Traditional Chinese medicine medical insurance reimbursement process management method and device, equipment and medium

Abstract

The invention relates to a traditional Chinese medicine medical insurance reimbursement process management method and device, equipment and a medium. According to the method, traditional Chinese medicine diagnosis and treatment feature data (including prescription composition, dose use and diagnosis and treatment item data) of a patient is acquired, compatibility analysis is performed by using a traditional Chinese medicine knowledge graph, and a dose deviation degree and therapy suitability are evaluated in combination with a clinical pathway specification database. And generating a comprehensive risk scoring level through a dynamic risk assessment model trained based on historical data. And when the score reaches a corresponding preset threshold value, generating a reimbursement decision instruction containing the risk level to guide the reimbursement process. By adopting the method, the efficiency and accuracy of traditional Chinese medicine medical insurance reimbursement are improved, the risk monitoring capability is enhanced, and safe use of medical insurance funds is effectively guaranteed.

Description

Technical Field

[0001] The present invention belongs to the field of medical system expense settlement, and in particular relates to a method, device, equipment and medium for managing the medical insurance reimbursement process of traditional Chinese medicine. Background Art

[0002] With the development of Traditional Chinese Medicine (TCM) technologies, its application in the healthcare system is becoming increasingly widespread. With its unique theoretical framework and rich clinical experience, TCM has demonstrated significant advantages in treating a wide range of diseases. However, with the increasing popularity of TCM services, the management of TCM medical insurance reimbursement processes is also facing new challenges.

[0003] Traditionally, medical insurance reimbursement for Traditional Chinese Medicine (TCM) has relied primarily on manual review. Medical insurance reviewers need to examine each reimbursement document individually, verifying the rationality of the prescription, compliance with dosage requirements, and the standardization of treatment items.

[0004] However, the current manual review method for reimbursement processing has problems such as low review efficiency, insufficient accuracy, lack of real-time risk monitoring, and difficulty in adapting to complex situations. Therefore, a more intelligent and automated solution is urgently needed to improve management level and ensure the safe use of medical insurance funds. Summary of the Invention

[0005] Based on this, it is necessary to provide a method, device, equipment and medium for managing the medical insurance reimbursement process of traditional Chinese medicine that can improve efficiency and accuracy while enhancing risk monitoring capabilities in response to the above technical problems.

[0006] In the first aspect, this application provides a method for managing the medical insurance reimbursement process for traditional Chinese medicine, including:

[0007] Obtaining the target patient's TCM diagnosis and treatment characteristic data, including prescription composition data, dosage data, and diagnosis and treatment item data;

[0008] Based on prescription composition data, compatibility analysis is performed through a pre-built Traditional Chinese Medicine knowledge graph to obtain compatibility taboo identification results and empirical prescription compliance identification results;

[0009] Based on the dosage usage data and diagnosis and treatment project data, the pre-stored TCM clinical pathway standard database is used for calculation and analysis to generate the dosage deviation score and therapy suitability identification results;

[0010] Based on the results of the incompatibility labeling, empirical prescription compliance labeling, dosage deviation score, and therapy suitability labeling, a dynamic risk assessment model is used to generate a risk score. The risk assessment model is a machine learning model trained on violation cases from historical TCM reimbursement data.

[0011] Based on the risk scoring result, a corresponding reimbursement decision instruction is generated; the reimbursement decision instruction includes the risk level, which is used to characterize the reimbursement process.

[0012] In one embodiment, based on prescription composition data, compatibility analysis is performed through a pre-built Traditional Chinese Medicine knowledge graph to obtain compatibility taboo identification results and empirical prescription compliance identification results, including:

[0013] Send the prescription composition data to the constructed Traditional Chinese Medicine knowledge graph, which includes a compatibility taboo rule base and an empirical prescription database;

[0014] According to the incompatibility rule library, the drug combinations in the prescription composition data are matched with the incompatibility rules to generate the incompatibility identification results;

[0015] Based on the empirical prescription database, the matching degree between the prescriptions in the prescription composition data and the historical empirical prescriptions is analyzed through a similarity calculation model to generate the empirical prescription compliance identification results.

[0016] In one embodiment, based on the dosage usage data and the diagnosis and treatment item data, a dosage deviation score and a treatment suitability identification result are generated by calculation and analysis using a pre-stored database of TCM clinical pathway specifications, including:

[0017] Based on the dosage usage data, the following formula is used to extract, calculate and analyze the single-drug dosage and medication course data to obtain the dosage deviation score:

[0018]

[0019] Where D is the dose deviation score, n is the number of medicinal ingredients in the prescription, ω i is the dosage weight coefficient of the i-th drug, d i is the actual dosage, d im It is the median of the standard dose range recorded in the clinical pathway specification database.

[0020] In one embodiment, based on the results of the incompatibility identification, the empirical prescription compliance identification, the dosage deviation score, and the therapy suitability identification, a comprehensive risk score is generated through dynamic risk assessment model analysis, including:

[0021] The results of incompatibility identification, empirical prescription compliance identification, dosage deviation score and therapy suitability identification were converted into standardized feature vectors;

[0022] The standardized feature vector is sent to the dynamic risk assessment model trained based on the XGBoost algorithm to generate a comprehensive risk score; the comprehensive risk score system is 0-100 points; the comprehensive risk score is calculated using the following formula:

[0023] C=α·TC+β·D+γ·(1-TM)+δ·EC+η·log(1+H)

[0024] Among them, C is the comprehensive risk score, TC is the incompatibility labeling result, D is the individual dose deviation score, TM is the therapy matching degree, EC is the empirical prescription compliance labeling score, H is the patient's historical reimbursement violation count, and α, β, γ, δ, and η are dynamic weight parameters. The dynamic weight parameters are actively learned by the XGBoost algorithm during the training process by minimizing the loss function between the prediction error of historical data and the actual violation label.

[0025] According to the preset risk level threshold range, the comprehensive risk score is mapped to low, medium and high risk levels to obtain the risk score level result.

[0026] In one embodiment, the comprehensive risk score is mapped to low, medium, and high risk levels according to a preset risk level threshold interval, and the risk score level result is obtained, including:

[0027] Based on the comprehensive risk score, the continuity risk index is generated by normalization through a preset logistic regression function. The calculation formula for the continuity risk index is:

[0028]

[0029] Among them, S is the continuity risk index, β is the parameter for adjusting the slope of the curve, and μ is the risk level classification threshold;

[0030] Based on the distribution characteristics of historical risk data, the initial preset risk level threshold interval is dynamically calibrated to generate an optimized preset risk level threshold interval;

[0031] Based on the optimized preset risk level threshold interval, the continuous risk index is divided into intervals to generate the corresponding discrete risk level results.

[0032] In one embodiment, the training process of the dynamic risk assessment model includes:

[0033] Construct a training dataset based on compliance cases and non-compliance cases in historical TCM reimbursement data;

[0034] Based on the training data set, the incompatibility features, empirical prescription matching features, dosage deviation features, and therapy adaptability features in each case of the training data set are extracted to obtain a feature data set;

[0035] Based on the feature data set, the dynamic risk assessment model is trained using the XGBoost algorithm to obtain a trained dynamic risk assessment model; wherein the dynamic risk assessment model is an ensemble learning model based on the gradient boosting tree;

[0036] The trained dynamic risk assessment model is verified through the validation set, and the model parameters are adjusted according to the verification results to obtain an optimized dynamic risk assessment model.

[0037] In one embodiment, the comprehensive risk score is mapped to low, medium, and high risk levels according to a preset risk level threshold interval. After the risk score level result is obtained, the following steps are further included:

[0038] When the risk level result is high risk, the generated reimbursement decision instruction is an electronic review work order containing the risk point identification; the electronic review work order is used to remind manual review and processing;

[0039] When the risk level result is medium risk, the reimbursement decision instruction generated is to dynamically adjust the reimbursement ratio based on the preset attenuation rules and generate reimbursement result data with risk tags;

[0040] When the risk level result is low risk, the reimbursement decision instruction generated is an electronic confirmation voucher.

[0041] Secondly, this application also provides a TCM medical insurance reimbursement process management device, including:

[0042] A data acquisition module is used to obtain the TCM diagnosis and treatment characteristic data of the target patient, which includes prescription composition data, dosage usage data, and diagnosis and treatment item data;

[0043] The compatibility analysis module is used to perform compatibility analysis based on prescription composition data through a pre-built traditional Chinese medicine knowledge graph to obtain compatibility taboo identification results and empirical prescription compliance identification results;

[0044] The pathway specification calculation and analysis module is used to calculate and analyze the dosage usage data and diagnosis and treatment project data through the pre-stored traditional Chinese medicine clinical pathway specification database to generate dosage deviation scores and treatment suitability identification results;

[0045] The dynamic risk assessment module is used to generate risk score results through dynamic risk assessment model analysis based on the results of incompatibility identification, empirical prescription compliance identification, dosage deviation score, and therapy suitability identification.

[0046] The reimbursement decision generation module is used to generate corresponding reimbursement decision instructions based on the risk scoring level results.

[0047] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements any step of the first aspect when executing the computer program.

[0048] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which implements any step of the above-mentioned first aspect when the computer program is executed by a processor.

[0049] The above-mentioned TCM medical insurance reimbursement process management methods, devices, equipment and storage media have significantly improved the efficiency and accuracy of TCM medical insurance reimbursement through intelligent analysis and automated decision-making, while enhancing risk monitoring capabilities and effectively ensuring the safe use of medical insurance funds. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0051] Figure 1 This is a flow chart of the method for managing the medical insurance reimbursement process for traditional Chinese medicine of the present invention;

[0052] Figure 2 Another flow chart of the method for managing the process of medical insurance reimbursement for traditional Chinese medicine of the present invention;

[0053] Figure 3 This is a structural block diagram of the TCM medical insurance reimbursement process management device of the present invention. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0055] In an exemplary embodiment, Figure 1 As shown, a method for managing the medical insurance reimbursement process for traditional Chinese medicine is provided. This method is described using the application of the method to a medical insurance reimbursement management terminal as an example. It is understood that the method can also be applied to a server, or to a system including a medical insurance reimbursement management terminal and a server, and implemented through interaction between the medical insurance reimbursement management terminal and the server. In this embodiment, the method includes the following steps S111 to S115. Among them:

[0056] S111, obtaining the TCM diagnosis and treatment characteristic data of the target patient, the TCM diagnosis and treatment characteristic data including prescription composition data, dosage usage data and diagnosis and treatment item data.

[0057] Optionally, TCM diagnosis and treatment characteristic data refers to detailed information related to the patient's TCM treatment, including prescription composition data (such as drug name, dosage, etc.), dosage usage data (such as whether the drug dosage meets the standards) and diagnosis and treatment item data (such as specific treatment methods and examination items). The above data is the basis for evaluating the rationality of TCM treatment and the compliance of medical insurance reimbursement.

[0058] Extract target patients' TCM diagnosis and treatment data from the medical information system. This data covers prescription composition, dosage, and treatment items. Through data interfaces or data import functions, the medical insurance reimbursement management terminal integrates this information into a structured dataset, providing data support for subsequent risk assessment and decision-making.

[0059] S112, based on prescription composition data, performs compatibility analysis through a pre-built traditional Chinese medicine knowledge graph to obtain the compatibility contraindication identification results and the empirical prescription compliance identification results.

[0060] For example, compatibility analysis refers to the evaluation of the interactions between drugs in a prescription based on traditional Chinese medicine theory to determine whether there are any incompatibilities or whether it complies with the compliance of the empirical prescription; the incompatibility identification result indicates whether there is a combination of drug interactions that may affect the efficacy or safety; and the empirical prescription compliance identification result determines whether the prescription follows a known effective prescription.

[0061] The prescription composition data is analyzed using a pre-built Traditional Chinese Medicine knowledge graph. The knowledge graph contains the interaction relationships between drugs and the composition rules of empirical prescriptions. By matching the information in the knowledge graph, the incompatibility taboos in the prescription and the compliance of the empirical prescription are identified, and the corresponding identification results are generated to provide key information for subsequent risk assessment.

[0062] S113, based on the dosage usage data and diagnosis and treatment item data, the dosage deviation score and treatment suitability identification result are generated through calculation and analysis of the pre-stored traditional Chinese medicine clinical pathway specification database.

[0063] Among them, the dose deviation score is a quantitative indicator to measure the difference between the actual dose and the standard dose, reflecting the rationality of dose use; the therapy adaptability identification result evaluates whether the diagnosis and treatment project complies with the clinical pathway standards of traditional Chinese medicine to ensure the appropriateness of the treatment means.

[0064] Based on the pre-stored database of TCM clinical pathway specifications, the dosage usage data and diagnosis and treatment item data are calculated and analyzed. By comparing the actual data with the standard specifications, the dosage deviation score and therapy suitability identification results are generated to provide data support for comprehensive risk assessment and ensure the standardization and rationality of treatment.

[0065] S114, based on the results of incompatibility labeling, empirical prescription compliance labeling, dosage deviation score and therapy adaptability labeling, generates risk score level results through dynamic risk assessment model analysis; the risk assessment model is a machine learning model trained based on violation cases in historical TCM reimbursement data.

[0066] Optionally, the dynamic risk assessment model is a model built based on machine learning technology. It is trained using violation cases in historical TCM reimbursement data and can predict the risk level of the current reimbursement application based on input feature data (such as incompatibility identification results, empirical prescription compliance identification results, etc.); the risk score level result is a quantitative representation of the risk level of the reimbursement application, which is used to guide subsequent reimbursement decisions.

[0067] The results of incompatibility labeling, empirical prescription compliance labeling, dosage deviation score, and therapy suitability labeling are transmitted to the dynamic risk assessment model, the characteristic data are analyzed, and the comprehensive risk score level is calculated based on the parameters obtained from historical violation case training, providing a scientific basis for reimbursement decisions.

[0068] S115, generating a corresponding reimbursement decision instruction based on the risk scoring result; the reimbursement decision instruction includes the risk level, which is used to characterize the reimbursement process.

[0069] Specifically, the reimbursement decision instruction is generated based on the risk scoring results and is used to guide the execution of the medical insurance reimbursement process. It contains risk level information, characterizes the risk level of the reimbursement application, and provides a decision-making reference for the reimbursement processing process. According to the risk scoring results, the corresponding reimbursement decision instruction is generated according to the preset rules. The reimbursement decision instruction contains risk level information, which is used to characterize the risk level of the reimbursement process, provide clear guidance for subsequent reimbursement operations, and ensure the rationality and security of medical insurance reimbursement.

[0070] In the above-mentioned TCM medical insurance reimbursement process management method, by systematically acquiring and analyzing TCM diagnosis and treatment characteristic data, and utilizing knowledge graphs, clinical pathway specification databases, and dynamic risk assessment models, it is possible to efficiently and accurately assess the risk level of reimbursement applications, thereby providing a scientific basis for medical insurance reimbursement decisions, improving management efficiency, and ensuring the safe use of medical insurance funds.

[0071] In one embodiment, based on prescription composition data, compatibility analysis is performed through a pre-built Traditional Chinese Medicine knowledge graph to obtain compatibility taboo identification results and empirical prescription compliance identification results, including:

[0072] S211, sending the prescription composition data to the constructed traditional Chinese medicine knowledge graph, which includes a compatibility taboo rule base and an empirical prescription database.

[0073] Optionally, the TCM knowledge graph is a structured knowledge base that integrates professional knowledge in the field of TCM, including a compatibility contraindication rule base and an empirical prescription database; the compatibility contraindication rule base records possible interaction contraindications between drugs, while the empirical prescription database stores verified TCM prescriptions and their application scenarios.

[0074] The collected prescription composition data is sent to the pre-built Traditional Chinese Medicine knowledge graph. The above process involves the adaptation and transmission of data formats to ensure that the prescription data can be correctly identified and processed by the rule base and database in the knowledge graph, providing accurate data input for subsequent analysis steps.

[0075] S212: Match the drug combinations in the prescription composition data with the incompatibility rule library to generate an incompatibility identification result.

[0076] Specifically, contraindication rule matching refers to comparing the drug combinations in the prescription one by one according to the preset rules in the incompatibility rule library to identify whether there are any drug incompatibilities that may affect the therapeutic effect or safety; the incompatibility identification result is the output of the above matching process, which clearly indicates whether there are any contraindications in the prescription.

[0077] Call the incompatibility rule library to conduct a detailed matching analysis of the drug combinations in the prescription composition data, check the interactions between drugs one by one through the algorithm, determine whether there are incompatibility contraindications based on the contraindication conditions in the rule library, and generate corresponding identification results to provide key information for subsequent risk assessment.

[0078] S213, based on the experience prescription database, analyze the matching degree between the prescription in the prescription composition data and the historical experience prescription through a similarity calculation model, and generate an experience prescription compliance identification result.

[0079] Exemplarily, the similarity calculation model is an algorithmic model used to evaluate the degree of similarity between the current prescription and the historical experience prescription; the experience prescription compliance identification result is based on the output of the above model, reflecting the matching degree between the prescription and the verified experience prescription, thereby judging whether the prescription complies with the diagnosis and treatment standards and practical experience of traditional Chinese medicine.

[0080] Using historical data in the empirical prescription database, the prescription composition data is analyzed through a similarity calculation model. The model calculates the similarity between the current prescription and the empirical prescription in the database, judges the compliance of the prescription based on the preset threshold, and generates the empirical prescription compliance identification result, providing data support for the rationality evaluation of the prescription.

[0081] In the above-mentioned TCM medical insurance reimbursement process management method, by matching and analyzing the prescription composition data with the incompatibility rule library and empirical prescription database in the TCM knowledge graph, it is possible to accurately identify the incompatibility in the prescription and evaluate the matching degree between the prescription and the empirical prescription, thereby providing a scientific basis for medical insurance reimbursement decision-making and ensuring the rationality and safety of reimbursement.

[0082] In one embodiment, based on the dosage usage data and the diagnosis and treatment item data, a dosage deviation score and a treatment suitability identification result are generated by calculation and analysis using a pre-stored database of TCM clinical pathway specifications, including:

[0083] S311: Based on the dosage usage data, use the following formula to extract, calculate and analyze the single-herb dosage and medication course data to obtain the dosage deviation score:

[0084]

[0085] Where D is the dose deviation score, n is the number of medicinal ingredients in the prescription, ω i is the dosage weight coefficient of the i-th drug, d i is the actual dosage, d im It is the median of the standard dose range recorded in the clinical pathway specification database.

[0086] Specifically, the actual dosage and course of treatment for each herb are extracted from the dosage usage data, while the median of the standard dosage range for the corresponding herb is obtained from a pre-stored database of Traditional Chinese Medicine clinical pathway specifications. Based on the preset dosage weight coefficient, the dosage deviation of each herb is calculated using the above formula, and the results are weighted and summed to generate a dosage deviation score. This score intuitively reflects the degree of deviation between the prescribed dosage and the standard, providing an important basis for subsequent risk assessment.

[0087] Specifically, taking Traditional Chinese Medicine prescription A as an example:

[0088] Drug name <![CDATA[d i ]]> <![CDATA[d im ]]> <![CDATA[ω i ]]> Drug A 10 grams 8 grams 0.3 Drug B 15 grams 12 grams 0.4 Drug C 20 grams 18 grams 0.3

[0089] The calculation is as follows:

[0090] Dose deviation of drug A:

[0091] Dose deviation of drug B:

[0092] Dose deviation of drug C:

[0093] D = 7.5 + 10 + 3.33 = 20.83;

[0094] Therefore, the dosage deviation score of this prescription is 20.83. The score shows that there is a certain deviation between the dosage of the overall prescription and the median of the standard dosage range, but the degree of deviation is within an acceptable range. The rationality and risk level of the prescription are further evaluated based on the given score to provide data support for medical insurance reimbursement decisions.

[0095] In the above-mentioned TCM medical insurance reimbursement process management method, by calculating the dosage deviation score based on dosage usage data, it is possible to quantitatively evaluate the degree of deviation between the drug dosage in the prescription and the clinical pathway standard, thereby providing a scientific basis for the rationality of dosage for medical insurance reimbursement decisions, effectively assisting in identifying possible problems such as excessive or insufficient dosage, and improving the accuracy and standardization of reimbursement review.

[0096] In one embodiment, based on the results of the incompatibility identification, the empirical prescription compliance identification, the dosage deviation score, and the therapy suitability identification, a comprehensive risk score is generated through dynamic risk assessment model analysis, including:

[0097] S411, convert the incompatibility identification results, empirical prescription compliance identification results, dosage deviation scores and therapy suitability identification results into standardized feature vectors.

[0098] Optionally, normalizing feature vectors refers to the process of converting data from different sources and dimensions into a unified format for easy processing by machine learning models, ensuring that each feature is comparable and consistent when input to the model, thereby improving the accuracy and stability of the model.

[0099] The results of incompatibility identification, empirical prescription compliance identification, dosage deviation score and therapy suitability identification are normalized and converted into standardized feature vectors. The above conversion process is implemented through mathematical formulas or algorithms to ensure that each feature value is within the same dimensional range, providing unified format input data for the subsequent dynamic risk assessment model.

[0100] S412: Send the standardized feature vector to the dynamic risk assessment model trained based on the XGBoost algorithm to generate a comprehensive risk score; the comprehensive risk score system is 0-100 points; the comprehensive risk score is calculated using the following formula:

[0101] C=α·TC+β·D+γ·(1-TM)+δ·EC+η·log(1+H)

[0102] Among them, C is the comprehensive risk score, TC is the incompatibility identification result, D is the individual dose deviation score, TM is the therapy matching, EC is the empirical prescription compliance identification score, H is the patient's historical reimbursement violation times, and α, β, γ, δ, and η are dynamic weight parameters; the dynamic weight parameters are actively learned by the XGBoost algorithm during the training process by minimizing the loss function between the prediction error of historical data and the actual violation label.

[0103] Specifically, the dynamic risk assessment model is a model built based on a machine learning algorithm that can predict the risk level based on the input feature vector; the XGBoost algorithm is an efficient gradient boosting algorithm that optimizes model parameters by minimizing the loss function between the prediction error and the actual violation label, thereby realizing dynamic risk assessment.

[0104] The standardized feature vector is sent to the dynamic risk assessment model trained based on the XGBoost algorithm. The model generates a comprehensive risk score based on the dynamic weight parameters learned during training and the transmitted feature vector using a preset calculation formula. The score is expressed in a range of 0 to 100 points, reflecting the overall risk level of the reimbursement application.

[0105] S413, mapping the comprehensive risk score to low, medium, and high risk levels according to the preset risk level threshold interval to obtain a risk score level result.

[0106] Preferably, the risk level threshold interval refers to a pre-set range for dividing continuous risk scores into different risk levels. By mapping the comprehensive risk score to the above interval, the risk can be quantified into three levels: low, medium and high, which is convenient for management and decision-making.

[0107] Based on the preset risk level threshold range, the generated comprehensive risk score is mapped to low, medium, and high risk levels. This mapping process is achieved through simple conditional judgment, converting continuous risk scores into discrete risk levels, providing intuitive risk assessment results for subsequent reimbursement decisions.

[0108] In the above-mentioned TCM medical insurance reimbursement process management method, by converting multiple risk assessment indicators into standardized feature vectors and sending them to a dynamic risk assessment model trained based on the XGBoost algorithm, it is possible to accurately calculate the comprehensive risk score and map the score to a risk level based on a preset threshold, thereby providing a scientific basis for medical insurance reimbursement decisions, effectively improving the accuracy and efficiency of decisions, and enhancing risk management capabilities.

[0109] In one embodiment, the comprehensive risk score is mapped to low, medium, and high risk levels according to a preset risk level threshold interval, and the risk score level result is obtained, including:

[0110] S511, based on the comprehensive risk score, normalization is performed using a preset logistic regression function to generate a continuity risk index; the calculation formula for the continuity risk index is:

[0111]

[0112] Among them, S is the continuity risk index, β is the parameter for adjusting the slope of the curve, and μ is the risk level division threshold.

[0113] For example, the continuous risk index is a value between 0 and 1 obtained by standardizing the comprehensive risk score, which is used to more smoothly represent the changes in risk levels; the logistic regression function is a commonly used S-shaped curve function that can map the input value to the interval (0,1), where the parameter β is used to adjust the steepness of the curve.

[0114] The comprehensive risk score is substituted into the logistic regression function model, and the score is nonlinearly transformed through the parameters β and μ, so that the original score is mapped into a continuous risk index that conforms to the S-shaped curve distribution. This process realizes the standardization of batch data through matrix operations, providing a unified measurement space for the subsequent discretization of risk levels.

[0115] S512 , dynamically calibrating the initially preset risk level threshold interval based on the distribution characteristics of the historical risk data to generate an optimized preset risk level threshold interval.

[0116] Specifically, the risk level threshold range is used to classify risk scores into different risk levels. Dynamic calibration refers to adjusting the initially preset threshold range based on the distribution characteristics of historical risk data to more accurately reflect the actual risk distribution and improve the accuracy of risk assessment.

[0117] Perform descriptive statistical analysis on historical risk score data, calculate quantiles (such as 25% and 75% quantiles) or fit the distribution curve through kernel density estimation, and adjust the initial threshold according to the skewness and peak characteristics of the data distribution. For example, set the threshold at the critical point where the risk probability suddenly changes, so that the sample size distribution of each risk level is more in line with the actual business scenario.

[0118] S513: Based on the optimized preset risk level threshold interval, the continuous risk index is divided into intervals to generate corresponding discrete risk level results.

[0119] Optionally, the discrete risk level results are generated by categorizing the continuous risk index into three risk levels: low, medium, and high, based on pre-set thresholds. This discretization facilitates quick understanding and decision-making in practical applications, simplifying complex continuous risk assessment results into intuitive level representations.

[0120] The standardized continuous risk index is numerically compared with the optimized threshold interval, and the index value is mapped to the corresponding risk level interval through conditional judgment logic (such as if-else rules) to generate a discretized risk level result. This process can achieve rapid classification of batch data through vector operations.

[0121] In the above-mentioned TCM medical insurance reimbursement process management method, a continuous risk index is generated by standardizing the comprehensive risk score, and the risk level threshold interval is dynamically calibrated based on historical data, and then the continuous risk index is divided into discrete risk levels. This can achieve accurate and dynamically adaptive risk assessment, provide a more scientific, accurate and easy-to-understand risk level division basis for medical insurance reimbursement decisions, and effectively improve decision-making efficiency and risk management capabilities.

[0122] In one embodiment, the training process of the dynamic risk assessment model includes:

[0123] S611, construct a training dataset based on compliance cases and non-compliance cases in historical TCM reimbursement data.

[0124] Among them, historical TCM reimbursement data includes previous medical records, prescription information and reimbursement review results. Compliance cases refer to normal reimbursement records that have passed the review, and violation cases refer to abnormal records with problems such as incompatibility and excessive medication. The training data set is a set of labeled data used for model learning.

[0125] Historical reimbursement data is extracted from the medical information system, outliers and missing values ​​are eliminated through data cleaning, compliance and non-compliance cases are marked according to preset rules, and a training dataset containing multi-dimensional features and corresponding labels is constructed to provide basic data support for model training.

[0126] S612: Based on the training data set, extract the incompatibility feature, the empirical prescription matching feature, the dosage deviation feature, and the therapy adaptability feature in each case in the training data set to obtain a feature data set.

[0127] Preferably, the incompatibility feature refers to the quantitative value of the degree to which the drug combination in the prescription violates the theory of traditional Chinese medicine; the empirical prescription matching feature is the similarity score between the prescription and the classic empirical prescription; the dosage deviation feature is the deviation ratio between the actual dosage and the standard dosage; the therapy adaptability feature reflects the fit between the diagnosis and treatment items and the symptoms.

[0128] For each case in the training dataset, the prescription text is parsed using natural language processing technology to extract the drug composition, and the compatibility score is calculated by comparing the knowledge graph; the similarity with the empirical prescription database is calculated based on the vector space model; the dosage deviation is quantified through the clinical pathway specification library; the therapy adaptability is evaluated in combination with the diagnosis and treatment guidelines, and finally integrated to form a multidimensional feature vector set.

[0129] S613, based on the feature data set, the dynamic risk assessment model is trained by the XGBoost algorithm to obtain a trained dynamic risk assessment model; wherein the dynamic risk assessment model is an ensemble learning model based on a gradient boosting tree.

[0130] Specifically, the XGBoost algorithm is an optimized distributed gradient boosting framework. The dynamic risk assessment model is an integrated learning model built based on this algorithm. The gradient boosting tree iteratively trains multiple weak classifiers and integrates their results to form a strong classifier.

[0131] The feature dataset is divided into a training set and a validation set, and the XGBoost model parameters (such as tree depth and learning rate) are configured. The decision tree model is iteratively trained using the gradient boosting algorithm to minimize the loss function between the predicted risk score and the actual label, ultimately generating an integrated model that can fuse multiple features for risk prediction.

[0132] S614: Based on the trained dynamic risk assessment model, the model is verified using a validation set, and model parameters are adjusted according to the verification results to obtain an optimized dynamic risk assessment model.

[0133] For example, the validation set is test data independent of the training set, which is used to evaluate the generalization ability of the model. The model parameters refer to the hyperparameters that affect the model performance (such as the number of iterations and the regularization coefficient). The optimized dynamic risk assessment model is a model with higher prediction accuracy after parameter tuning.

[0134] Use the validation set to evaluate the performance of the trained model, calculate indicators such as precision, recall, and area under the ROC curve, adjust model parameters using methods such as grid search or Bayesian optimization, and ensure the stability of parameter optimization through cross-validation, ultimately obtaining the risk assessment model that performs best on the validation set.

[0135] In the above-mentioned TCM medical insurance reimbursement process management method, by constructing a training data set containing compliance and non-compliance cases, it is possible to achieve an accurate quantitative assessment of TCM reimbursement risks.

[0136] In one embodiment, Figure 2 As shown, another method for managing the medical insurance reimbursement process for traditional Chinese medicine is provided. According to the preset risk level threshold range, the comprehensive risk score is mapped into low, medium and high risk levels. After obtaining the risk score level result, it also includes:

[0137] S711, when the risk level result is high risk, the reimbursement decision instruction generated is an electronic review work order containing a risk point identifier; the electronic review work order is used to remind manual review processing.

[0138] Optionally, a high-risk level result corresponds to a situation where the comprehensive risk score exceeds a preset threshold. The electronic review work order is a to-do task generated by the medical insurance reimbursement management terminal that contains risk point details. The risk point identifier refers to the location information of the specific violation (such as incompatibility and excessive medication). When the risk level is determined to be high risk, the medical insurance reimbursement management terminal automatically generates a formatted electronic work order, extracts key violation information from the risk characteristics (such as the name of the drug involved in the contraindication and the dosage value that deviates from the standard), attaches it to the work order, and pushes it to the manual review queue. At the same time, it records the work order generation timestamp and the relevant diagnosis and treatment information index.

[0139] S712, when the risk level result is medium risk, the reimbursement decision instruction generated is to dynamically adjust the reimbursement ratio based on the preset attenuation rule, and generate reimbursement result data with risk tags.

[0140] Specifically, the medium-risk level result refers to the interval where the risk score is between the high and low thresholds. The attenuation rule is the calculation logic that dynamically reduces the reimbursement ratio according to the degree of risk. The risk mark is a warning sign attached to the reimbursement result.

[0141] For medium-risk cases, the medical insurance reimbursement management terminal calls a preset attenuation function (such as linear attenuation or exponential attenuation), calculates the reimbursement ratio that should be reduced based on the risk index, and generates result data containing the adjusted reimbursement amount. At the same time, a risk level mark and a brief violation description are added to the reimbursement record for subsequent audit tracking.

[0142] S713: When the risk level result is low risk, the reimbursement decision instruction generated is an electronic confirmation voucher.

[0143] For example, a low risk level result indicates that the risk score is below a safe range of a preset threshold, and the electronic confirmation certificate is a reimbursement approval certificate automatically generated by the medical insurance reimbursement management terminal.

[0144] When the risk level is determined to be low risk, the medical insurance reimbursement management terminal automatically generates an electronic confirmation certificate containing key data such as patient information, reimbursement amount, medical treatment items, etc., and stores it in the database after being stamped with an electronic signature. The reimbursement result is notified to the patient or medical institution through a message push mechanism, and the relevant business status fields are updated at the same time.

[0145] In the above-mentioned TCM medical insurance reimbursement process management method, corresponding medical insurance reimbursement process processing is carried out according to the high, medium and low risk level results, which can realize differentiated intelligent management and control of the medical insurance reimbursement process, while ensuring the security of the fund, improving the audit efficiency and optimizing the level of management refinement.

[0146] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0147] Based on the same inventive concept, the embodiments of the present application also provide a device for implementing the aforementioned method for managing the process of medical insurance reimbursement for traditional Chinese medicine. The implementation solution provided by the device is similar to the implementation solution described in the aforementioned method. Therefore, the specific limitations in one or more embodiments of the device for managing the process of medical insurance reimbursement for traditional Chinese medicine provided below can be found in the above-mentioned limitations of the method for managing the process of medical insurance reimbursement for traditional Chinese medicine, and will not be repeated here.

[0148] In an exemplary embodiment, Figure 3 As shown, a TCM medical insurance reimbursement process management device 10 is provided, comprising:

[0149] The data acquisition module 11 is used to obtain the TCM diagnosis and treatment characteristic data of the target patient, which includes prescription composition data, dosage usage data and diagnosis and treatment item data;

[0150] The compatibility analysis module 12 is used to perform compatibility analysis based on prescription composition data through a pre-built traditional Chinese medicine knowledge graph to obtain compatibility taboo identification results and empirical prescription compliance identification results;

[0151] The pathway specification calculation and analysis module 13 is used to calculate and analyze the dosage usage data and the diagnosis and treatment item data through the pre-stored traditional Chinese medicine clinical pathway specification database to generate a dosage deviation score and a treatment suitability identification result;

[0152] A dynamic risk assessment module 14 is configured to generate a risk score level result through dynamic risk assessment model analysis based on the incompatibility identification result, the empirical prescription compliance identification result, the dosage deviation score, and the therapy suitability identification result;

[0153] The reimbursement decision generating module 15 is used to generate corresponding reimbursement decision instructions based on the risk scoring result.

[0154] In one embodiment, the compatibility analysis module is further configured to:

[0155] Send the prescription composition data to the constructed Traditional Chinese Medicine knowledge graph, which includes a compatibility taboo rule base and an empirical prescription database;

[0156] According to the incompatibility rule library, the drug combinations in the prescription composition data are matched with the incompatibility rules to generate the incompatibility identification results;

[0157] Based on the empirical prescription database, the matching degree between the prescriptions in the prescription composition data and the historical empirical prescriptions is analyzed through a similarity calculation model to generate the empirical prescription compliance identification results.

[0158] In one embodiment, the path specification calculation and analysis module is further configured to:

[0159] Based on the dosage usage data, the following formula is used to extract, calculate and analyze the single-drug dosage and medication course data to obtain the dosage deviation score:

[0160]

[0161] Where D is the dose deviation score, n is the number of medicinal ingredients in the prescription, ω i is the dosage weight coefficient of the i-th drug, d i is the actual dosage, d im It is the median of the standard dose range recorded in the clinical pathway specification database.

[0162] In one embodiment, the dynamic risk assessment module is further configured to:

[0163] The results of incompatibility identification, empirical prescription compliance identification, dosage deviation score and therapy suitability identification were converted into standardized feature vectors;

[0164] The standardized feature vector is sent to the dynamic risk assessment model trained with the XGBoost algorithm to generate a comprehensive risk score. The comprehensive risk score system is 0-100 points. The comprehensive risk score is calculated using the following formula:

[0165] C=α·TC+β·D+γ·(1-TM)+δ·EC+η·log(1+H)

[0166] Among them, C is the comprehensive risk score, TC is the incompatibility labeling result, D is the individual dose deviation score, TM is the therapy matching degree, EC is the empirical prescription compliance labeling score, H is the patient's historical reimbursement violation count, and α, β, γ, δ, and η are dynamic weight parameters. The dynamic weight parameters are actively learned by the XGBoost algorithm during the training process by minimizing the loss function between the prediction error of historical data and the actual violation label.

[0167] According to the preset risk level threshold range, the comprehensive risk score is mapped to low, medium and high risk levels to obtain the risk score level result.

[0168] In one embodiment, the dynamic risk assessment module is further configured to:

[0169] Based on the comprehensive risk score, the continuity risk index is generated by normalization through a preset logistic regression function. The calculation formula for the continuity risk index is:

[0170]

[0171] Among them, S is the continuity risk index, β is the parameter for adjusting the slope of the curve, and μ is the risk level classification threshold;

[0172] Based on the distribution characteristics of historical risk data, the initial preset risk level threshold interval is dynamically calibrated to generate an optimized preset risk level threshold interval;

[0173] Based on the optimized preset risk level threshold interval, the continuous risk index is divided into intervals to generate the corresponding discrete risk level results.

[0174] In one embodiment, the dynamic risk assessment module is further configured to:

[0175] Construct a training dataset based on compliance cases and non-compliance cases in historical TCM reimbursement data;

[0176] Based on the training data set, the incompatibility features, empirical prescription matching features, dosage deviation features, and therapy adaptability features in each case of the training data set are extracted to obtain a feature data set;

[0177] Based on the feature data set, the dynamic risk assessment model is trained using the XGBoost algorithm to obtain a trained dynamic risk assessment model; wherein the dynamic risk assessment model is an ensemble learning model based on the gradient boosting tree;

[0178] The trained dynamic risk assessment model is verified through the validation set, and the model parameters are adjusted according to the verification results to obtain an optimized dynamic risk assessment model.

[0179] In one embodiment, the reimbursement decision generation module is further configured to:

[0180] When the risk level result is high risk, the generated reimbursement decision instruction is an electronic review work order containing the risk point identification; the electronic review work order is used to remind manual review and processing;

[0181] When the risk level result is medium risk, the reimbursement decision instruction generated is to dynamically adjust the reimbursement ratio based on the preset attenuation rules and generate reimbursement result data with risk tags;

[0182] When the risk level result is low risk, the reimbursement decision instruction generated is an electronic confirmation voucher.

[0183] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the method for managing the medical insurance reimbursement process of traditional Chinese medicine as described above are implemented.

[0184] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0185] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial description of the method embodiments. The device embodiments described above are merely illustrative, wherein the components described as separate parts may or may not be physically separated, and the parts displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the disclosed solution. A person of ordinary skill in the art can understand and implement it without expending creative work.

[0186] The above-described embodiments merely represent several implementation methods of the embodiments of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that a person skilled in the art may make various modifications and improvements without departing from the concept of the embodiments of the present application, and these modifications and improvements fall within the scope of protection of the embodiments of the present application.

Claims

1. A method for managing the process of medical insurance reimbursement for traditional Chinese medicine, characterized in that: The method comprises: Obtaining TCM diagnosis and treatment characteristic data of the target patient, wherein the TCM diagnosis and treatment characteristic data includes prescription composition data, dosage usage data, and diagnosis and treatment item data; Based on the prescription composition data, a compatibility analysis is performed through a pre-built traditional Chinese medicine knowledge graph to obtain the compatibility taboo identification results and the empirical prescription compliance identification results; Based on the dosage usage data and diagnosis and treatment item data, a dosage deviation score and a treatment suitability identification result are generated by calculation and analysis using a pre-stored database of TCM clinical pathway specifications; Based on the results of the incompatibility labeling, the empirical prescription compliance labeling, the dosage deviation score, and the therapy suitability labeling, a dynamic risk assessment model is used to analyze and generate a risk score level result; the risk assessment model is a machine learning model trained based on violation cases in historical TCM reimbursement data; Based on the risk scoring result, a corresponding reimbursement decision instruction is generated; the reimbursement decision instruction includes a risk level, which is used to characterize the reimbursement process.

2. The method according to claim 1, characterized in that Based on the prescription composition data, the compatibility analysis is performed through a pre-built traditional Chinese medicine knowledge graph to obtain the compatibility taboo identification results and the empirical prescription compliance identification results, including: Sending the prescription composition data to a constructed traditional Chinese medicine knowledge graph, wherein the traditional Chinese medicine knowledge graph includes a compatibility taboo rule base and an empirical prescription database; According to the incompatibility rule library, the drug combination in the prescription composition data is matched with the incompatibility rule to generate the incompatibility identification result; Based on the experience prescription database, the matching degree between the prescription in the prescription composition data and the historical experience prescription is analyzed by a similarity calculation model to generate the experience prescription compliance identification result.

3. The method according to claim 1, characterized in that The method generates a dose deviation score and a therapy suitability identification result based on the dosage usage data and the diagnosis and treatment item data by calculating and analyzing the pre-stored TCM clinical pathway specification database, including: Based on the dosage usage data, the following formula is used to extract, calculate and analyze the single-herb dosage and medication course data to obtain the dosage deviation score: Where D is the dose deviation score, n is the number of medicinal ingredients in the prescription, ω i is the dosage weight coefficient of the i-th drug, d i is the actual dosage, d im It is the median of the standard dose range recorded in the clinical pathway specification database.

4. The method according to claim 1, wherein The risk score level results are generated by analyzing the dynamic risk assessment model based on the incompatibility identification results, the empirical prescription compliance identification results, the dosage deviation score and the therapy suitability identification results, including: Converting the incompatibility identification results, empirical prescription compliance identification results, dosage deviation scores, and therapy suitability identification results into standardized feature vectors; The standardized feature vector is sent to a dynamic risk assessment model trained based on the XGBoost algorithm to generate a comprehensive risk score; the comprehensive risk score system is 0-100 points; the comprehensive risk score is obtained by the following calculation formula: C=α·TC+β·D+γ·(1-TM)+δ·EC+η·log(1+H) Where C is the comprehensive risk score, TC is the incompatibility labeling result, D is the individual dose deviation score, TM is the therapy matching degree, EC is the empirical prescription compliance labeling score, H is the patient's historical reimbursement violation count, and α, β, γ, δ, and η are dynamic weight parameters. These dynamic weight parameters are actively learned by the XGBoost algorithm during training by minimizing the loss function between the prediction error of historical data and the actual violation label. According to the preset risk level threshold range, the comprehensive risk score is mapped into low, medium and high risk levels to obtain a risk score level result.

5. The method according to claim 4, characterized in that Mapping the comprehensive risk score to low, medium, and high risk levels according to a preset risk level threshold interval to obtain the risk score level result includes: Based on the comprehensive risk score, a standardized process is performed using a preset logistic regression function to generate a continuity risk index. The calculation formula for the continuity risk index is: Among them, S is the continuity risk index, β is the parameter for adjusting the slope of the curve, and μ is the risk level classification threshold; Based on the distribution characteristics of historical risk data, the initial preset risk level threshold interval is dynamically calibrated to generate an optimized preset risk level threshold interval; Based on the optimized preset risk level threshold interval, the continuous risk index is divided into intervals to generate corresponding discrete risk level results.

6. The method according to claim 4, characterized in that The training process of the dynamic risk assessment model includes: Construct a training dataset based on compliance cases and non-compliance cases in historical TCM reimbursement data; Based on the training data set, extracting the incompatibility feature, the empirical prescription matching feature, the dosage deviation feature, and the therapy adaptability feature in each case of the training data set to obtain a feature data set; According to the feature data set, the dynamic risk assessment model is trained by the XGBoost algorithm to obtain a trained dynamic risk assessment model; wherein the dynamic risk assessment model is an ensemble learning model based on a gradient boosting tree; The trained dynamic risk assessment model is verified using a validation set, and model parameters are adjusted according to the verification results to obtain an optimized dynamic risk assessment model.

7. The method according to claim 1, characterized in that Generating a corresponding reimbursement decision instruction based on the risk scoring result includes: When the risk level result is high risk, the generated reimbursement decision instruction is an electronic review worksheet containing a risk point identifier; the electronic review worksheet is used to remind manual review processing; When the risk level result is medium risk, the reimbursement decision instruction generated is to dynamically adjust the reimbursement ratio based on a preset attenuation rule, and generate reimbursement result data with a risk mark; When the risk level result is low risk, the reimbursement decision instruction generated is an electronic confirmation certificate.

8. A TCM medical insurance reimbursement process management device, characterized in that: The device comprises: A data acquisition module is used to obtain the TCM diagnosis and treatment characteristic data of the target patient, wherein the TCM diagnosis and treatment characteristic data includes prescription composition data, dosage usage data, and diagnosis and treatment item data; A compatibility analysis module is used to perform compatibility analysis based on the prescription composition data through a pre-built traditional Chinese medicine knowledge graph to obtain compatibility taboo identification results and empirical prescription compliance identification results; A pathway specification calculation and analysis module is used to calculate and analyze the dosage usage data and diagnosis and treatment item data through a pre-stored traditional Chinese medicine clinical pathway specification database to generate a dosage deviation score and a therapy suitability identification result; A dynamic risk assessment module, configured to generate a comprehensive risk score based on the incompatibility identification results, the empirical prescription compliance identification results, the dosage deviation score, and the therapy suitability identification results through dynamic risk assessment model analysis; The reimbursement decision generating module is used to generate corresponding reimbursement decision instructions based on the risk scoring level result.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images