Methodological evaluation system and method for medical examination

Abstract

The present invention discloses a methodological evaluation system and method for medical testing, which relate to the field of medical testing technology. The evaluation system includes a data acquisition and preprocessing module, a methodological evaluation module, a result analysis and optimization module, a knowledge base construction and update module, and a feedback and continuous improvement module, which are run in sequence. The technical key points are as follows: the scheme can automatically calculate the difference between indicator data and evaluation standards according to abnormal fluctuations in equipment parameters or differences in sample processing, and trigger different optimization instructions according to the significance of the difference; for significant abnormal fluctuations in equipment parameters, the maintenance frequency is rapidly increased in an exponential increase manner to ensure stable equipment performance; for non-significant abnormal fluctuations, the maintenance frequency is appropriately adjusted, which not only ensures the effectiveness of equipment maintenance but also avoids excessive maintenance; at the same time, based on the differences in the sample processing process, suggestions for optimizing the sample processing process can be given to further improve the quality of testing.

Description

Technical Field

[0001] The present invention relates to the field of medical testing technology, and in particular to a methodology evaluation system and method for medical testing. Background Art

[0002] Medical testing is a medical method that uses laboratory technology and equipment to detect and analyze samples taken from the human body (such as blood, urine, tissue, etc.); it aims to obtain important information about diseases, health conditions or disease processes, and provide a scientific basis for clinical diagnosis and treatment; medical testing involves multiple disciplines, including microbiology, immunology, biochemistry, genetics, hematology, etc., through professional techniques and methods such as biochemical testing, immunological testing, hematological testing, microbiological testing, etc., to provide doctors with accurate diagnostic basis; medical testing plays a vital role in modern medicine and is an indispensable part of protecting patient health and improving the quality of medical services.

[0003] The existing application number is 201710083277.9, and the technical solution specified in the document named "A Methodological Evaluation System and Method for Medical Examination" includes: an indicator storage module is used to establish and store single indicators and combined indicators; a plan formulation module is used to formulate an evaluation plan to be reviewed, which forms an evaluation plan to be entered after the authority holder reviews and approves it, and the evaluation plan to be entered serves as the execution framework for sample testing; a data statistics module can derive methodological parameters; a plan approval module is used to enter the test results and methodological parameters corresponding to each single indicator to form an evaluation plan to be approved; after the evaluation plan to be approved is approved, a feasible evaluation plan is formed; the main detection module is used to store the test items, detection systems and methodological parameters that have passed the review; the present invention is conducive to promoting the standardization and data management of methodological evaluation; but it does not provide a targeted response method to solve the results after the evaluation.

[0004] In combination with the above documents and prior art:

[0005] Suppose a hospital introduces a new blood tumor marker detection method. The traditional evaluation plan for this detection method is simply to directly compare the existing standards with the standards after analyzing the new detection method, and understand the problems or advantages of the new detection method based on the comparison results. However, the results obtained by this method alone are insufficiently accurate, and potential problems are difficult to identify. At the same time, when problems are found in the new detection method, the optimization suggestions given are extremely broad. For example: Traditional systems can identify potential deviations or deficiencies, such as abnormal fluctuations in equipment parameters and differences in sample processing, but the suggestions given are mostly to adjust equipment parameters or strengthen equipment maintenance, without giving specific suggestions. The frequency of equipment maintenance still requires human control and adjustment. Summary of the Invention

[0006] (1) Technical problems solved

[0007] In view of the deficiencies in the prior art, the present invention provides a methodological evaluation system and method for medical testing, which solves the problems raised in the background art.

[0008] (2) Technical solution

[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0010] A methodological evaluation system for medical testing, the system comprising:

[0011] The data acquisition and preprocessing module collects raw data from various medical testing equipment, including at least sample information, test process parameters, and environmental conditions, and preprocesses the raw data;

[0012] The methodological evaluation module evaluates each test method based on preset evaluation criteria, analyzes each test method using statistical methods and AI models, and outputs the analysis and identification results;

[0013] The result analysis and optimization module receives the output results and generates an evaluation report, which includes optimization suggestions and strategies, after comparing the indicator data with the corresponding evaluation criteria for the second time;

[0014] If the result of triggering the potential identification mechanism is:

[0015] If there is abnormal fluctuation in equipment parameters, the calculated index data will exceed the difference of the corresponding evaluation standard;

[0016] When the difference exceeds a preset threshold, a first optimization instruction is triggered to adjust the equipment parameters and perform maintenance work according to the first equipment maintenance frequency; otherwise, a second optimization instruction is triggered to perform maintenance work according to the second equipment maintenance frequency;

[0017] If the result of triggering the potential identification mechanism is:

[0018] If there are differences in the sample processing, the third optimization instruction is triggered;

[0019] The knowledge base construction and update module establishes a comprehensive knowledge base and uses NLP technology to update it in real time. The comprehensive knowledge base is used to record real-time evaluation standards. When the evaluation standards change, the changed evaluation standards are fed back and stored in the comprehensive knowledge base.

[0020] The feedback and continuous improvement module collects feedback data, builds a feedback evaluation calculation model, generates a comprehensive index, compares the comprehensive index with the preset standard threshold, and executes corresponding strategies based on the comparison results.

[0021] Furthermore, the sample information includes the patient's basic information, sample type, and sampling time;

[0022] Among them, basic information includes age, gender and medical history, and sample types include blood, urine and tissue;

[0023] The test process parameters include the test equipment settings, reagent information, and the execution of the test steps;

[0024] The setting parameters include temperature, humidity and light; the reagent information includes batch number, expiration date and concentration; the execution status includes reaction time and sample volume;

[0025] Environmental conditions include room temperature, humidity, and air pressure.

[0026] Furthermore, the raw data are pre-processed including: data cleaning, format unification and outlier processing;

[0027] In data cleaning, machine learning algorithms are used to identify and correct data errors. The implementation process is as follows:

[0028] Algorithm selection: choose supervised learning or unsupervised learning algorithm;

[0029] Model training: Use a known correct dataset as a training set to train the machine learning model;

[0030] Real-time application: The trained model is deployed in the data acquisition and preprocessing module to process newly collected data in real time; the machine learning model identifies errors in the raw data and corrects them based on the learned rules.

[0031] Furthermore, the process of using statistical methods to analyze each test method is as follows:

[0032] Extract the pre-processed raw data and simultaneously obtain the test results, including positive or negative, calculate the index data, including sensitivity and specificity, and compare the index data with the corresponding evaluation criteria;

[0033] If any indicator data exceeds the corresponding evaluation standard, it is judged as abnormal; otherwise, it is judged as fuzzy normal;

[0034] Furthermore, the process of using AI models to analyze each inspection method is as follows:

[0035] Under the condition that the statistical method determines that the result is fuzzy and normal, a deep learning model is introduced to trigger a strategy of secondary analysis of the inspection process. The steps of this strategy are as follows:

[0036] Feature extraction: Use the CNN model to extract features from the original data;

[0037] Pattern recognition and performance evaluation: The extracted features are input into the RNN model for pattern recognition and performance evaluation; the time dependency and long-term trends in the original data are captured; the RNN model is trained to predict the detection results under different conditions and calculate the index data, which is then compared with the corresponding evaluation criteria;

[0038] If any indicator data exceeds the corresponding evaluation standard, the potential identification mechanism will be triggered to identify potential deviations, including at least abnormal fluctuations in equipment parameters and differences in sample processing; otherwise, it will be judged as normal.

[0039] Furthermore, the calculation process of the first equipment maintenance frequency is as follows:

[0040]

[0041] Where Pw1 represents the first equipment maintenance frequency, Pw0 represents the original equipment maintenance frequency, k1 represents the first impact coefficient, and k1>0, Cz represents the difference between the indicator data and the corresponding evaluation standard, Pow represents the definition threshold, m represents the power of exponential growth, and m>0;

[0042] The calculation process of the second equipment maintenance frequency is as follows:

[0043]

[0044] Wherein, Pw2 represents the second equipment maintenance frequency, k2 represents the second influence coefficient, and k2>0.

[0045] Furthermore, the process of collecting feedback data is as follows:

[0046] Select L target users, record the satisfaction score, problem-solving time, and problem-solving rate of each target user, and average the L satisfaction scores, problem-solving time, and problem-solving rates to obtain feedback data, namely the mean satisfaction score, mean problem-solving time, and mean problem-solving rate.

[0047] Furthermore, the formula used to construct the feedback evaluation calculation model is as follows:

[0048]

[0049] Where, I represents the comprehensive index, W s represents the weight of the mean satisfaction, S represents the mean satisfaction, W T represents the weight of the mean problem-solving time, T represents the mean problem-solving time, W s +W T=1, k3 represents the adjustment coefficient, and k3>0, R represents the mean problem solving rate, R0 represents the benchmark problem solving rate, v represents the power value, and v>1.

[0050] Furthermore, the implementation promotion strategy is: to promote the current inspection method;

[0051] The implementation improvement strategy is: prompting the current inspection method to be improved and developed.

[0052] A method for evaluating medical testing methodology comprises the following steps:

[0053] S1. Collect raw data from various medical testing equipment, including at least sample information, test process parameters, and environmental conditions, and pre-process the raw data;

[0054] S2. Evaluate each inspection method based on pre-set evaluation criteria, analyze each inspection method using statistical methods and AI models, and output the analysis and identification results;

[0055] S3. Receive the output results and generate an evaluation report, which includes optimization recommendations, after comparing the indicator data with the corresponding evaluation criteria for the second time;

[0056] If the result of triggering the potential identification mechanism is:

[0057] If there is abnormal fluctuation in equipment parameters, the calculated index data will exceed the difference of the corresponding evaluation standard;

[0058] When the difference exceeds a preset threshold, a first optimization instruction is triggered to adjust the equipment parameters and perform maintenance work according to the first equipment maintenance frequency; otherwise, a second optimization instruction is triggered to perform maintenance work according to the second equipment maintenance frequency;

[0059] If the result of triggering the potential identification mechanism is:

[0060] If there are differences in the sample processing, the third optimization instruction is triggered;

[0061] S4. Establish a comprehensive knowledge base and use NLP technology to update it in real time. The comprehensive knowledge base is used to record the real-time evaluation criteria. When the evaluation criteria change, the changed evaluation criteria are fed back and stored in the comprehensive knowledge base.

[0062] S5. Collect feedback data, build a feedback evaluation calculation model, generate a comprehensive index, compare the comprehensive index with the preset standard threshold, and execute corresponding strategies based on the comparison results.

[0063] (3) Beneficial effects

[0064] The present invention provides a methodological evaluation system and method for medical testing, which has the following beneficial effects:

[0065] (1) This solution combines statistical methods and deep learning models to conduct a comprehensive and in-depth analysis and evaluation of each inspection method; statistical methods can quickly screen out abnormal indicators, while deep learning models can not only further verify but also mine the potential features in the data and identify potential deviations or deficiencies in fuzzy normality, such as abnormal fluctuations in equipment parameters and differences in sample processing, thereby improving the accuracy and reliability of the evaluation;

[0066] (2) This solution can intelligently receive evaluation results, generate intuitive evaluation reports, and provide targeted optimization suggestions. Based on abnormal fluctuations in equipment parameters or differences in sample processing, it automatically calculates the difference between the index data and the evaluation standard, and triggers different optimization instructions based on the significance of the difference. For significant abnormal fluctuations in equipment parameters, the maintenance frequency is rapidly increased by exponential increase to ensure stable equipment performance. For non-significant abnormal fluctuations, the maintenance frequency is appropriately adjusted, which not only ensures the effectiveness of equipment maintenance but also avoids excessive maintenance. At the same time, based on the differences in the sample processing process, it can provide suggestions for optimizing the sample processing process to further improve the inspection quality.

[0067] (3) A comprehensive feedback and continuous improvement plan was constructed to achieve effective evaluation and continuous optimization of medical laboratory methodology; by collecting feedback data from target users, a scientific feedback evaluation calculation model was constructed, and a comprehensive index was generated to quantify the performance of the evaluation methodology; the close linkage between the intelligent update of the knowledge base and the evaluation module not only ensured the accuracy and timeliness of the evaluation, but also improved the sensitivity and influence of the problem-solving rate in the evaluation by adjusting the model design of coefficients and power values, further ensuring the effectiveness of the comprehensive index, and facilitating the subsequent implementation of corresponding strategies based on the comparison results. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 It is a modular schematic diagram of the evaluation system of the present invention;

[0069] Figure 2 Schematic diagram of the overall process of the evaluation method of the present invention. DETAILED DESCRIPTION

[0070] The following will provide a clear and complete description of 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. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0071] System Overview

[0072] SMTMES (the evaluation system) aims to comprehensively, objectively, and efficiently evaluate medical testing methods by integrating modern information technology, big data analytics, artificial intelligence algorithms, and medical laboratory expertise. This aims to improve the accuracy and reliability of test results and promote the innovative development of medical testing technology. The system is divided into five functional modules: data acquisition and preprocessing, methodological evaluation, result analysis and optimization, knowledge base construction and updating, and user feedback and continuous improvement, forming a closed-loop evaluation and optimization system.

[0073] Example 1:

[0074] See also Figure 1 ,This embodiment provides a methodological evaluation system for medical testing, which includes a data acquisition and preprocessing module, a methodological evaluation module, a result analysis and optimization module, a knowledge base construction and update module, and a feedback and continuous improvement module, which are operated in sequence;

[0075] The functional modules are described in detail as follows

[0076] Data acquisition and preprocessing module:

[0077] Collect raw data from various medical testing equipment, including at least sample information, testing process parameters, and environmental conditions, and pre-process the raw data;

[0078] Among them, sample information: including basic information of the patient (such as age, gender and medical history), sample type (such as blood, urine and tissue) and sampling time;

[0079] Test process parameters: These include test equipment settings (such as temperature, humidity, and light), reagent information (such as batch number, expiration date, and concentration), and test step execution (such as reaction time and sample volume).

[0080] Environmental conditions: Record the laboratory environmental conditions during the test, such as room temperature, humidity, and air pressure, which may affect the accuracy of the test results;

[0081] Preprocessing of raw data includes:

[0082] Data cleaning: remove duplicate data, fill in missing values, and correct obvious data entry errors; format unification: convert data from different testing equipment and formats into a unified format to facilitate subsequent processing and analysis; outlier processing: identify and process outliers in the data, which may be caused by equipment failure, operational errors, or abnormal samples themselves;

[0083] In data cleaning, machine learning algorithms are used to automatically identify and correct data errors, improving the efficiency and accuracy of data processing. The specific implementation process is as follows:

[0084] Algorithm selection: Choose supervised or unsupervised learning algorithms, such as random forest, support vector machine (SVM), K-nearest neighbor (KNN), or clustering algorithms, and select the appropriate algorithm based on data characteristics and error types;

[0085] Model training: Using a known or manually verified dataset as the training set, the machine learning model is trained. The model learns how to identify and correct errors in the data, such as outliers outside the numerical range and logical inconsistencies.

[0086] Real-time application: The trained model is deployed in the data acquisition and preprocessing module to process newly collected data in real time. The model automatically identifies errors in the raw data and corrects them based on the learned rules.

[0087] For example:

[0088] Suppose that during a blood test, the blood glucose level of a sample is mistakenly entered as 1000 mg / dL (the normal blood glucose range is typically 70-140 mg / dL). Traditional data cleaning methods may require manual inspection and correction of this error. However, in this system, the machine learning model can automatically identify this outlier and, based on the blood glucose distribution of other samples and other test parameters of the sample (such as age, gender, sampling time, etc.), infer the correct blood glucose range and automatically correct the error. This not only improves the accuracy of data processing, but also greatly reduces the time and cost of manual intervention.

[0089] Methodological Assessment Module:

[0090] Evaluate each inspection method based on preset evaluation criteria, analyze each inspection method using statistical methods and AI models, and output the analysis and identification results;

[0091] The process of using statistical methods to analyze each test method is as follows:

[0092] Extract the pre-processed raw data and simultaneously obtain the test results, including positive or negative, calculate the index data, which includes sensitivity and specificity, and compare the index data with the corresponding evaluation criteria. If any index data exceeds the corresponding evaluation criteria, it is judged as abnormal; otherwise, it is judged as fuzzy normal;

[0093] The process of analyzing each inspection method using the AI ​​model is as follows:

[0094] Under the condition that the statistical method determines that the result is fuzzy and normal, a deep learning model is introduced to trigger a strategy of secondary analysis of the inspection process. The steps of this strategy are as follows:

[0095] Feature extraction: CNN models are used to extract features from the data during the test (i.e., raw data). CNNs can automatically learn the spatial structure and local features in the data, such as trends in tumor marker concentrations and fluctuations in device parameters. These features are crucial for evaluating the performance of test methods.

[0096] Pattern recognition and performance evaluation: The extracted features are input into the RNN model for pattern recognition and performance evaluation. RNN can process sequential data and capture temporal dependencies and long-term trends in the data. By training the RNN model, we can predict detection results under different conditions and calculate sensitivity and specificity. The indicator data is then compared with the corresponding evaluation criteria. If any indicator exceeds the corresponding evaluation criteria, the potential recognition mechanism is triggered.

[0097] Otherwise, it is judged as normal;

[0098] Among them, the potential identification mechanism is used to identify potential deviations or deficiencies, including at least: abnormal fluctuations in equipment parameters and differences in sample processing; the AI ​​model is a deep learning model, and the deep learning module includes CNN and then RNN.

[0099] Specifically, the system not only achieves comprehensive and accurate collection and preprocessing of raw data from medical testing equipment, effectively removing duplicate, missing, and erroneous data, but also improves the efficiency and accuracy of data processing through machine learning algorithms;

[0100] At the same time, a comprehensive and in-depth analysis and evaluation of each inspection method was conducted by combining statistical methods with deep learning models. Statistical methods can quickly screen out abnormal indicators, while deep learning models can not only further verify them but also mine the underlying features in the data, identifying potential deviations or deficiencies in fuzzy normals, such as abnormal fluctuations in equipment parameters and differences in sample processing, thereby improving the accuracy and reliability of the assessment.

[0101] In summary, the above solution solves technical problems such as low efficiency, insufficient accuracy, and difficulty in identifying potential problems in traditional data processing and analysis. It provides strong support for the performance evaluation of medical testing methods, helps to improve the accuracy and reliability of medical tests, and thus promotes the improvement of medical standards and patient health.

[0102] Result analysis and optimization module:

[0103] Receive the output results and use visualization tools to generate an evaluation report, which includes optimization recommendations. After comparing the indicator data with the corresponding evaluation criteria for the second time, if the result of triggering the potential identification mechanism is:

[0104] If there is abnormal fluctuation in equipment parameters, the calculated index data will exceed the difference of the corresponding evaluation standard;

[0105] When the difference exceeds a preset threshold, it indicates that the abnormal fluctuation of the equipment parameters has a significant impact on the detection results, triggering a first optimization instruction, which is to adjust the equipment parameters and perform maintenance work according to the first equipment maintenance frequency;

[0106] The calculation process of the maintenance frequency of the first equipment is as follows:

[0107]

[0108] Where Pw1 represents the first equipment maintenance frequency, Pw0 represents the original equipment maintenance frequency, k1 represents the first impact coefficient, which indicates the increase in maintenance frequency when the difference exceeds the defined threshold, and k1>0, Cz represents the difference between the indicator data and the corresponding evaluation standard, Pow represents the defined threshold, m represents the power of exponential growth, which indicates the accelerating effect of abnormal fluctuations on the increase in maintenance frequency, and m>0;

[0109] It should be noted that the part multiplied by k1 represents the relative distance between the difference and the defined threshold, which is used to measure the degree of abnormal fluctuation. When the difference exceeds the defined threshold, the maintenance frequency is rapidly increased through exponential increase to cope with significant abnormal fluctuations in equipment parameters.

[0110] The difference Cz of the indicator data exceeding the corresponding evaluation standard:

[0111] If the difference Cz is just any indicator data, then a normal difference calculation is performed to obtain a positive difference CZ. If there is more than one indicator data, then the difference of each indicator data exceeding the corresponding evaluation standard is calculated, and each difference is weighted to obtain the required difference Cz.

[0112] If the difference does not exceed the preset threshold, it means that the abnormal fluctuation of the equipment parameters has an impact on the test results, but it is not significant enough, triggering the second optimization instruction, which is: perform maintenance work according to the second equipment maintenance frequency;

[0113] The calculation process of the second equipment maintenance frequency is as follows:

[0114]

[0115] Where Pw2 represents the maintenance frequency of the second equipment, k2 represents the second impact coefficient, which indicates the increase in the maintenance frequency when the difference exceeds the defined threshold, and k2>0;

[0116] Summarize

[0117] First, the frequency of equipment maintenance increases exponentially to quickly respond to significant abnormal fluctuations in equipment parameters;

[0118] Second, the frequency of equipment maintenance is increased exponentially to appropriately address non-significant abnormal fluctuations in equipment parameters;

[0119] Both formulas take into account the relative distance between the difference and the defined threshold, as well as the original maintenance frequency, to ensure the pertinence and rationality of the maintenance strategy;

[0120] If the result of triggering the potential identification mechanism is:

[0121] If there are differences in the sample processing process, the third optimization instruction is triggered, and the instruction content is: optimize the sample processing process;

[0122] In addition, in the result analysis and optimization module:

[0123] The evaluation report also automatically proposes other targeted improvement suggestions based on the evaluation results, such as adjusting reagent ratios, changing reaction times, and optimizing the test process. When evaluating a new urine analysis test method, the methodological evaluation module provides the method's sensitivity and specificity and points out that the method's performance needs to be improved under certain specific conditions (such as specific reagent ratios and reaction times).

[0124] In the result analysis and optimization module, intelligent optimization technology will be used to explore the best combination of inspection conditions;

[0125] The specific steps are as follows:

[0126] Define the optimization problem:

[0127] Identify optimization goals, such as improving sensitivity or specificity;

[0128] Determine optimization variables, such as reagent ratio, reaction time, etc.;

[0129] Define constraints, such as reagent cost, detection time, etc.

[0130] Select an optimization algorithm:

[0131] Select appropriate intelligent optimization algorithms, such as genetic algorithms or particle swarm optimization, based on the complexity and characteristics of the problem;

[0132] Algorithm implementation and iteration:

[0133] Model and solve optimization problems using selected algorithms;

[0134] Through iterative calculations, the optimal solution is continuously approached; in each iteration, the algorithm is adjusted and optimized based on the performance of the current solution;

[0135] Result analysis and verification:

[0136] Conduct detailed analysis on the optimal solution output by the algorithm to verify whether it meets the optimization goal;

[0137] If the requirements are met, the optimal solution is provided to the user as an improvement suggestion; if not, iterative calculations are continued or algorithm parameters are adjusted;

[0138] Example application:

[0139] Suppose that after optimization through a genetic algorithm, we have found a new set of reagent ratios and reaction times that increases the sensitivity of the urine analysis test method by 10%. We provide this set of optimal solutions to the user as improvement suggestions and recommend actual verification. If the verification results are good, the combination can be promoted as a new standard detection method.

[0140] By introducing intelligent optimization technology, the result analysis and optimization module can automatically search for the best combination of test conditions, greatly accelerating the method optimization process and improving the accuracy and efficiency of the test method; at the same time, the linkage with the methodology evaluation module ensures that the optimization process is based on scientific evaluation results, ensuring the correctness and effectiveness of the optimization direction.

[0141] By adopting the above technical solution, a result analysis and optimization module was built. This module can intelligently receive evaluation results, generate intuitive evaluation reports, and provide targeted optimization suggestions. Based on abnormal fluctuations in equipment parameters or differences in sample processing, it automatically calculates the difference between the indicator data and the evaluation standard, and triggers different optimization instructions based on the significance of the difference.

[0142] For significant abnormal fluctuations in equipment parameters, the module exponentially increases the maintenance frequency to ensure stable equipment performance. For less significant abnormal fluctuations, the maintenance frequency is adjusted appropriately, ensuring effective equipment maintenance while avoiding excessive maintenance. Furthermore, based on differences in sample processing, the module provides recommendations for optimizing the sample processing process to further improve inspection quality.

[0143] This technical solution not only improves the accuracy and efficiency of result analysis, but also solves the problems of traditional maintenance strategies lacking in specificity and flexibility, as well as potential hidden dangers in the sample processing process. It provides strong guarantees for the accuracy and reliability of medical tests, and helps to improve the quality of medical services and patient satisfaction.

[0144] Knowledge base construction and update module:

[0145] Establish a comprehensive knowledge base and use NLP technology to update it in real time to provide an authoritative reference for evaluation. The comprehensive knowledge base is used to record real-time evaluation standards. When the evaluation standards change, the changed evaluation standards are fed back and stored in the comprehensive knowledge base.

[0146] Specific instructions and examples

[0147] Automatically extract key information:

[0148] Principle: NLP technology uses text parsing, entity recognition, and relationship extraction to automatically identify and extract key information from literature, including the principles, operational procedures, and evaluation criteria of a test method. For example, consider a paper on a new blood tumor marker detection method. NLP technology can automatically extract the method's detection principle (e.g., based on immunochromatography), operational procedures (e.g., sample collection, reagent preparation, and test steps), evaluation criteria (e.g., sensitivity, specificity, and accuracy), and experimental data.

[0149] Intelligent induction and classification:

[0150] Principle: NLP technology can also intelligently summarize and categorize the extracted information, organizing it into different themes or categories, such as by test method type, disease type, and evaluation indicators. For example, using the blood tumor marker detection method mentioned above as an example, NLP technology can summarize the extracted information under the theme of "blood tumor marker detection" and further organize it into subcategories such as "test method" and "evaluation criteria."

[0151] Linkage with the Methodology Assessment Module:

[0152] Principle: The knowledge base construction and update module maintains a close linkage with the evaluation module. When the evaluation module needs to refer to the principles, operating procedures or evaluation standards of a certain type of testing method, it can directly obtain relevant information from the knowledge base. At the same time, new data and new experiences generated during the evaluation process can also be promptly fed back to the knowledge base for updating and expansion. For example: when evaluating a new blood tumor marker detection method, the evaluation module can obtain information such as the principles, operating procedures and evaluation standards of the method from the knowledge base as a reference. After the evaluation is completed, the evaluation results, experimental data and possible improvement suggestions can be automatically extracted by NLP technology and classified into the corresponding categories in the knowledge base, realizing the dynamic update of the knowledge base.

[0153] Rapid expansion and intelligent management:

[0154] Principle: Through the continuous application of NLP technology, knowledge bases can rapidly expand and intelligently manage information from large volumes of literature. This not only increases the capacity and coverage of the knowledge base, but also makes its management more efficient and convenient. For example, with the continuous development of the field of medical testing, a large number of new documents are published daily. Through the automatic extraction and summarization capabilities of NLP technology, these knowledge base construction and update modules can quickly incorporate key information from these new documents into the knowledge base, providing the latest reference for evaluation work.

[0155] In summary, the knowledge base construction and update module has achieved rapid expansion and intelligent management of the knowledge base by introducing natural language processing (NLP) technology, providing strong support for the performance evaluation of medical testing methods; at the same time, the close linkage with the methodological evaluation module also ensures the timely updating and effective utilization of knowledge.

[0156] Feedback and Continuous Improvement Module:

[0157] Collect feedback data, build a feedback evaluation calculation model, generate a comprehensive index, and compare the comprehensive index with the preset standard threshold. If the comprehensive index exceeds the standard threshold, it means that the methodology used for medical testing meets the standard and a promotion strategy is implemented; if the comprehensive index does not exceed the standard threshold, it means that the methodology used for medical testing does not meet the standard and an improvement strategy is implemented;

[0158] The process of collecting feedback data is as follows:

[0159] Select L target users, who are inspectors or researchers, and record the satisfaction score, problem-solving time, and problem-solving rate of each target user. Average the L satisfaction scores, problem-solving time, and problem-solving rates to obtain feedback data, namely the mean satisfaction score, mean problem-solving time, and mean problem-solving rate.

[0160] L is a positive integer, and the value of L is at least 3. In this embodiment, the value of L is 50;

[0161] The formula used to construct the feedback evaluation calculation model is as follows:

[0162]

[0163] Where, I represents the comprehensive index, W s represents the weight of the mean satisfaction score, S represents the mean satisfaction score, and the satisfaction score ranges from 1 to 10, with 10 being the best. T represents the weight of the mean problem-solving time, T represents the mean problem-solving time, W s +W T=1, k3 represents the adjustment coefficient, and k3>0, R represents the mean problem resolution rate, R0 represents the benchmark problem resolution rate, such as the industry average or historical data, v represents the power value, which is used to adjust the impact of the problem resolution rate on the comprehensive index, and v>1;

[0164] Logic description:

[0165] The purpose of introducing weights is to allow users or decision makers to adjust the relative importance of the mean satisfaction and mean problem-solving time in the comprehensive index according to actual needs. The sum of the weights is 1, which ensures that the contributions of the two to the comprehensive index are complementary. The introduction of the adjustment coefficient k3, the baseline problem-solving rate R0, and the power value v is to more finely adjust the contribution of the mean problem-solving rate in the comprehensive index. When the mean problem-solving rate is higher than the baseline level, the k3*(R-R0) v In the form of , its positive impact on the comprehensive index can be amplified; conversely, when the mean problem-solving rate is lower than the benchmark level, its negative impact will also be amplified accordingly;

[0166] The implementation promotion strategy is: to promote the current medical laboratory methodology (i.e., current testing methods);

[0167] The implementation improvement strategy is: prompting the current inspection method to be improved and developed.

[0168] By adopting the above technical solutions, a comprehensive feedback and continuous improvement system was built, achieving effective evaluation and continuous optimization of medical laboratory methodology. By collecting feedback data from target users, a scientific feedback evaluation calculation model was constructed, and a comprehensive index was generated to quantify the performance of the evaluation methodology. The intelligent update of the knowledge base and the close linkage of the evaluation module not only ensured the accuracy and timeliness of the evaluation, but also improved the sensitivity and influence of the problem-solving rate in the evaluation through the design of adjusting coefficients and power values.

[0169] When the methodology meets the standards, the system automatically executes the promotion strategy to accelerate the popularization of excellent methods;

[0170] When the standards are not met, improvement strategies are triggered to guide further development of the methodology;

[0171] This technical solution effectively solves the problems of strong subjectivity, delayed feedback, and unclear improvement direction in traditional evaluation methods. It provides strong technical support for the continuous improvement and performance enhancement of medical testing methodology, and helps promote the rapid development of the medical testing field and the overall improvement of medical service quality.

[0172] Example 2:

[0173] See also Figure 2Based on Example 1, this example further provides a methodological evaluation method for medical testing, comprising the following steps:

[0174] S1. Collect raw data from various medical testing equipment, including at least sample information, test process parameters, and environmental conditions, and pre-process the raw data;

[0175] S2. Evaluate each inspection method based on pre-set evaluation criteria, analyze each inspection method using statistical methods and AI models, and output the analysis and identification results;

[0176] S3. Receive the output results and generate an evaluation report, which includes optimization recommendations, after comparing the indicator data with the corresponding evaluation criteria for the second time;

[0177] If the result of triggering the potential identification mechanism is:

[0178] If there is abnormal fluctuation in equipment parameters, the calculated index data will exceed the difference of the corresponding evaluation standard;

[0179] When the difference exceeds a preset threshold, a first optimization instruction is triggered to adjust the equipment parameters and perform maintenance work according to the first equipment maintenance frequency; otherwise, a second optimization instruction is triggered to perform maintenance work according to the second equipment maintenance frequency;

[0180] If the result of triggering the potential identification mechanism is:

[0181] If there are differences in the sample processing, the third optimization instruction is triggered;

[0182] S4. Establish a comprehensive knowledge base and use NLP technology to update it in real time. The comprehensive knowledge base is used to record the real-time evaluation criteria. When the evaluation criteria change, the changed evaluation criteria are fed back and stored in the comprehensive knowledge base.

[0183] S5. Collect feedback data, build a feedback evaluation calculation model, generate a comprehensive index, compare the comprehensive index with the preset standard threshold, and execute corresponding strategies based on the comparison results.

[0184] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will appreciate that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution.

[0185] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment as needed.

[0186] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A methodological evaluation system for medical testing, characterized by: The system includes: The data acquisition and preprocessing module collects raw data from various medical testing equipment, including at least sample information, test process parameters, and environmental conditions, and preprocesses the raw data; The methodological evaluation module evaluates each test method based on the preset evaluation criteria, analyzes each test method using statistics and AI models, and outputs the results of the analysis and identification; the process of using statistics to analyze each test method is as follows: extract the pre-processed raw data, and simultaneously obtain the test results, including positive or negative, calculate the index data, and the index data includes sensitivity and specificity, and compare the index data with the corresponding evaluation criteria; if any index data exceeds the corresponding evaluation criteria, it is judged to be abnormal; otherwise, it is judged to be fuzzy normal; the process of using AI models to analyze each test method is as follows: under the condition that the statistical judgment result is fuzzy normal, deep learning is introduced The learning model triggers a strategy for secondary analysis of the inspection process. The steps of this strategy are as follows: Feature extraction: Use the CNN model to extract features from the original data; Pattern recognition and performance evaluation: Input the extracted features into the RNN model for pattern recognition and performance evaluation; Capture the time dependency and long-term trend in the original data; Train the RNN model to predict the inspection results under different conditions and calculate the indicator data, and then compare the indicator data with the corresponding evaluation criteria; If any indicator data exceeds the corresponding evaluation criteria, the potential recognition mechanism is triggered to identify potential deviations, including at least abnormal fluctuations in equipment parameters and differences in sample processing; Otherwise, it is judged to be normal; The result analysis and optimization module receives the output results and generates an evaluation report, which includes optimization suggestions and strategies, after comparing the indicator data with the corresponding evaluation criteria for the second time; If the result of triggering the potential identification mechanism is: If there is abnormal fluctuation in equipment parameters, the calculated index data will exceed the difference of the corresponding evaluation standard; When the difference exceeds a preset threshold, a first optimization instruction is triggered to adjust the equipment parameters and perform maintenance work according to the first equipment maintenance frequency. Otherwise, a second optimization instruction is triggered to perform maintenance work according to the second equipment maintenance frequency. The calculation process of the first equipment maintenance frequency is as follows: ； Where Pw1 represents the first equipment maintenance frequency, Pw0 represents the original equipment maintenance frequency, k1 represents the first impact coefficient, and k1>0, Cz represents the difference between the indicator data and the corresponding evaluation standard, Pow represents the definition threshold, m represents the power of exponential growth, and m>0; The calculation process of the second equipment maintenance frequency is as follows: ； Where Pw2 represents the maintenance frequency of the second equipment, k2 represents the second influence coefficient, and k2>0; If the result of triggering the potential identification mechanism is: If there are differences in the sample processing process, the third optimization instruction is triggered, and the instruction content is: optimize the sample processing process; The knowledge base construction and update module establishes a comprehensive knowledge base and uses NLP technology to update it in real time. The comprehensive knowledge base is used to record real-time evaluation standards. When the evaluation standards change, the changed evaluation standards are fed back and stored in the comprehensive knowledge base. The feedback and continuous improvement module collects feedback data, builds a feedback evaluation calculation model, generates a comprehensive index, compares the comprehensive index with the preset standard threshold, and executes corresponding strategies based on the comparison results.

2. A methodological evaluation system for medical testing according to claim 1, characterized in that: Sample information includes the patient's basic information, sample type, and sampling time; Among them, basic information includes age, gender and medical history, and sample types include blood, urine and tissue; The test process parameters include the test equipment settings, reagent information, and the execution of the test steps; The setting parameters include temperature, humidity and light; the reagent information includes batch number, expiration date and concentration; the execution status includes reaction time and sample volume; Environmental conditions include room temperature, humidity, and air pressure.

3. A methodological evaluation system for medical testing according to claim 1, characterized in that: Preprocessing of raw data includes: data cleaning, format unification and outlier processing; In data cleaning, machine learning algorithms are used to identify and correct data errors. The implementation process is as follows: Algorithm selection: Choose supervised learning or unsupervised learning algorithm; Model training: Use a known correct dataset as a training set to train the machine learning model; Real-time application: The trained model is deployed in the data acquisition and preprocessing module to process newly collected data in real time; the machine learning model identifies errors in the raw data and corrects them based on the learned rules.

4. A methodological evaluation system for medical testing according to claim 1, characterized in that: The process of collecting feedback data is: Select L target users, record the satisfaction score, problem-solving time, and problem-solving rate of each target user, and average the L satisfaction scores, problem-solving time, and problem-solving rates to obtain feedback data, namely the mean satisfaction score, mean problem-solving time, and mean problem-solving rate.

5. A methodological evaluation system for medical testing according to claim 4, characterized in that: The formula used to construct the feedback evaluation calculation model is as follows: ； Where I represents the comprehensive index, represents the weight of the mean satisfaction, S represents the mean satisfaction, represents the weight of the mean problem-solving time, T represents the mean problem-solving time, , k3 represents the adjustment coefficient, and k3>0, R represents the mean problem solving rate, represents the benchmark problem solving rate, v represents the power value, and v>

1.

6. A methodological evaluation system for medical testing according to claim 1, characterized in that: The process of executing the corresponding strategy based on the comparison results is as follows: If the comprehensive index exceeds the standard threshold, the promotion strategy is executed to promote the current inspection method; If the comprehensive index does not exceed the standard threshold, the improvement strategy will be implemented, prompting the current inspection method to be improved and developed.

7. A method for evaluating a method for medical testing, using the system according to any one of claims 1 to 6, characterized in that: The steps include: S1. Collect raw data from various medical testing equipment, including at least sample information, test process parameters, and environmental conditions, and pre-process the raw data; S2. Evaluate each inspection method based on pre-set evaluation criteria, analyze each inspection method using statistics and AI models, and output the analysis and identification results; S3. Receive the output results and generate an evaluation report, which includes optimization recommendations, after comparing the indicator data with the corresponding evaluation criteria for the second time; If the result of triggering the potential identification mechanism is: If there is abnormal fluctuation in equipment parameters, the calculated index data will exceed the difference of the corresponding evaluation standard; When the difference exceeds a preset threshold, a first optimization instruction is triggered to adjust the equipment parameters and perform maintenance work according to the first equipment maintenance frequency; otherwise, a second optimization instruction is triggered to perform maintenance work according to the second equipment maintenance frequency; If the result of triggering the potential identification mechanism is: If there are differences in the sample processing, the third optimization instruction is triggered; S4. Establish a comprehensive knowledge base and use NLP technology to update it in real time. The comprehensive knowledge base is used to record the real-time evaluation criteria. When the evaluation criteria change, the changed evaluation criteria are fed back and stored in the comprehensive knowledge base. S5. Collect feedback data, build a feedback evaluation calculation model, generate a comprehensive index, compare the comprehensive index with the preset standard threshold, and execute corresponding strategies based on the comparison results.

Images