important. With the rapid development of big data, cloud accounting and AI technology, the work of financial data verification is further deepened. In order to further improve the high quality of hospital services and strengthen the management of hospitals, at present, there are multiple charges in the existing medical financial data verification. Because different patients may have different charges, it is difficult to detect the multiple charges in the financial data verification.

[0003] The existing paper "Research on Intelligent Internal Audit of Medical Charging Behavior of A District Traditional Chinese Medicine Hospital Based on RPA+AI - He Jiayu" proposes a correlation detection method based on the Apriori algorithm to detect multi-item charging phenomena and reduce the amount of medical financial data audit. The paper mentions that based on the Apriori association rule algorithm, pathological characteristics, drugs, medical charges, prescriptions and other medical behaviors are analyzed for correlation, and relevant drugs are prescribed according to pathological characteristics, and appropriate charges are made. Then, the drugs are compared with the prescriptions prescribed by the doctors to analyze whether there are multiple prescriptions, random prescriptions, overcharges and random charges, so as to achieve the effect of automated review and standardize medical behavior.

[0004] At present, due to the large amount of patient data in medical financial data, the efficiency of using the Apriori algorithm to detect the correlation of medical projects is low. The Apriori algorithm needs to generate all possible candidate item sets to find frequent item sets. This process grows exponentially with the increase of the item set length. After the strong correlation rules between the treatment project data are found, they still need to be reviewed one by one, and the accuracy of the verification must be ensured. Summary of the invention

[0005] In order to solve the problem that there is a large amount of patient data in medical financial data, but the efficiency of using the Apriori algorithm to detect the association of medical items is low, the present invention provides solutions in the following aspects.

[0006] In a first aspect, a financial data verification method comprises: obtaining financial verification data in an HRP system and an HIS system, wherein the financial verification data comprises symptoms, symptom numbers and treatment items, obtaining a set of treatment items corresponding to each person for each symptom, and using an Apriori algorithm to obtain an item set, and calculating support, confidence and lift of the item set; dividing the set of treatment items for each patient in each symptom into multiple segments in time sequence, obtaining a set of strong association rules corresponding to each segment, classifying the set of strong association rules, and merging the strong association rules between similar items in multiple segments as comprehensive association rules; and calculating the support, confidence and lift of each patient based on the comprehensive association rules. The audit value coefficient is used to characterize whether the financial verification data is abnormal. The audit value coefficient of each patient is sorted and audited to obtain the audit value coefficient curve of the audited patients; based on the audit value coefficient curve, the ratio of the average value of the audit value coefficient of each patient with a preset number of consecutive normal audits to the average value of the audit value coefficient of all patients in the corresponding category with normal audits is calculated, the variance of the audit value coefficient of each patient in the category with a preset number of consecutive normal audits is calculated, and the product of the ratio and the variance is used as the audit cutoff coefficient; in response to the audit cutoff coefficient being less than the preset threshold, the remaining patients in the category will no longer be audited or random sampling audits will be adopted.

[0007] The effect is as follows: using the Apriori algorithm to obtain the set of treatment items corresponding to the disease, and calculating the support, confidence and lift, it is helpful to identify common treatment patterns and potential abnormal patterns. By time-series segmenting the set of treatment items for each disease, it is possible to capture treatment patterns and trends that change over time, and sort the audit value coefficients of patients so that the audit work can give priority to those patients with higher coefficients, thereby improving the targeted nature of the audit. Through the audit value coefficient curve of the audited patients, it is possible to observe the changing trend of the audit value over time, providing a dynamic perspective for the audit decision-making. By calculating the audit cutoff coefficient, a quantitative cutoff point is provided for the audit process, which helps to reduce unnecessary audit workload while ensuring the quality of the audit, improves the accuracy and efficiency of the hospital's financial data verification, and helps the hospital better manage and optimize its financial processes.

[0008] Preferably, the strong association rule is extracted from the association rules of each segment, that is, the rule that meets the minimum support and minimum confidence threshold is used as the strong association rule, and the strong association rule is a triple: (project, related project, promotion degree).

[0009] Preferably, classifying the set of strong association rules includes:

[0010] Take any segment as the target segment, take the items and related items in the strong association rules of the target segment as nodes, and the promotion degree as the edge weight, establish graph structure data for the strong association rules of each segment, classify the graph structure data, and obtain multiple segment classification results.

[0011] The effect is: through graph structure data, complex strong association rules can be displayed in a graphical way, and the use of lift as edge weight can quantify the strength of association between different items. Through classification results, strong association rules can be managed and maintained more effectively. Through the classification of graph structure data, association rules that need to be reviewed can be quickly located, thereby improving the efficiency of financial data verification.

[0012] Preferably, the merging of strong association rules between similar items in multiple segments as comprehensive association rules includes:

[0013] Obtain strong association rules between similar items in the classification results of multiple segments, calculate the average of the lifts in the strong association rules between similar items as the combined lift, and use the strong association rules corresponding to the combined lift as the comprehensive association rules.

[0014] The effect is that merging the mean of the lift helps to extract more stable and representative comprehensive association rules, reducing the excessive influence of the particularity of individual segments on the rules. Through mean calculation, the influence of outliers or extreme values ​​on the strength of association rules can be reduced, making the comprehensive association rules more robust.

[0015] Preferably, the calculation of the audit value coefficient for each patient includes:

[0016] Take any patient as the target patient, take the category where the target patient belongs as the target category, obtain all the diagnosis and treatment item sets of the target patient, and calculate the matching degree between the diagnosis and treatment item set of the target patient and the comprehensive association rule of the target category where the target patient belongs;

[0017] Sort the audit value coefficient sequence that has been audited in the segment, obtain the risk items that have been audited, calculate the proportion of the risk items in the risk items that have been audited, and use the maximum value of the proportion of the risk items in the diagnosis and treatment item set of the target patient to the risk items that have been audited as the audit value adjustment coefficient;

[0018] The product of the matching degree and the audit value adjustment coefficient is used as the audit value coefficient.

[0019] The effect is: through the calculation of the matching degree, the similarity between the target patient's diagnosis and treatment item set and the comprehensive association rules can be quantified, which helps to identify abnormal situations that do not conform to the normal pattern. The audit value adjustment coefficient is based on the proportion of risky items that have been audited, so that the audit pays more attention to those risky items that may have problems. By giving priority to verifying patients with high audit value coefficients, audit resources can be allocated more effectively and audit efficiency can be improved.

[0020] Preferably, obtaining the matching degree includes the steps of: taking any patient as a target patient, obtaining all diagnosis and treatment item sets of the target patient, calculating the intersection of the diagnosis and treatment item set of the target patient and all item sets in the comprehensive association rule of the corresponding classification result, and taking the ratio of the number of diagnosis and treatment items in the intersection to the number of diagnosis and treatment item sets of the target patient as the approximation;

[0021] Calculate the cumulative sum of the lifts corresponding to the subset of strong association rules for the diagnosis and treatment items of the target patient in the intersection, calculate the cumulative sum of the lifts corresponding to the subset of strong association rules for the diagnosis and treatment items of all patients in the target category where the target patient is located, and take the ratio of the cumulative sum of the target patient's lifts to the cumulative sum of the lifts of all patients as the mean difference; take the product of the approximation and the mean difference as the matching degree between each patient's diagnosis and treatment items and the comprehensive association rules.

[0022] Preferably, the audit value adjustment coefficient includes:

[0023] In response to the number of audits of the target category being less than a preset number or the number of patients being less than a preset number, the audit value adjustment coefficient is set to 1.

[0024] In response to the audit value adjustment coefficient being not 1, the audited patients are sorted according to their audit value coefficients, risk items are obtained, the proportion of the risk items in the audited risk items is calculated, and the maximum value of the proportion is obtained as the audit value coefficient.

[0025] The effect is: for situations where the number of initial audits is insufficient or the number of patients is small, the audit value adjustment coefficient is set to 1, which simplifies the calculation of the audit value coefficient and speeds up the audit process. As the number of audits increases and the number of patients accumulates, the audit value adjustment coefficient can be adaptively adjusted to reflect a more accurate risk assessment. By sorting and calculating the proportion of risk items for audited patients, the audit value adjustment coefficient can focus on high-risk items and improve the targeted nature of the audit.

[0026] In a second aspect, a financial data verification system includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned financial data verification method is implemented.

[0027] The present invention has the following effects:

[0028] 1. The present invention realizes a dynamic assessment of the risk of patient financial data by calculating the audit value coefficient curve. The introduction of the audit cutoff coefficient allows the system to intelligently adjust the audit strategy according to the current risk assessment results, optimize the use of audit resources, and reduce the audit frequency of low-risk patients, thereby focusing on high-risk areas where problems may exist. When the audit cutoff coefficient is lower than the preset threshold, the system will reduce or stop the audit of the remaining patients, or adopt a random sampling method for audit, which helps to improve the overall efficiency of the audit process and avoid repeated and possibly unnecessary audit work on patients with normal continuous audits and low risks.

[0029] 2. The present invention provides a flexible audit mechanism for financial data verification by introducing an audit cutoff coefficient, thereby dynamically adjusting according to actual audit results and risk assessment to ensure that the audit work is more targeted while maintaining the ability to respond quickly to new risks, thereby improving the adaptability and effectiveness of financial data verification. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0031] Figure 1 It is a method flow chart of steps S1 to S4 in a financial data verification method according to an embodiment of the present invention;

[0032] Figure 2 It is a structural block diagram of a financial data verification system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0033] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0034] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0035] Reference Figure 1 A financial data verification method includes steps S1 to S4, which are as follows:

[0036] S1: Obtain the financial verification data from the HRP system and HIS system. The financial verification data includes symptoms, symptom numbers, and treatment items. Obtain the set of treatment items corresponding to each person's symptoms, and use the Apriori algorithm to obtain the item set, and calculate the support, confidence, and lift of the item set.

[0037] The collection and processing of financial verification data is completed through the HRP and HIS systems, where HRP is the hospital resource management system and HIS is the hospital information management system.

[0038] It should be noted that the number is set according to each symptom in the financial verification data. , get the Symptoms All treatment items, including but not limited to medication, laboratory tests, surgery, physical examinations and other items that require payment, etc.; Get the first Symptoms Middle A collection of treatment programs for each patient , use the Apriori algorithm to obtain item sets, calculate the support of each item set, filter frequent item sets in the item set, and calculate the confidence and lift values ​​of each frequent item set.

[0039] For example, if the symptoms are cold, fever, acute bronchitis, gastritis, gastric ulcer, etc., taking cold as an example, its label is set to ; Common colds are treated with drugs, but are not limited to drug treatment items. In this embodiment, drug treatment items are taken as examples, such as medicines: 999 Ganmao Ling Granules, Amoxicillin Capsules, Cough Syrup, Ibuprofen Capsules, etc.;

[0040] Using association rule learning methods such as the Apriori algorithm, the proportion of the number of times each item set appears in all treatment item sets to the total data set is calculated, and the proportion is used as the support degree corresponding to the two data.

[0041] Specifically, the support satisfies the following relationship:

[0042] ;

[0043] In the formula, Representation data and data The support of Representation data and data The total number of occurrences of Indicates the total number of data sets.

[0044] Generally speaking, data with high support does not necessarily constitute a frequent itemset, but data with too low support definitely does not constitute a frequent itemset. Support is for frequent itemsets, so this scheme sets the minimum support threshold to 0.2, retains itemsets greater than the minimum support threshold as frequent itemsets, and also plays the role of an itemset filter.

[0045] Get frequent item set data Data , and further analyze the correlation between the two data of the frequent item set;

[0046] Specifically, the confidence satisfies the following relationship:

[0047] ;

[0048] In the formula, Representation data Data The confidence level, Representation data and data The total number of occurrences of Representation data The total number of occurrences of .

[0049] Generally speaking, the greater the confidence level of the data, Including data The greater the possibility of occurrence.

[0050] Further analysis will obtain data Data Confidence and data The ratio of the probability of occurrence of the whole is taken as data Data The degree of improvement;

[0051] Specifically, the lift satisfies the following relationship:

[0052] ;

[0053] In the formula, Representation data Data The degree of improvement, Representation data Data The confidence level, Representation data The total number of occurrences of .

[0054] Further explanation: if the lift is greater than 1, it means that the data data is a valid strong association rule. If the lift is less than or equal to 1, it means it is an invalid strong association rule. If it is equal to 1, it means that the data and data Independent; and then obtain the strong association rules of frequent item sets.

[0055] S2: Divide the set of treatment items for each patient in each disease into multiple segments in time sequence, obtain the set of strong association rules corresponding to each segment, classify the set of strong association rules, and merge the strong association rules between similar items in multiple segments as comprehensive association rules.

[0056] Further explanation, the division rule is: divide according to the order of seeing a doctor for each patient in each disease, wherein, when dividing, the treatment project data of every thousand patients is a segment, and multiple segments are obtained.

[0057] Among them, the strong association rule extracts the strong association rule from the association rules of each segment, that is, the rule that meets the minimum support and minimum confidence threshold is taken as the strong association rule, and the strong association rule is a triple: (project, related project, promotion degree).

[0058] Take any segment as the target segment, take the items and related items in the strong association rules of the target segment as nodes, and the promotion degree as the edge weight, establish graph structure data for the strong association rules of each segment, classify the graph structure data, and obtain multiple segment classification results.

[0059] The strong association rules between similar items in the classification results of multiple segments are obtained, the average of the lifts in the strong association rules between similar items is calculated as the combined lift, and the strong association rules corresponding to the combined lift are used as comprehensive association rules.

[0060] S3: Based on the comprehensive association rules, the audit value coefficient of each patient is calculated. The audit value coefficient is used to characterize whether the financial verification data is abnormal. The audit value coefficient of each patient is sorted and audited to obtain the audit value coefficient curve of the audited patients.

[0061] Take any patient as the target patient, take the category where the target patient belongs as the target category, obtain all the diagnosis and treatment item sets of the target patient, and calculate the matching degree between the diagnosis and treatment item set of the target patient and the comprehensive association rule of the target category where the target patient belongs;

[0062] The specific steps to obtain the matching degree include:

[0063] Take any patient as the target patient, obtain all the diagnosis and treatment item sets of the target patient, calculate the intersection of the diagnosis and treatment item set of the target patient and all the item sets in the comprehensive association rules of the corresponding classification results, and take the ratio of the number of diagnosis and treatment items in the intersection to the number of diagnosis and treatment item sets of the target patient as the approximation;

[0064] In other words, the smaller the approximation, the more likely it is that the diagnosis and treatment items in the intersection are The more similar the set of all medical treatment items for a patient is, the lower the audit value is, and vice versa.

[0065] At the same time, there may also be a problem of overcharging, and then further analyze the cumulative sum of the corresponding improvement degrees of the strong association rules between the diagnosis and treatment items of each patient. Among them, the subset items between non-strong association rules do not participate in the calculation of the cumulative sum. The specific steps are as follows:

[0066] Calculate the cumulative sum of the lifts corresponding to the subset of the strong association rule for the diagnosis and treatment items of the target patient in the intersection, calculate the cumulative sum of the lifts corresponding to the subset of the strong association rule for the diagnosis and treatment items of all patients in the target classification where the target patient is located, and take the ratio of the cumulative sum of the target patient's lifts to the cumulative sum of the lifts of all patients as the mean difference;

[0067] The product of the similarity and the mean difference is taken as the matching degree between each patient's diagnosis and treatment items and the comprehensive association rules.

[0068] In other words, the smaller the matching value is, the more the number of treatment items in the strong association rules within this category is close to or less than the mean, where a small number of treatment items is uncommon. The closer the mean is or the smaller the number of treatment items is, the lower the audit value is, and vice versa.

[0069] Sort the audit value coefficient sequence that has been audited in the segment, obtain the risk items that are being audited, calculate the proportion of risk items in the risk items that have been audited, and use the maximum value of the proportion of risk items in the target patient's diagnosis and treatment item set to the risk items that have been audited as the audit value adjustment coefficient;

[0070] The product of the matching degree and the audit value adjustment coefficient is taken as the audit value coefficient.

[0071] Specifically, the audit value coefficient satisfies the following relationship:

[0072] ;

[0073] In the formula, Indicates The audit value coefficient of each patient, Indicates The diagnosis and treatment items of each patient The matching degree of the comprehensive association rules of the classification of each patient, Indicates The audit value adjustment factor for each patient.

[0074] That is to say, the larger the audit value coefficient value, the The higher the patient's audit value coefficient, the more likely it is to be audited.

[0075] In response to the number of audits of the target category being less than a preset number or the number of patients being less than a preset number, the audit value adjustment coefficient is set to 1.

[0076] In response to the audit value adjustment coefficient being not 1, the audited patients are sorted according to their audit value coefficients, risk items are obtained, the proportion of risk items in the audited risk items is calculated, and the maximum value of the proportion is obtained as the audit value coefficient.

[0077] Further explanation: the preset number of audit times is set to 10 times.

[0078] It should be noted that the unreviewed patients are sorted in descending order according to the review value coefficient, and the patients with high review value coefficients are reviewed first. If the review is carried out all the time, all of them will be reviewed in the end, and then the patients with high review value coefficients should be selected for review. When the review value coefficient becomes low and stable, the remaining patients will no longer be reviewed for medical items to reduce the workload. The specific operation steps are as follows:

[0079] S4: Based on the audit value coefficient curve, calculate the ratio of the average value of the audit value coefficient of each patient with a preset number of consecutive normal audits to the average value of the audit value coefficient of all patients with normal audits in the corresponding category, calculate the variance of the audit value coefficient of each patient in the category with a preset number of consecutive normal audits, and take the product of the ratio and the variance as the audit cutoff coefficient; in response to the audit cutoff coefficient being less than the preset threshold, the remaining patients in the category will no longer be audited or a random sampling audit will be adopted.

[0080] To further explain, the audit value coefficient curve of the audited patients in the classification is obtained. When the audit value coefficient value is low and relatively stable, it means that the audit can be terminated.

[0081] Specifically, the audit cut-off coefficient satisfies the following relationship:

[0082] ;

[0083] In the formula, Indicates the current The audit cut-off coefficient for each patient, Indicates The average value of the first 10 consecutive normal audits of the patient is The ratio of the average value of the normal audit value coefficient of all patients in the corresponding classification to the patient. Indicates The variance of the coefficient of normal audit value for the first 10 consecutive audits of each patient.

[0084] It should be noted that The lower the value, the lower the audit cut-off coefficient of consecutive patients, and the lower the risk, and the value of continuing the audit is not great. If one of the 10 consecutive audits is an abnormal patient, the audit cut-off coefficient ; Among them, the preset threshold is 0.08, which can be adjusted by technical personnel in this field according to the specific implementation scenario.

[0085] The present invention also provides a financial data verification system. Figure 2 As shown, the system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a financial data verification method according to the first aspect of the present invention is implemented.

[0086] The system also includes other components familiar to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

[0087] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0088] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0089] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A financial data verification method, characterized in that: include: Obtain financial verification data from the HRP system and HIS system, the financial verification data including symptoms, symptom numbers and treatment items, obtain the set of treatment items corresponding to each symptom, and use the Apriori algorithm to obtain item sets, and calculate the support, confidence and lift of the item sets; Divide the treatment item set of each patient in each disease into multiple segments in time sequence, obtain a set of strong association rules corresponding to each segment, classify the set of strong association rules, obtain strong association rules between similar items in the classification results of multiple segments, calculate the average of the lifts in the strong association rules between similar items as the combined lift, and use the strong association rules corresponding to the combined lift as the comprehensive association rules; Take any patient as the target patient and the category to which the target patient belongs as the target category, obtain all the diagnosis and treatment item sets of the target patient, calculate the intersection of the diagnosis and treatment item set of the target patient and all the item sets in the comprehensive association rules of the corresponding classification results, and take the ratio of the number of diagnosis and treatment items in the intersection to the number of diagnosis and treatment item sets of the target patient as the approximation; Calculate the cumulative sum of the lifts corresponding to the subsets of the strong association rules for the diagnosis and treatment items of the target patient in the intersection, calculate the cumulative sum of the lifts corresponding to the subsets of the strong association rules for the diagnosis and treatment items of all patients in the target category where the target patient is located, and take the ratio of the cumulative sum of the target patient's lifts to the cumulative sum of the lifts of all patients as the mean difference; The product of the approximation and the mean difference is used as the matching degree between the diagnosis and treatment items of each patient and the comprehensive association rule; Sort the audit value coefficient sequence that has been audited in the segment, obtain the risk items that have been audited, calculate the proportion of the risk items in the risk items that have been audited, and use the maximum value of the proportion of the risk items in the diagnosis and treatment item set of the target patient in the risk items that have been audited as the audit value adjustment coefficient; in response to the number of audits of the target classification being less than a preset number or the number of patients being less than a preset number, the audit value adjustment coefficient is set to 1; in response to the audit value adjustment coefficient not being 1, sort according to the audit value coefficients of the audited patients, obtain the risk items, calculate the proportion of the risk items in the audited risk items, and obtain the maximum value of the proportion as the audit value coefficient; The product of the matching degree and the audit value adjustment coefficient is used as the audit value coefficient; Sorting and reviewing the audit value coefficient of each patient to obtain an audit value coefficient curve of the reviewed patients; Based on the audit value coefficient curve, the ratio of the average value of the audit value coefficient of each patient with a preset number of consecutive normal audits to the average value of the audit value coefficient of all patients in the corresponding category is calculated, and the variance of the audit value coefficient of each patient in the category with a preset number of consecutive normal audits is calculated, and the product of the ratio and the variance is used as the audit cutoff coefficient; in response to the audit cutoff coefficient being less than the preset threshold, the remaining patients in the category will no longer be audited or a random sampling audit will be adopted.

2. A financial data verification method according to claim 1, characterized in that: in, The strong association rule extracts a strong association rule from the association rules of each segment, that is, a rule that meets the minimum support and minimum confidence threshold is used as a strong association rule, and the strong association rule is a triple: (item, associated item, promotion degree).

3. A financial data verification method according to claim 1, characterized in that: The set of strong association rules is classified, including: Take any segment as the target segment, take the items and related items in the strong association rules of the target segment as nodes, and the promotion degree as the edge weight, establish graph structure data for the strong association rules of each segment, classify the graph structure data, and obtain multiple segment classification results.

4. A financial data verification system, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the financial data verification method according to any one of claims 1-3 is implemented.