A compliance inspection method and system for medical data based on large models

Abstract

The present invention relates to the field of data processing technology, and in particular to a compliance inspection method and system for medical data based on a large model. The method extracts core target medical indicators from medical compliance inspection statements through a preset large language model, integrates multi-source medical basic information, basic inspection rules and target medical indicators through the preset large language model, automatically associates the target medical indicators with the field names of the initial medical data, generates targeted candidate inspection rules, improves the accuracy and applicability of the rules, extracts target medical data from the initial medical data, screens the target inspection rules from the candidate inspection rules, inputs the target medical data and the target inspection rules into the preset large language model, obtains a compliance inspection report, and achieves comprehensive consideration of resource consumption and detection effect, thereby improving the accuracy of data anomaly detection while reducing resource consumption costs.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a compliance checking method and system for medical data based on a large model. Background Art

[0002] In today's digital healthcare era, medical data contains a large amount of sensitive patient information, such as personal identity, health status, treatment records, etc. The compliant use and storage of medical data not only affects the privacy protection of patients, but also affects the standardized development of the medical industry. Therefore, compliance checks on medical data are of vital importance.

[0003] Existing methods for medical data compliance checks primarily rely on pre-setting a series of compliance check rules based on relevant medical regulations and industry standards. A rule engine then performs a one-by-one check of medical data based on these rules to determine whether the data complies with the rules. While this method can quickly and accurately check for simple, clear compliance requirements, and the results are highly interpretable, it suffers from limited flexibility. When encountering new regulatory requirements or complex business scenarios, rules must be manually added or modified, which is time-consuming and labor-intensive. Furthermore, it's difficult to develop comprehensive and accurate rules for unstructured or semi-structured data, resulting in poor compliance check accuracy and difficulty meeting data compliance requirements in medical scenarios.

[0004] Therefore, how to improve the accuracy of data anomaly detection has become an urgent problem to be solved. Summary of the Invention

[0005] In response to the above technical problems, in a first aspect, the present invention provides a compliance checking method for medical data based on a large model, the compliance checking method for medical data based on a large model comprising the following steps:

[0006] S1: Input the preset medical compliance inspection sentence into the preset large language model to obtain the target medical indicators for compliance inspection.

[0007] S2. Input the basic medical information corresponding to the initial medical data, the basic inspection rules in the preset inspection rule library, and the target medical indicators into the preset large language model to obtain the target basic information and candidate inspection rules corresponding to the target medical indicators. The basic medical information includes all field names corresponding to the initial medical data, the basic inspection rules are in the form of SQL statements, the target basic information includes several target field names to be checked, and each candidate inspection rule is used to check the data corresponding to the corresponding several target field names in the initial medical data.

[0008] S3, extracting target medical data from the initial medical data based on the target basic information.

[0009] S4, based on the target medical data, screen the target inspection rules from all candidate inspection rules.

[0010] S5: Input the target medical data and target inspection rules into the preset large language model to obtain the compliance inspection report corresponding to the initial medical data.

[0011] In a second aspect, the present invention provides a compliance checking system for medical data based on a large model, the compliance checking system for medical data based on a large model comprising:

[0012] The indicator acquisition module is used to input the preset medical compliance inspection statement into the preset large language model to obtain the target medical indicators for compliance inspection.

[0013] The data screening module is used to input the basic medical information corresponding to the initial medical data, the basic inspection rules in the preset inspection rule library, and the target medical indicators into the preset large language model, and obtain the target basic information and candidate inspection rules corresponding to the target medical indicators, wherein the basic medical information includes all field names corresponding to the initial medical data, the basic inspection rules are in the form of SQL statements, the target basic information includes several target field names to be checked, and each candidate inspection rule is used to check the data corresponding to the corresponding several target field names in the initial medical data.

[0014] The data extraction module is used to extract target medical data from the initial medical data based on the target basic information.

[0015] The rule screening module is used to screen the target inspection rules from all candidate inspection rules according to the target medical data.

[0016] The compliance inspection module is used to input the target medical data and target inspection rules into the preset large language model to obtain the compliance inspection report corresponding to the initial medical data.

[0017] The present invention has at least the following beneficial effects: through the natural language processing capability of a preset large language model, the core target medical indicators are quickly and accurately extracted from preset medical compliance inspection statements, clarifying the direction and focus for subsequent inspection work; through the preset large language model, multi-source medical basic information, basic inspection rules and target medical indicators are integrated, the target medical indicators are automatically associated with the field names of the initial medical data, and targeted candidate inspection rules are generated, thereby improving the accuracy and applicability of the rules; the target medical data is extracted from the initial medical data to avoid interference from irrelevant data, which can reduce the amount of data for subsequent inspections, improve inspection efficiency and the accuracy of inspection results; the target inspection rules are screened from all candidate inspection rules, the target medical data and the target inspection rules are input into the preset large language model, and the compliance inspection report corresponding to the initial medical data is obtained, thereby achieving comprehensive consideration of resource consumption and detection effects, reducing resource consumption costs while improving the accuracy of data anomaly detection, and facilitating medical data managers and decision makers to quickly understand the compliance status of the data and make corresponding decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 A schematic diagram of a compliance checking method for medical data based on a large model provided in the first embodiment of the present invention;

[0020] Figure 2 A module diagram of a compliance checking system for medical data based on a large model provided in the second embodiment of the present invention. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 making any creative efforts shall fall within the scope of protection of the present invention.

[0022] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It is understood that, where appropriate, the above-mentioned terms used to distinguish similar objects can be interchanged so that the present invention can also implement other embodiments other than the above-mentioned illustrated embodiments or described embodiments. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0023] Example 1

[0024] This embodiment provides a compliance inspection method for medical data based on a large model. Figure 1 As shown, the compliance checking method for medical data based on a large model includes the following steps:

[0025] S1: Input the preset medical compliance inspection sentence into the preset large language model to obtain the target medical indicators for compliance inspection.

[0026] Among them, the preset medical compliance inspection statements are pre-written statements based on medical regulations, industry standards and business needs, which describe the requirements for medical data compliance inspections. They are used to summarize the inspection objects and contents in this compliance inspection, and may include several statements such as "check whether the surgical process meets the prescribed requirements" and "confirm whether the use of drugs is within the prescribed range".

[0027] A preset large language model refers to an artificial intelligence model that has been trained with a large amount of text data, such as GPT models. It has strong language understanding and generation capabilities, and can perform semantic understanding and analysis on the input medical compliance inspection statements, identify the key medical information involved, and thus extract target medical indicators.

[0028] Target medical indicators are specific metrics extracted from pre-defined medical compliance check statements that are used to measure whether medical data is compliant. For example, in the aforementioned medical compliance check statement "Confirm whether drug usage is within the prescribed range," "drug usage" is a target medical indicator.

[0029] In a specific embodiment, a preset medical compliance check sentence and a first preset prompt word are input into a preset large language model to obtain target medical indicators for compliance checking.

[0030] The first preset prompt is used to guide the preset large language model to generate a specific result, namely, the target medical indicator for compliance inspection. The first preset prompt can be pre-set based on the actual needs of the implementer. For example, the first preset prompt may be "Based on the input content, provide key compliance indicators" or "Based on the input content, provide risk assessment indicators," etc., which are prompts for obtaining the target medical indicator.

[0031] As mentioned above, the intelligent analysis capabilities of the large language model are used to quickly and accurately extract target medical indicators from complex medical compliance inspection statements, avoiding the tediousness and accuracy issues of manual indicator extraction.

[0032] S2. Input the basic medical information corresponding to the initial medical data, the basic inspection rules in the preset inspection rule library, and the target medical indicators into the preset large language model to obtain the target basic information and candidate inspection rules corresponding to the target medical indicators. The basic medical information includes all field names corresponding to the initial medical data, the basic inspection rules are in the form of SQL statements, the target basic information includes several target field names to be checked, and each candidate inspection rule is used to check the data corresponding to the corresponding several target field names in the initial medical data.

[0033] Among them, initial medical data refers to medical data that needs to be inspected for compliance. It can come from the hospital's information system, medical equipment, etc., and includes various medical information such as the patient's basic information, diagnosis results, treatment records, etc.

[0034] Basic medical information includes all the field names corresponding to the original medical data. For example, basic medical information may include fields such as "patient name," "patient age," "medical record number," "diagnosis," "symptoms," "disease code," "treatment plan," "operation name," "operation record," "medication record," and "attending physician."

[0035] The default check rule library is a collection of pre-established rules based on medical regulations, industry standards, and business needs. It is used to perform compliance checks on medical data. The default check rule library includes several basic check rules, which use SQL statements to perform compliance checks on medical data.

[0036] A pre-set large language model performs comprehensive analysis and reasoning on basic medical information, basic inspection rules, and target medical indicators to identify target basic information and candidate inspection rules related to the target medical indicator. Target basic information refers to the names of several target fields to be inspected related to the target medical indicator. For example, if the target medical indicator is "drug usage," the target basic information might be the field name "medication record."

[0037] Candidate check rules are a set of check rules generated based on the target medical indicator, basic medical information, and basic check rules. Each candidate check rule is used to check the data corresponding to several target field names in the initial medical data. For example, for the target medical indicator "drug usage," candidate check rules may include rules that check whether the medication type, medication time, and medication dosage are correct.

[0038] In a specific embodiment, the basic medical information corresponding to the initial medical data, the basic inspection rules in the preset inspection rule library, the target medical indicators and the second preset prompt word are input into the preset large language model to obtain the target basic information and candidate inspection rules corresponding to the target medical indicators.

[0039] Among them, the second preset prompt word is used to guide the preset large language model to screen and generate target basic information and candidate inspection rules corresponding to the target medical indicators from the medical basic information corresponding to the input initial medical data, the basic inspection rules in the preset inspection rule library, and the target medical indicators.

[0040] The second preset prompt word can be pre-set according to the actual needs of the implementer. For example, the second preset prompt word can be "based on the input content, filter and output the target basic information and candidate inspection rules corresponding to the target medical indicators."

[0041] As described above, the target medical indicators are automatically associated with the field names of the initial medical data through the large language model to generate targeted candidate inspection rules, which improves the accuracy and applicability of the rules and improves their applicability to different medical data structures and compliance inspection requirements.

[0042] S3, extracting target medical data from the initial medical data based on the target basic information.

[0043] The target medical data refers to the specific data extracted from the initial medical data that corresponds to the target basic information. For example, if the target basic information is the field name "drug usage," the target medical data is all data related to the used drugs in the initial medical data.

[0044] As described above, extracting target medical data from initial medical data and avoiding interference from irrelevant data can reduce the amount of data for subsequent examinations, thereby improving examination efficiency and the accuracy of examination results.

[0045] S4, based on the target medical data, screen the target inspection rules from all candidate inspection rules.

[0046] Among them, each candidate inspection rule corresponds to one, two or more target field names, and is used to perform a comprehensive inspection on the data corresponding to the corresponding target field names. Each target field name can be subject to compliance inspection through one, two or more different candidate inspection rules. Therefore, after screening a number of candidate inspection rules from the preset inspection rule library in combination with basic medical information, different candidate inspection rules can perform compliance inspection on the data corresponding to the same target field name. When compliance inspection is performed according to all candidate inspection rules, there are redundant inspections, resulting in the consumption of computing resources, and thus the increase of inspection costs and time.

[0047] Therefore, this embodiment further screens the candidate inspection rules to obtain target inspection rules for performing compliance inspection on target medical data, thereby reducing inspection costs and improving inspection efficiency while ensuring the accuracy of compliance inspection.

[0048] In a specific embodiment, S4 includes the following steps:

[0049] S41, according to the several target field names corresponding to each candidate inspection rule, screen out several candidate inspection rule combinations from all candidate inspection rules, wherein each candidate inspection rule combination includes several candidate inspection rules, and each candidate inspection rule combination is used to inspect the target medical data.

[0050] S42, for any candidate inspection rule combination, obtain the spatiotemporal processing cost corresponding to the current candidate inspection rule combination based on the data volume corresponding to several target field names corresponding to each candidate inspection rule in the current candidate inspection rule combination and the spatiotemporal processing parameters corresponding to each candidate inspection rule.

[0051] S43 , obtaining an execution efficiency score corresponding to the current candidate inspection rule combination according to the data volume and execution efficiency parameters corresponding to several target field names corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0052] S44 , obtaining the priority corresponding to each candidate inspection rule combination according to the spatiotemporal processing cost and execution efficiency score corresponding to each candidate inspection rule combination.

[0053] S45 , determining several candidate inspection rules corresponding to the candidate inspection rule combination with the highest priority as target inspection rules.

[0054] Among them, each candidate inspection rule corresponds to several target field names. By analyzing the association and combination between the target field names, different rule combinations are screened out from all candidate inspection rules as the basis for screening out the target inspection rules.

[0055] The amount of data corresponding to each target field name is the amount of data corresponding to each target field name in the target medical data.

[0056] Spatiotemporal processing parameters are parameters related to the time and space costs involved in executing a candidate inspection rule. These parameters, such as the speed of data compliance checks and the storage space consumed, are used to evaluate the spatiotemporal processing costs of a candidate inspection rule combination. The spatiotemporal processing cost represents the overall time and space resource costs required to process the target medical data using the current candidate inspection rule combination.

[0057] Execution performance parameters are related to the performance of candidate inspection rule execution, such as query accuracy, recall, and stability. These parameters are used to calculate the execution performance score of each candidate inspection rule combination. The execution performance score represents the overall execution efficiency and accuracy of the current candidate inspection rule combination when processing target medical data.

[0058] The priority of each candidate check rule combination can be determined by weighted calculation, combining the temporal and spatial processing costs and the execution efficiency score. The specific weighting method can be adjusted according to actual needs and application scenarios.

[0059] Priority is used to indicate the relative importance and applicability of each candidate check rule combination among all other combinations. A higher priority indicates a better overall performance of the candidate check rule combination in terms of temporal and spatial processing costs and execution efficiency.

[0060] The candidate inspection rule combination with the highest priority has the best comprehensive performance in terms of time and space processing cost and execution efficiency. Selecting the candidate inspection rule corresponding to the candidate inspection rule combination with the highest priority as the target inspection rule can maximize the inspection efficiency and reduce resource consumption while ensuring the accuracy of the inspection.

[0061] As mentioned above, by calculating the spatiotemporal processing cost, the efficiency of each candidate inspection rule combination in resource utilization can be evaluated, and the execution efficiency score can reflect the efficiency and accuracy of each candidate inspection rule combination in processing data. Selecting a combination with lower spatiotemporal processing cost and higher execution efficiency score can reduce the burden on the system, avoid waste of resources, and improve inspection efficiency and the accuracy of inspection results.

[0062] In a specific embodiment, S41 includes the following steps:

[0063] S411: All candidate inspection rules are randomly combined to obtain several initial inspection rule combinations.

[0064] S412 , for any initial inspection rule combination, obtain a target field name set corresponding to the current initial inspection rule combination according to several target field names corresponding to each candidate inspection rule in the current initial inspection rule combination.

[0065] S413: If the target field name set corresponding to the current initial check rule combination includes all target field names to be checked and there is no duplicate target field name, the current initial check rule combination is determined as a candidate check rule combination.

[0066] S414, traverse all initial inspection rule combinations, and repeat steps S12 and S13 to obtain all candidate inspection rule combinations.

[0067] The initial inspection rule combination refers to a set of rules obtained by arbitrarily combining all candidate inspection rules, which provides a comprehensive range of options for subsequent screening of suitable candidate inspection rule combinations. Each initial inspection rule combination includes several candidate inspection rules.

[0068] The target field name set is a set consisting of the target field names corresponding to all candidate check rules in the current initial check rule combination. It can clearly understand all target fields involved in the current initial check rule combination, which is used for subsequent judgment of whether the current initial check rule combination meets the requirements.

[0069] The target field name set corresponding to the candidate inspection rule combination includes all target field names to be checked and there are no repeated target field names. That is, all candidate inspection rules in the candidate inspection rule combination can perform comprehensive inspections on the data corresponding to all target field names to be checked, and no repeated inspections will be performed, thereby ensuring the integrity of abnormal data detection and reducing detection costs.

[0070] As described above, by screening out the initial inspection rule combinations that meet the conditions as candidate inspection rule combinations, it can be ensured that the final candidate inspection rule combinations can comprehensively and efficiently inspect the target medical data, avoiding the problems of missing inspection fields or repeated inspections, and improving the accuracy and efficiency of the inspection.

[0071] In a specific embodiment, the spatiotemporal processing parameters include time processing parameters and space processing parameters, and S42 includes the following steps:

[0072] S421 , obtaining a first total amount of data corresponding to each candidate inspection rule in the current candidate inspection rule combination according to the amount of data corresponding to several target field names corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0073] S422 , obtaining a time processing cost corresponding to each candidate inspection rule in the current candidate inspection rule combination according to the first data volume and the time processing parameter corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0074] S423 , acquiring the temporal and spatial processing costs corresponding to each candidate inspection rule in the current candidate inspection rule combination according to the temporal processing costs and spatial processing parameters corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0075] Among them, the first total data volume refers to the sum of the data volumes contained in all target field names corresponding to all candidate inspection rules in the current candidate inspection rule combination, which is used to clarify the data scale that needs to be processed by each candidate inspection rule combination, and provide basic data for subsequent calculation of time processing costs and space processing costs.

[0076] The time processing parameter can be the speed at which the corresponding candidate inspection rule performs data anomaly detection. Based on the total amount of first data to be detected and the time processing parameter, the time it takes for the corresponding candidate inspection rule to process the data corresponding to the corresponding target field name can be calculated as the corresponding time processing cost.

[0077] The space processing parameter can be the amount of CPU, memory, disk, or other resources occupied by a unit of data volume processed by the candidate inspection rule, for example, the amount of disk space required to store a certain amount of data. Based on the total amount of first data to be inspected and the space processing parameter, the amount of resources occupied by the candidate inspection rule for the data corresponding to the corresponding target field name can be calculated as the corresponding space processing cost.

[0078] The time processing cost and space processing cost of each candidate inspection rule are comprehensively calculated. For example, the time and space processing costs corresponding to each candidate inspection rule are obtained by weighted summation, which reflects the processing cost of each candidate inspection rule during the execution process and serves as the basis for screening the candidate inspection rule combination.

[0079] As mentioned above, by calculating the time processing cost and space processing cost, we can quantify the time and resource size required for each candidate inspection rule to process data, which helps to evaluate the processing costs of different rule combinations in terms of time and space, and thus provide a basis for selecting a better inspection rule combination.

[0080] In a specific embodiment, the execution performance parameters include an accuracy parameter, a recall parameter, and a stability parameter. S43 includes the following steps:

[0081] S431, according to the accuracy parameter, recall parameter, stability parameter, the first preset weight corresponding to the accuracy parameter, the second preset weight corresponding to the recall parameter and the third preset weight corresponding to the stability parameter corresponding to each candidate inspection rule in the current candidate inspection rule combination, obtain the reference score corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0082] S432 , obtaining an execution efficiency score corresponding to the current candidate inspection rule combination according to the total amount of first data and the reference score corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0083] The accuracy parameter indicates the percentage of anomalies that a candidate check rule correctly detects when checking data, reflecting the accuracy of the candidate check rule's results. For example, if a candidate check rule correctly detects 80 out of 100 actual anomalies, its accuracy is 80%.

[0084] The recall parameter indicates the ratio of anomaly data detected by a candidate check rule to all actual anomaly data, reflecting the candidate check rule's ability to find all anomalies. For example, if there are 100 actual anomaly data and the candidate check rule detects 90, the recall rate is 90%.

[0085] The stability parameter measures the consistency and reliability of a candidate check rule over multiple executions. This can be quantified by observing the fluctuations in the corresponding check results after executing the candidate check rule multiple times. For example, if the deviations in the check results of a candidate check rule over multiple executions are within a small range, the stability is high.

[0086] The first preset weight, the second preset weight, and the third preset weight are used to reflect the relative importance of the accuracy parameter, the recall parameter, and the stability parameter in the comprehensive evaluation. The specific values can be set by the implementer according to the actual situation.

[0087] As described above, the accuracy parameters, recall parameters, and stability parameters of the candidate inspection rules are comprehensively used to fully reflect the effectiveness of each candidate inspection rule during execution, providing a data basis for screening candidate inspection rule combinations with higher effectiveness.

[0088] In a specific embodiment, S432 includes the following steps:

[0089] S4321, according to the total amount of first data corresponding to each candidate inspection rule in the current candidate inspection rule combination, obtain the total amount of second data corresponding to the current candidate inspection rule combination.

[0090] S4322: Determine the ratio of the first total amount of data corresponding to each candidate inspection rule in the current candidate inspection rule combination to the second total amount of data corresponding to the current candidate inspection rule combination as a reference weight corresponding to the current candidate inspection rule combination.

[0091] S4323: Obtain the execution efficiency score corresponding to the current candidate inspection rule combination based on the reference weight and reference score corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0092] Among them, the reference weight reflects the proportion of the data volume processed by each candidate inspection rule in the entire candidate inspection rule combination. The candidate inspection rule with a larger data volume proportion has a higher relative importance in the entire candidate inspection rule combination. Correspondingly, the greater the weight it occupies in the subsequent calculation of the execution efficiency score, so that the execution efficiency score can more accurately reflect the actual performance of the candidate inspection rule combination in processing large-scale data, avoiding ignoring the impact of data volume differences on the overall performance during evaluation, thereby providing a more reliable basis for selecting the optimal candidate inspection rule combination.

[0093] In a specific embodiment, S44 includes the following steps:

[0094] S441 , obtaining a cost threshold according to the spatiotemporal processing costs corresponding to all candidate inspection rule combinations.

[0095] S442: Obtain a first scoring threshold and a second scoring threshold according to the execution efficiency scores corresponding to all candidate inspection rule combinations.

[0096] S443 : For any candidate inspection rule combination, if the execution efficiency score corresponding to the current candidate inspection rule combination is greater than or equal to the second score threshold, the current candidate inspection rule combination is determined as an intermediate inspection rule combination.

[0097] S444: If the execution efficiency score corresponding to the current candidate inspection rule combination is greater than or equal to the first score threshold and less than the second score threshold, and the corresponding spatiotemporal processing cost is less than or equal to the cost threshold, the current candidate inspection rule combination is determined as the intermediate inspection rule combination.

[0098] S445, obtain the priority corresponding to each intermediate inspection rule combination based on the time-space processing cost, execution efficiency score, the fourth preset weight corresponding to the time-space processing cost and the fifth preset weight corresponding to the execution efficiency score corresponding to each intermediate inspection rule combination.

[0099] S446: If the execution efficiency score corresponding to the current candidate inspection rule combination is less than the first score threshold, or the corresponding spatiotemporal processing cost is greater than the cost threshold, the priority corresponding to the current candidate inspection rule combination is determined to be the lowest priority.

[0100] The cost threshold refers to the limit of the spatiotemporal processing cost for screening candidate inspection rule combinations. This threshold is obtained by analyzing the spatiotemporal processing costs of all candidate inspection rule combinations and is used to determine whether the spatiotemporal processing costs of a particular combination are within an acceptable range. The first and second scoring thresholds are the limit values used to categorize the execution performance levels of candidate inspection rule combinations. This threshold is obtained by analyzing the execution performance scores of all candidate inspection rule combinations and is used to determine the execution performance level of a particular combination. The first scoring threshold is lower than the second scoring threshold.

[0101] Candidate inspection rule combinations with execution efficiency scores greater than or equal to the second scoring threshold perform well in terms of execution effect. Correspondingly, some low-cost requirements can be sacrificed and the corresponding candidate inspection rule combinations can be screened as intermediate inspection rule combinations, which helps to give priority to combinations with better performance for subsequent priority calculations, thereby improving the accuracy of the detection results of the data to be detected.

[0102] The candidate inspection rule combinations whose execution efficiency scores are between the first and second scoring thresholds and whose spatiotemporal processing costs are less than or equal to the cost thresholds perform well in terms of execution effect and low cost. Screening the corresponding candidate inspection rule combinations as intermediate inspection rule combinations will help to comprehensively consider high-accuracy and low-cost combinations for data anomaly detection and achieve a balance between accuracy and cost.

[0103] By comprehensively considering the spatiotemporal processing costs and execution efficiency scores through weighted summation, a reasonable priority is quantified for each intermediate inspection rule combination, which facilitates the subsequent selection of the optimal combination for data anomaly detection.

[0104] For combinations with poor execution performance or too high spatiotemporal processing costs, their priority is set to the lowest, reducing the possibility of being selected for data anomaly detection and avoiding negative impacts on system performance and detection results.

[0105] As mentioned above, by setting a cost threshold and two scoring thresholds, the candidate inspection rule combinations are screened and evaluated in multiple dimensions, taking into account the two key factors of spatiotemporal processing cost and execution efficiency, to ensure that the final selected combination achieves a good balance between resource consumption and detection effect.

[0106] In a specific embodiment, S441 includes the following steps:

[0107] S4411, according to the spatiotemporal processing costs corresponding to all candidate inspection rule combinations, a cost distribution function is fitted.

[0108] S4412: Determine the spatiotemporal processing cost corresponding to the second quartile of the cost distribution function as a cost threshold.

[0109] Among them, the cost distribution function describes the distribution law of the spatiotemporal processing costs corresponding to all candidate inspection rule combinations. The second quartile is the median. Find the point where the cumulative distribution function value of the cost distribution function is 0.5. The spatiotemporal processing cost corresponding to this point is the cost threshold.

[0110] As described above, based on the distribution law of the spatiotemporal processing costs corresponding to all candidate inspection rule combinations, an appropriate cost threshold is selected so that the screened combinations are representative and reasonable in terms of spatiotemporal resource consumption, thereby improving the accuracy of screening the subsequent optimal candidate inspection rule combination.

[0111] In a specific embodiment, S442 includes the following steps:

[0112] S4421: According to the execution efficiency scores corresponding to all candidate inspection rule combinations, a score distribution function is fitted.

[0113] S4422: Determine the execution efficiency score corresponding to the second quartile of the score distribution function as a first score threshold.

[0114] S4423: Determine the execution efficiency score corresponding to the third quartile of the score distribution function as the second score threshold

[0115] Among them, the score distribution function reflects the distribution of the execution efficiency scores corresponding to all candidate inspection rule combinations, which helps to understand the distribution characteristics of the execution efficiency of different combinations in the overall distribution.

[0116] The second quartile of the score distribution function is the point where the cumulative distribution function value of the score distribution function is 0.5, and the third quartile of the score distribution function is the point where the cumulative distribution function value of the score distribution function is 0.75. The execution performance score corresponding to this point is found as the second score threshold.

[0117] As described above, based on the distribution pattern of the execution efficiency scores corresponding to all candidate inspection rule combinations, appropriate first scoring thresholds and second scoring thresholds are set, so that all candidate inspection rule combinations can be divided into different intervals according to execution efficiency, which facilitates the subsequent adoption of different processing strategies according to different intervals and more reasonably screens out the optimal candidate inspection rule combination.

[0118] In the above, by screening out candidate inspection rule combinations that can detect data corresponding to all target field names, comprehensive detection of various key parts of the data to be detected is ensured. Based on the data volume of the target field name corresponding to each candidate inspection rule and the spatiotemporal processing parameters, the consumption of time and space resources of each candidate inspection rule combination during execution is accurately evaluated, and based on the data volume and execution efficiency parameters of the target field name corresponding to the candidate inspection rule, the execution efficiency score of each candidate inspection rule combination is evaluated. Finally, the spatiotemporal processing cost and the execution efficiency score are combined to determine the priority of each candidate inspection rule combination, which helps to intuitively compare the performance of different combinations in detection accuracy, recall rate, stability, etc., so that in actual applications, according to resource conditions and performance requirements, a combination with low execution cost and excellent performance can be reasonably selected, so as to achieve comprehensive consideration of resource consumption and detection effect, and reduce resource consumption cost while improving the accuracy of data anomaly detection.

[0119] S5: Input the target medical data and target inspection rules into the preset large language model to obtain the compliance inspection report corresponding to the initial medical data.

[0120] Among them, after the target medical data and target inspection rules are input into the preset large language model, the large language model will understand the target inspection rules, clarify the specific inspection requirements and logic, and then conduct one-by-one inspections of the target medical data based on the target inspection rules. Specifically, during the inspection process, it will determine whether the data meets the requirements of the rules. Once data that does not meet the rules is found, it will be recorded. Finally, the large language model will organize and analyze the inspection results, predict potential risks and trends, and generate compliance inspection reports in the form of natural language, which greatly reduces the workload of manual review and report writing, improves work efficiency, and facilitates medical data managers and decision makers to quickly understand the compliance status of the data and make corresponding decisions.

[0121] In a specific embodiment, the target medical data, the target inspection rules and the third preset prompt word are input into a preset large language model to obtain a compliance inspection report corresponding to the initial medical data.

[0122] Among them, the third preset prompt word is used to guide the preset large language model to output a compliance inspection report corresponding to the initial medical data.

[0123] The third preset prompt word can be pre-set according to the actual needs of the implementer. For example, the third preset prompt word can be "perform a compliance check on the target medical data according to the input target inspection rules, generate and output a compliance inspection report corresponding to the initial medical data, wherein the data that does not comply with the rules is recorded in the compliance inspection report, and potential risks and trends are predicted."

[0124] As described above, through the natural language processing capabilities of the preset large language model, the core target medical indicators are quickly and accurately extracted from the preset medical compliance inspection statements, which clarifies the direction and focus for subsequent inspection work. The preset large language model integrates multi-source medical basic information, basic inspection rules and target medical indicators, automatically associates the target medical indicators with the field names of the initial medical data, generates targeted candidate inspection rules, improves the accuracy and applicability of the rules, extracts target medical data from the initial medical data, avoids interference from irrelevant data, reduces the amount of data for subsequent inspections, improves inspection efficiency and the accuracy of inspection results, screens target inspection rules from all candidate inspection rules, inputs the target medical data and target inspection rules into the preset large language model, and obtains the compliance inspection report corresponding to the initial medical data, realizing comprehensive consideration of resource consumption and detection effect, while improving the accuracy of data anomaly detection and reducing resource consumption costs, making it convenient for medical data managers and decision makers to quickly understand the compliance status of the data and make corresponding decisions.

[0125] Example 2

[0126] This second embodiment provides a compliance inspection system for medical data based on a large model, such as Figure 2 As shown, the compliance inspection system for medical data based on a large model includes:

[0127] The indicator acquisition module 21 is used to input the preset medical compliance inspection sentence into the preset large language model to obtain the target medical indicator for compliance inspection.

[0128] The data screening module 22 is used to input the basic medical information corresponding to the initial medical data, the basic inspection rules in the preset inspection rule library, and the target medical indicators into the preset large language model, and obtain the target basic information and candidate inspection rules corresponding to the target medical indicators, wherein the basic medical information includes all field names corresponding to the initial medical data, the basic inspection rules are in the form of SQL statements, the target basic information includes several target field names to be checked, and each candidate inspection rule is used to check the data corresponding to the corresponding several target field names in the initial medical data.

[0129] The data extraction module 23 is used to extract target medical data from the initial medical data according to the target basic information.

[0130] The rule screening module 24 is used to screen the target inspection rules from all candidate inspection rules according to the target medical data.

[0131] The compliance checking module 25 is used to input the target medical data and target checking rules into a preset large language model to obtain a compliance checking report corresponding to the initial medical data.

[0132] In a specific embodiment, the rule screening module 24 includes:

[0133] The rule screening submodule is used to screen out several candidate inspection rule combinations from all candidate inspection rules according to several target field names corresponding to each candidate inspection rule, wherein each candidate inspection rule combination includes several candidate inspection rules, and each candidate inspection rule combination is used to inspect the target medical data.

[0134] The spatiotemporal processing cost acquisition submodule is used to obtain the spatiotemporal processing cost corresponding to any candidate inspection rule combination based on the data volume corresponding to several target field names corresponding to each candidate inspection rule in the current candidate inspection rule combination and the spatiotemporal processing parameters corresponding to each candidate inspection rule.

[0135] The execution efficiency score acquisition submodule is used to obtain the execution efficiency score corresponding to the current candidate inspection rule combination based on the data volume and execution efficiency parameters corresponding to several target field names corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0136] The priority acquisition submodule is used to obtain the priority corresponding to each candidate inspection rule combination according to the time-space processing cost and execution efficiency score corresponding to each candidate inspection rule combination.

[0137] The target inspection rule acquisition submodule is used to determine several candidate inspection rules corresponding to the candidate inspection rule combination with the highest priority as target inspection rules.

[0138] In a specific embodiment, the rule screening submodule includes:

[0139] The rule combination unit is used to arbitrarily combine all candidate inspection rules to obtain a number of initial inspection rule combinations.

[0140] The target field name set acquisition unit is used to obtain the target field name set corresponding to the current initial inspection rule combination according to several target field names corresponding to each candidate inspection rule in the current initial inspection rule combination for any initial inspection rule combination.

[0141] The first candidate inspection rule acquisition unit is configured to determine the current initial inspection rule combination as a candidate inspection rule combination if the target field name set corresponding to the current initial inspection rule combination includes all target field names to be inspected and there are no duplicate target field names.

[0142] The second candidate inspection rule acquisition unit is used to traverse all initial inspection rule combinations and repeatedly execute the steps of the target field name set acquisition unit and the first candidate inspection rule acquisition unit to obtain all candidate inspection rule combinations.

[0143] In a specific embodiment, the spatiotemporal processing parameters include time processing parameters and space processing parameters, and the spatiotemporal processing cost acquisition submodule includes:

[0144] The first data total amount acquisition subunit is used to acquire the first data total amount corresponding to each candidate inspection rule in the current candidate inspection rule combination according to the data amount corresponding to several target field names corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0145] The time processing cost acquisition subunit is used to acquire the time processing cost corresponding to each candidate inspection rule in the current candidate inspection rule combination according to the first data volume and time processing parameter corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0146] The time-space processing cost acquisition subunit is used to acquire the time-space processing cost corresponding to each candidate inspection rule in the current candidate inspection rule combination according to the time processing cost and space processing parameters corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0147] In a specific embodiment, the execution performance parameters include accuracy parameters, recall parameters, and stability parameters, and the execution performance score acquisition submodule includes:

[0148] The reference score acquisition unit is used to obtain the reference score corresponding to each candidate inspection rule in the current candidate inspection rule combination based on the accuracy parameter, recall parameter, stability parameter, the first preset weight corresponding to the accuracy parameter, the second preset weight corresponding to the recall parameter and the third preset weight corresponding to the stability parameter corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0149] The execution efficiency score acquisition unit is used to acquire the execution efficiency score corresponding to the current candidate inspection rule combination according to the total amount of first data and the reference score corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0150] In a specific embodiment, the execution performance score acquisition unit includes:

[0151] The second data total amount obtaining subunit is configured to obtain the second data total amount corresponding to the current candidate inspection rule combination according to the first data total amount corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0152] The reference weight acquisition subunit is used to determine the ratio of the first total data corresponding to each candidate inspection rule in the current candidate inspection rule combination to the second total data corresponding to the current candidate inspection rule combination as the reference weight corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0153] The execution efficiency score acquisition subunit is used to acquire the execution efficiency score corresponding to the current candidate inspection rule combination based on the reference weight and reference score corresponding to each candidate inspection rule in the current candidate inspection rule combination.

[0154] In a specific embodiment, the priority acquisition submodule includes:

[0155] The cost threshold acquisition unit is used to acquire a cost threshold according to the spatiotemporal processing costs corresponding to all candidate inspection rule combinations.

[0156] The scoring threshold acquisition unit is configured to acquire a first scoring threshold and a second scoring threshold according to the execution efficiency scores corresponding to all candidate inspection rule combinations.

[0157] The first intermediate inspection rule combination acquisition unit is configured to determine, for any candidate inspection rule combination, the current candidate inspection rule combination as the intermediate inspection rule combination if the execution efficiency score corresponding to the current candidate inspection rule combination is greater than or equal to the second score threshold.

[0158] The second intermediate inspection rule combination acquisition unit is used to determine the current candidate inspection rule combination as an intermediate inspection rule combination if the execution efficiency score corresponding to the current candidate inspection rule combination is greater than or equal to the first score threshold and less than the second score threshold, and the corresponding space-time processing cost is less than or equal to the cost threshold.

[0159] The first priority acquisition unit is used to obtain the priority corresponding to each intermediate inspection rule combination based on the time-space processing cost corresponding to each intermediate inspection rule combination, the execution efficiency score, the fourth preset weight corresponding to the time-space processing cost, and the fifth preset weight corresponding to the execution efficiency score.

[0160] The second priority acquisition unit is used to determine that the priority corresponding to the current candidate inspection rule combination is the lowest priority if the execution efficiency score corresponding to the current candidate inspection rule combination is less than the first score threshold, or the corresponding spatiotemporal processing cost is greater than the cost threshold.

[0161] In a specific embodiment, the cost threshold acquisition unit includes:

[0162] The cost distribution function acquisition subunit is used to fit the cost distribution function according to the spatiotemporal processing costs corresponding to all candidate inspection rule combinations.

[0163] The cost threshold acquisition subunit is used to determine the spatiotemporal processing cost corresponding to the second quartile of the cost distribution function as the cost threshold.

[0164] In a specific embodiment, the scoring threshold acquisition unit includes:

[0165] The score distribution function acquisition subunit is used to fit the score distribution function according to the execution efficiency scores corresponding to all candidate inspection rule combinations.

[0166] The first score threshold acquisition subunit is configured to determine the execution efficiency score corresponding to the second quartile of the score distribution function as the first score threshold.

[0167] The second scoring threshold obtaining subunit is configured to determine the execution efficiency score corresponding to the third quartile of the scoring distribution function as the second scoring threshold.

[0168] Although some specific embodiments of the present invention have been described in detail by way of example, it should be understood by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present invention. It should also be understood by those skilled in the art that various modifications may be made to the embodiments without departing from the scope and spirit of the present invention. The scope of the present invention is defined by the appended claims.

Claims

1. A compliance checking method for medical data based on a large model, characterized in that: The compliance checking method for medical data based on a large model includes the following steps: S1, inputting a preset medical compliance inspection sentence into a preset large language model to obtain target medical indicators for compliance inspection; S2, inputting basic medical information corresponding to the initial medical data, basic inspection rules in a preset inspection rule library, and the target medical indicator into the preset large language model, and obtaining target basic information and candidate inspection rules corresponding to the target medical indicator, wherein the basic medical information includes all field names corresponding to the initial medical data, the basic inspection rules are in the form of SQL statements, the target basic information includes several target field names to be inspected, and each candidate inspection rule is used to inspect data corresponding to the corresponding several target field names in the initial medical data; S3, extracting target medical data from the initial medical data based on the target basic information; S4, screening a target inspection rule from all candidate inspection rules according to the target medical data; S5, inputting the target medical data and the target inspection rules into the preset large language model, and obtaining a compliance inspection report corresponding to the initial medical data.

2. The compliance checking method for medical data based on a large model according to claim 1, characterized in that: S4 includes the following steps: S41, based on the target field names corresponding to each candidate inspection rule, screening out a plurality of candidate inspection rule combinations from all candidate inspection rules, wherein each candidate inspection rule combination includes a plurality of candidate inspection rules, and each candidate inspection rule combination is used to inspect the target medical data; S42, for any candidate inspection rule combination, obtaining the spatiotemporal processing cost corresponding to the current candidate inspection rule combination based on the data volume corresponding to the target field names corresponding to each candidate inspection rule in the current candidate inspection rule combination and the spatiotemporal processing parameters corresponding to each candidate inspection rule; S43, obtaining an execution efficiency score corresponding to the current candidate inspection rule combination based on the data volume and execution efficiency parameters corresponding to the target field names corresponding to each candidate inspection rule in the current candidate inspection rule combination; S44, obtaining the priority corresponding to each candidate inspection rule combination based on the spatiotemporal processing cost and execution efficiency score corresponding to each candidate inspection rule combination; S45 , determining several candidate inspection rules corresponding to the candidate inspection rule combination with the highest priority as target inspection rules.

3. The compliance checking method for medical data based on a large model according to claim 2, characterized in that: S41 includes the following steps: S411, arbitrarily combine all candidate inspection rules to obtain several initial inspection rule combinations; S412, for any initial inspection rule combination, obtain a target field name set corresponding to the current initial inspection rule combination based on a number of target field names corresponding to each candidate inspection rule in the current initial inspection rule combination; S413, if the target field name set corresponding to the current initial check rule combination includes all target field names to be checked and there are no duplicate target field names, then the current initial check rule combination is determined as a candidate check rule combination; S414, traverse all initial inspection rule combinations, and repeat steps S412 and S413 to obtain all candidate inspection rule combinations.

4. The compliance checking method for medical data based on a large model according to claim 2, characterized in that: The spatiotemporal processing parameters include time processing parameters and space processing parameters. S42 includes the following steps: S421, obtaining a first total amount of data corresponding to each candidate inspection rule in the current candidate inspection rule combination based on the amount of data corresponding to the multiple target field names corresponding to each candidate inspection rule in the current candidate inspection rule combination; S422, obtaining a time processing cost corresponding to each candidate inspection rule in the current candidate inspection rule combination based on the first data volume and time processing parameter corresponding to each candidate inspection rule in the current candidate inspection rule combination; S423 , acquiring the temporal and spatial processing costs corresponding to each candidate inspection rule in the current candidate inspection rule combination according to the temporal processing costs and spatial processing parameters corresponding to each candidate inspection rule in the current candidate inspection rule combination.

5. The compliance checking method for medical data based on a large model according to claim 4, characterized in that: The execution performance parameters include accuracy parameters, recall rate parameters and stability parameters. S43 includes the following steps: S431, obtaining a reference score corresponding to each candidate inspection rule in the current candidate inspection rule combination based on the accuracy parameter, recall parameter, stability parameter, first preset weight corresponding to the accuracy parameter, second preset weight corresponding to the recall parameter, and third preset weight corresponding to the stability parameter corresponding to each candidate inspection rule in the current candidate inspection rule combination; S432 , obtaining an execution efficiency score corresponding to the current candidate inspection rule combination according to the total amount of first data and the reference score corresponding to each candidate inspection rule in the current candidate inspection rule combination.

6. The compliance checking method for medical data based on a large model according to claim 5, characterized in that: S432 includes the following steps: S4321, obtaining the second total amount of data corresponding to the current candidate inspection rule combination based on the first total amount of data corresponding to each candidate inspection rule in the current candidate inspection rule combination; S4322, determining a ratio of the first total amount of data corresponding to each candidate inspection rule in the current candidate inspection rule combination to the second total amount of data corresponding to the current candidate inspection rule combination as a reference weight corresponding to each candidate inspection rule in the current candidate inspection rule combination; S4323: Obtain the execution efficiency score corresponding to the current candidate inspection rule combination based on the reference weight and reference score corresponding to each candidate inspection rule in the current candidate inspection rule combination.

7. The compliance checking method for medical data based on a large model according to claim 2, characterized in that: S44 includes the following steps: S441, obtaining a cost threshold based on the spatiotemporal processing costs corresponding to all candidate inspection rule combinations; S442, obtaining a first scoring threshold and a second scoring threshold based on the execution efficiency scores corresponding to all candidate inspection rule combinations; S443, for any candidate inspection rule combination, if the execution efficiency score corresponding to the current candidate inspection rule combination is greater than or equal to the second score threshold, determine the current candidate inspection rule combination as an intermediate inspection rule combination; S444: If the execution efficiency score corresponding to the current candidate inspection rule combination is greater than or equal to the first score threshold and less than the second score threshold, and the corresponding spatiotemporal processing cost is less than or equal to the cost threshold, then the current candidate inspection rule combination is determined as an intermediate inspection rule combination; S445 , obtaining the priority corresponding to each intermediate inspection rule combination based on the spatiotemporal processing cost, the execution efficiency score, the fourth preset weight corresponding to the spatiotemporal processing cost, and the fifth preset weight corresponding to the execution efficiency score corresponding to each intermediate inspection rule combination; S446: If the execution efficiency score corresponding to the current candidate inspection rule combination is less than the first score threshold, or the corresponding spatiotemporal processing cost is greater than the cost threshold, then determine that the priority corresponding to the current candidate inspection rule combination is the lowest priority.

8. The compliance checking method for medical data based on a large model according to claim 7, characterized in that: S441 includes the following steps: S4411, fitting a cost distribution function based on the spatiotemporal processing costs corresponding to all candidate inspection rule combinations; S4412: Determine the spatiotemporal processing cost corresponding to the second quartile of the cost distribution function as a cost threshold.

9. The compliance checking method for medical data based on a large model according to claim 7, characterized in that: S442 includes the following steps: S4421, fitting a score distribution function based on the execution efficiency scores corresponding to all candidate inspection rule combinations; S4422, determining the execution efficiency score corresponding to the second quartile of the score distribution function as a first score threshold; S4423: Determine the execution efficiency score corresponding to the third quartile of the score distribution function as a second score threshold.

10. A compliance checking system for medical data based on a large model, characterized in that: The compliance checking system for medical data based on a large model includes: An indicator acquisition module is used to input a preset medical compliance inspection sentence into a preset large language model to obtain target medical indicators for compliance inspection; a data screening module, configured to input basic medical information corresponding to the initial medical data, basic inspection rules in a preset inspection rule library, and the target medical indicator into the preset large language model, and obtain target basic information and candidate inspection rules corresponding to the target medical indicator, wherein the basic medical information includes all field names corresponding to the initial medical data, the basic inspection rules are in the form of SQL statements, the target basic information includes several target field names to be inspected, and each candidate inspection rule is used to inspect data corresponding to the corresponding several target field names in the initial medical data; a data extraction module, configured to extract target medical data from the initial medical data based on the target basic information; A rule screening module, configured to screen target inspection rules from all candidate inspection rules according to the target medical data; The compliance checking module is used to input the target medical data and the target checking rules into the preset large language model to obtain a compliance checking report corresponding to the initial medical data.

Images