Multi-dimensional evaluation and analysis system for clinical medicine practical skills

Abstract

The invention discloses a clinical medicine practical skill multi-dimensional evaluation and analysis system, and relates to the technical field of clinical medicine education and skill evaluation, and the system comprises a multi-source data collection module which obtains operation and other multi-dimensional data; the data standardization preprocessing module unifies formats, corrects exceptions, encrypts and desensitizes; the data intelligent quantitative calculation module converts a quantitative result by using an algorithm; the multi-dimensional dynamic evaluation module constructs a three-dimensional model, sets three stages of periods, and combines AI and manual output scores; the personalized feedback module generates a report and a scheme and supports multi-scene data linkage; according to the method, multi-source data is integrated, the limitation of single data of traditional evaluation is broken through, the difficult-to-quantify ability is converted into objective indexes through an intelligent algorithm, a three-dimensional evaluation model and a three-level evaluation period are constructed, skill development is dynamically tracked, AI and manual rechecking are fused, efficiency and accuracy are balanced, and a personalized scheme and a multi-scene report are generated. And clinical medicine evaluation upgrading is promoted.

Description

Technical Field

[0001] This invention relates to the field of clinical medical education and skills assessment technology, specifically a multi-dimensional evaluation and analysis system for clinical medical practice skills. Background Technology

[0002] In the field of clinical medicine, practical skills are a core element in measuring the professional competence of medical practitioners, directly impacting the quality of medical services and patient safety. Therefore, the scientific evaluation of clinical medical practical skills has become a crucial link in medical education, standardized physician training, and industry quality supervision. With the deepening of medical education reform and the rapid development of medical information technology, clinical skills simulation training equipment, electronic medical record systems, and audio-visual acquisition technologies have become increasingly widespread, providing diverse data acquisition channels for practical skills evaluation. Currently, the industry's demand for practical skills evaluation has shifted from simple operational assessments to the evaluation of practitioners' comprehensive abilities. It not only focuses on the standardization and accuracy of operations but also increasingly emphasizes the logic of clinical thinking, the effectiveness of doctor-patient communication, and the coordination of teamwork. However, existing technologies have not yet formed a complete system capable of integrating multi-source data and achieving multi-dimensional competence quantification and dynamic evaluation, making it difficult to fully meet the needs of modern clinical medical talent training for precise and systematic evaluation. Therefore, it is urgent to construct a more comprehensive practical skills evaluation framework through technology.

[0003] Traditional clinical medical practice skills assessment methods have significant limitations and are difficult to adapt to the development requirements of modern medical education and practice. First, traditional assessments are mostly based on manual examinations, relying on the subjective judgment of examiners. The assessment results are easily influenced by factors such as personal experience and cognitive biases, lacking objective quantitative evidence. In particular, there is a lack of effective means to assess implicit abilities such as clinical thinking and professional ethics, resulting in a one-sided assessment dimension that cannot fully reflect the comprehensive ability level of practitioners. Second, the data sources for traditional assessments are relatively scattered, relying mostly on paper records or data from single devices. There is a lack of integration of multi-source data such as operation trajectories, communication audio and video, and team collaboration logs. Moreover, the data processing lacks standardized procedures, which easily leads to problems such as inconsistent data formats and abnormal data interference, affecting the validity and utilization rate of the data. In addition, traditional assessments mostly focus on final assessments, lacking real-time tracking and phased evaluation of the practice process. It is difficult to identify problems in skills training in a timely manner, and it is also difficult to form a continuous feedback and improvement mechanism for individuals, which restricts the guiding role of assessment in talent training. At the same time, some assessment methods lack adequate protection measures for learner and patient information, posing information security risks. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a multi-dimensional evaluation and analysis system for clinical medical practice skills. This system acquires data such as operation trajectories and medical record texts through a multi-source data acquisition module; a data standardization preprocessing module performs data transformation, outlier handling, and encryption / desensitization; a data intelligent quantification calculation module uses algorithms such as dynamic risk weighting to transform operation and medical record data into quantitative results; a multi-dimensional dynamic evaluation module constructs a three-dimensional model, sets three-level evaluation cycles, and combines AI pre-scoring with manual review to output a comprehensive score; and a personalized feedback module generates evaluation reports and improvement plans, supporting multi-scenario data linkage.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a multi-dimensional evaluation and analysis system for clinical medical practice skills, the system comprising:

[0006] Multi-source data acquisition module: By connecting to clinical skills simulation equipment and electronic medical record system, deploying high-definition cameras and microphones to collect operation action videos and doctor-patient communication audio and video, and linking with training log platform to collect team collaboration data, obtain operation trajectory, medical record text, audio and video streams and collaboration logs;

[0007] Data standardization preprocessing module: Converts video into frame sequences, transcribes audio into text, filters and corrects outliers in device data and text data, standardizes data format and units, and encrypts and desensitizes all data;

[0008] Data intelligence quantification and calculation module: Based on standardized data, it transforms operation-related data into operation skill quantification results through dynamic risk weighting algorithm, transforms medical records and dialogue texts into clinical thinking quantification results through fuzzy information association algorithm, and transforms communication audio and video and team logs into professional quality quantification results through interactive collaborative entropy algorithm;

[0009] Multi-dimensional dynamic evaluation module: Constructs a three-dimensional evaluation model that includes operational skills, clinical thinking, and professional qualities, sets up three-level evaluation cycles of real-time, stage, and final evaluation, combines AI pre-scoring and manual review mechanisms, and uses an adaptive dimension fusion algorithm to output a comprehensive score and level;

[0010] Personalized feedback and scenario-based reporting module: Generates learner's personal evaluation report and personalized improvement plan based on comprehensive evaluation results, outputs summary report adapted to medical colleges, hospital standardized training, and health administration scenarios, and supports data linkage with academic affairs system and HIS system.

[0011] Furthermore, the multi-source data acquisition module includes the following acquisition methods and data: acquiring data on operation trajectory, clamping force, puncture depth, suture distance, and operation speed via USB or TCP / IP protocol interface to connect to the clinical skills simulation device; acquiring medical history records, diagnostic reports, and treatment plan text data via HL7FHIR interface interface to connect to the electronic medical record system; acquiring video data of operation actions and body language via high-definition cameras deployed in the operation and communication areas; acquiring audio data of doctor-patient dialogue via microphone; and acquiring collaborative log data on team division of labor, emergency response time, and information transmission records via API interface connection to the training log platform.

[0012] Furthermore, the data standardization preprocessing module includes outlier handling and encryption / desensitization, which include:

[0013] The Z-score method is used to determine outlier values ​​in equipment data. When a data value deviates from the mean by 3 standard deviations, it is marked as an outlier. The correction method is to take the arithmetic mean of the five sampling points before and after the mean.

[0014] Missing key fields in text data are marked as missing, and duplicate diagnostic records retain the content of the first entry;

[0015] Storage encryption uses the AES-256 algorithm, and the key is rotated periodically;

[0016] The desensitization rule involves anonymizing learner identity information and patient information;

[0017] The specific implementation of the AES-256 algorithm includes: Algorithm parameters: a 256-bit key length, a 128-bit block length, a CBC working mode, and an initial vector of 128 random values ​​generated by a random number generator based on the NISTSP800-22 standard; Key generation: The key is generated by a random number generator based on CTR-DRBG, and the generation process incorporates the device hardware fingerprint as an entropy source to ensure key uniqueness.

[0018] Furthermore, in the data intelligent quantification calculation module, the expression for the dynamic risk weighting algorithm is: ,in, It is a quantitative score of operational skills. for Real-time operational precision This is the proficiency decay coefficient. For decay rate parameters, This is the start time of the operation. It is the end time of the operation. As for the operational risk level, for Time of the first Operational efficiency indicators This is the standard value for the indicator. For the accuracy-efficiency balance factor, This indicates the number of operational efficiency indicators.

[0019] Furthermore, in the data intelligent quantification calculation module, the expression for the fuzzy information association algorithm is: ,in, It is a quantitative score of clinical reasoning. For medical history information For diagnosis Information gain For diagnosis Compared with standard treatment pathways Fuzzy membership degree, This refers to the actual processing time. Standard processing time, For information-execution balance factor, It is the ratio of the relative deviation between the actual time and the standard time. This refers to the actual processing time. This is the standard processing time.

[0020] Furthermore, in the data intelligent quantification calculation module, the expression for the interactive collaborative entropy algorithm is: ,in, Professional competence quantification score, For learners of language Feedback from patients mutual information, For team collaboration entropy, For the maximum possible entropy, To ensure compliance with professional ethics, It serves as a communication-collaboration balancing factor.

[0021] Furthermore, the specific content of the three-dimensional evaluation model in the multi-dimensional dynamic evaluation module is as follows:

[0022] The operational skills dimension includes indicators of precision, efficiency, and stability. Precision involves needle distance error and puncture angle deviation; efficiency involves operation time and amount of operation per unit time; and stability involves the amplitude and force fluctuation of hand tremors.

[0023] Clinical thinking dimensions include indicators for medical history taking, diagnosis, and emergency management. Among them, medical history taking involves the coverage rate of core keywords, diagnosis involves the degree of matching with standard pathways, and emergency management involves the response speed to critical values.

[0024] The professional competence dimension includes indicators of doctor-patient communication, empathy, and teamwork. Among them, doctor-patient communication involves the proportion of empathetic language, empathy involves the proportion of positive emotions in patients, and teamwork involves the degree of division of labor and matching.

[0025] Furthermore, the AI ​​pre-scoring and manual review mechanism in the multi-dimensional dynamic evaluation module includes:

[0026] AI pre-scoring covers 70% of the indicators, including the accuracy, efficiency, and stability of operational skills; the completeness of medical history collection and the degree of matching between diagnosis and standard pathway in clinical thinking; the proportion of empathetic language and the degree of compliance with professional norms in professional ethics; and the scores are automatically calculated based on quantitative data.

[0027] Manual review covers 30% of complex scenarios, including the degree of tacit understanding in multidisciplinary collaboration, the innovativeness of clinical decision-making, and the rationality of handling rare cases. After reviewing the original audio and video, medical record text, and team logs through the system, examiners will correct the scores, and the specific reasons for the corrections must be recorded.

[0028] Furthermore, in the multi-dimensional dynamic evaluation module, the expression for the adaptive dimension fusion algorithm is: ,in, It is a three-dimensional comprehensive evaluation score. For the first Coefficient of variation of dimensional scores Quantify the score for the corresponding dimension. This is a scene correction factor. It is the sum of the reciprocals of the coefficients of variation of all dimensions. It is a dimension identifier. It is a skill of operation. It's clinical thinking. It is professional ethics; ≥90 points is excellent; 80 ≤ <90 points is good; 60 ≤ A score of 80 or below is considered passing. A score of 60 or below is considered unqualified.

[0029] Compared with existing technologies, this multi-dimensional evaluation and analysis system for clinical medical practice skills has the following beneficial effects:

[0030] I. This invention integrates multi-source data acquisition modules, linking clinical skills simulation equipment, electronic medical record systems, audio and video acquisition equipment, and training log platforms to achieve comprehensive data acquisition of operation trajectories, medical record texts, communication audio and video, and collaboration logs. This breaks through the limitations of traditional evaluations that rely solely on single data sources. Simultaneously, a data standardization preprocessing module completes data format unification, outlier correction, and encryption / desensitization processing, ensuring data quality while protecting the information security of learners and patients. Furthermore, through intelligent algorithms such as dynamic risk weighting, fuzzy information association, and interactive collaborative entropy, difficult-to-quantify abilities such as operational skills, clinical reasoning, and professional ethics are transformed into objective indicators. This addresses the problems of high subjective factors and one-sided dimensions in traditional evaluations, providing a more comprehensive and accurate quantitative basis for clinical medical practice skills evaluation and helping to efficiently identify learners' skill strengths and weaknesses.

[0031] Second, this invention constructs a three-dimensional evaluation model encompassing operational skills, clinical reasoning, and professional ethics, coupled with a three-tiered evaluation cycle of real-time, phased, and final assessments. This enables dynamic tracking of learners' skill development, overcoming the limitations of traditional evaluations that focus only on a single time point. Furthermore, it integrates AI pre-scoring with manual review mechanisms. AI efficiently covers most routine evaluation indicators, while manual review ensures accuracy in complex scenarios, balancing efficiency and precision. In addition, it generates personalized improvement plans and summary reports adapted to medical schools, hospital residency training, and health administration scenarios based on evaluation results. It also supports data linkage with academic affairs systems and HIS systems, providing learners with targeted improvement directions and decision support for different stakeholders. This promotes the upgrading of clinical medical practice skills evaluation from a single assessment to a full-process, personalized, and multi-scenario application, helping to improve the quality of clinical medical talent training and the level of practical teaching management.

[0032] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0034] Figure 1 A flowchart of a multi-dimensional evaluation and analysis system for clinical medical practice skills;

[0035] Figure 2A framework and data flow diagram for a multi-dimensional evaluation and analysis system for clinical medical practice skills. Detailed Implementation

[0036] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0037] Example 1:

[0038] Surgical suturing skills training and assessment for third-year students majoring in clinical medicine in five-year medical colleges.

[0039] Multi-source data acquisition module:

[0040] By connecting to the surgical suturing mannequin via USB protocol, data on students' suturing movements, stitch length, clamping force, and operating speed are collected. This data accurately captures key details of students' suturing actions, providing a concrete and analyzable data source for subsequent quantification of operational skills. Through the HL7FHIR interface, the system connects to the school's simulated electronic medical record system, collecting textual data such as simulated patient medical histories, preliminary diagnostic reports, and proposed suturing treatment plans. This comprehensively presents the students' collection and processing of case information, providing textual support for clinical reasoning evaluation. Two high-definition cameras are deployed in both the suturing operation area and the doctor-patient communication area to collect video data of students' suturing movements and body language. Simultaneously, audio data of dialogues between students and simulated patients regarding patient information and suturing risks is collected via microphones, comprehensively recreating the operation and communication scene and aiding in the assessment of communication skills and operational standardization in professional competence. Through the API interface, the system connects to the school's training log platform to collect team collaboration data, comprehensively understanding group collaboration performance and ensuring that professional competence evaluation includes the team collaboration dimension. Figure 1 As shown.

[0041] Data standardization preprocessing module:

[0042] The acquired suturing operation videos and doctor-patient communication videos were converted into continuous frame sequences, and the dialogue audio was transcribed into text format. This transformed unstructured audio and video data into structured data suitable for subsequent algorithmic processing, laying the foundation for quantitative calculations. For equipment data such as operation speed and clamping force obtained from the suturing manipulator, the Z-score method was used to identify outliers. If a data value deviated from the mean by 3 standard deviations, it was marked as an outlier. The correction method involved replacing the outlier data with the arithmetic mean of the five sampling points before and after it, eliminating invalid data and correcting deviations to ensure the accuracy of the equipment data and avoid outliers interfering with subsequent quantitative results. For text data in the simulated electronic medical record system, if the medical history record lacked "injury time" or "past medical history,"... Key fields such as "surgical history" are marked as missing. If there are duplicate entries in the diagnostic report, the first entry is retained to ensure the integrity and uniqueness of the text data, providing reliable material for the quantification of clinical reasoning. All processed data is stored and encrypted using the AES-256 algorithm. This algorithm uses a 256-bit key length, a 128-bit block length, and operates in CBC mode. The initial vector is a 128-bit random value. The key is generated by a CTR-DRBG-based random number generator. The generation process incorporates the hardware fingerprint of the training equipment as an entropy source. At the same time, the student's name, student ID, and the virtual identity information of the simulated patient are anonymized and desensitized to ensure data storage security and meet privacy protection requirements, avoiding the leakage of personal information.

[0043] Data intelligent quantification calculation module:

[0044] Based on standardized data, a dynamic risk-weighted algorithm is used to transform operation-related data such as suture trajectory, stitch length, force, and speed into quantitative results of operational skills. The expression for the dynamic risk-weighted algorithm is as follows: ,in, It is a quantitative score of operational skills. for Real-time operational precision This is the proficiency decay coefficient. For decay rate parameters, This is the start time of the operation. It is the end time of the operation. As for the operational risk level, for Time of the first Operational efficiency indicators This is the standard value for the indicator. For the accuracy-efficiency balance factor, This algorithm, which combines the risk level and efficiency indicators of suturing operations, objectively reflects the accuracy, proficiency, and risk control capabilities of students' suturing skills. It employs a fuzzy information association algorithm to transform medical history records, diagnostic reports, and transcribed doctor-patient communication texts from simulated electronic medical records into quantitative results of clinical reasoning. The expression for the fuzzy information association algorithm is: ,in, It is a quantitative score of clinical reasoning. For medical history information For diagnosis Information gain For diagnosis Compared with standard treatment pathways Fuzzy membership degree, This refers to the actual processing time. Standard processing time, For information-execution balance factor, It is the ratio of the relative deviation between the actual time and the standard time. This refers to the actual processing time. This is the standard processing time, which, by linking medical history, diagnosis, and communication content, accurately measures the logic, completeness, and alignment with standard treatment pathways of students' clinical thinking. Using an interactive collaborative entropy algorithm, it transforms collaborative log data collected from the training log platform, including audio and video recordings of doctor-patient communication, team division of labor, emergency response time, and information transmission records, into quantitative results of professional competence. The expression for the interactive collaborative entropy algorithm is: ,in, Professional competence quantification score, For learners of language Feedback from patients mutual information, For team collaboration entropy, For the maximum possible entropy, To ensure compliance with professional ethics, As a communication-collaboration balancing factor, it comprehensively evaluates students' doctor-patient communication effectiveness, empathy ability, and teamwork smoothness, fully reflecting their performance in all dimensions of professional competence.

[0045] Multi-dimensional dynamic evaluation module:

[0046] A three-dimensional evaluation model is constructed, encompassing operational skills, clinical reasoning, and professional ethics. The operational skills dimension includes indicators of accuracy, efficiency, and stability; the clinical reasoning dimension includes indicators of history taking, diagnosis, and emergency response; and the professional ethics dimension includes indicators of doctor-patient communication, empathy, and teamwork. This clearly defines the evaluation scope, ensuring comprehensive coverage of the core competencies required for students' surgical suturing training. A three-tiered evaluation cycle—real-time, phased, and final—is established to track the entire training process, enabling timely identification of real-time problems and periodic assessments of learning outcomes, ultimately providing a comprehensive evaluation of ability levels. AI pre-scoring covers 70% of the indicators, automatically calculating scores based on quantitative data, rapidly processing large amounts of standardized data, improving evaluation efficiency, and reducing repetitive manual work. Manual review covers 30% of complex scenarios, where surgical training instructors review original operation videos, doctor-patient communication audio and video, simulated medical records, and team logs through the system to correct scores and record the reasons for corrections. This relies on the instructors' professional experience to solve complex problems that AI struggles to accurately assess, ensuring the accuracy and fairness of the evaluation results. Finally, an adaptive dimension fusion algorithm is used to output the student's comprehensive score and grade. The expression for the adaptive dimension fusion algorithm is: ,in, It is a three-dimensional comprehensive evaluation score. For the first Coefficient of variation of dimensional scores Quantify the score for the corresponding dimension. This is a scene correction factor. It is the sum of the reciprocals of the coefficients of variation of all dimensions. It is a dimension identifier. It is a skill of operation. It's clinical thinking. It is professional ethics; ≥90 points is excellent; 80 ≤ <90 points is good; 60 ≤ A score of 80 or below is considered passing. A score of 60 or below is considered unqualified. The weights will be adjusted based on the variation in data across each dimension to make the comprehensive evaluation more closely reflect the students' actual abilities and provide a clear and scientific basis for subsequent feedback.

[0047] Personalized feedback and scenario-based reporting modules:

[0048] Based on the comprehensive evaluation results, a personal evaluation report is generated for each student. The report clearly points out their strengths and weaknesses in areas such as suturing precision, clinical thinking, and professional competence, and provides a personalized improvement plan to help students clearly identify their shortcomings, clarify their future learning direction, and target their weak areas for improvement. A summary report adapted to the practical training management scenario of medical colleges is also generated, including the overall pass rate of surgical suturing skills for all students in that grade, average scores in each dimension, and statistics on common problems. This provides the school's practical training management department with overall teaching effectiveness data, helping to optimize practical training curriculum design and adjust teaching methods. Furthermore, the system supports integration with the school's academic affairs system, synchronizing student practical training evaluation data to the system, enabling linked management of practical training grades and course credits, simplifying the academic performance calculation process, and improving the efficiency of the connection between practical training management and academic affairs.

[0049] In summary, in the surgical suturing skills training assessment for third-year students in the five-year clinical medicine program at medical colleges, this system operates systematically in modules: the multi-source data acquisition module comprehensively acquires operational, medical record, audio-visual, and collaborative data, providing basic materials for evaluation; the data standardization and preprocessing module converts data formats, corrects outliers, and encrypts and desensitizes data to ensure usability and security; the intelligent data quantification calculation module uses algorithms such as dynamic risk weighting to transform data into quantitative results across three dimensions; the multi-dimensional dynamic evaluation module achieves comprehensive assessment through a three-dimensional model, three-level cycle, and AI + human review; and the personalized feedback and scenario-based reporting module generates exclusive reports and school-wide summary reports and links them to the academic affairs system. Overall, it covers the core competency evaluation of the training, providing scientific and efficient support for student skills improvement and school training management.

[0050] Example 2:

[0051] Evaluation of standardized training for resident physicians in tertiary hospitals in internal medicine patient reception and team collaboration.

[0052] Multi-source data acquisition module:

[0053] By connecting to internal medicine simulation equipment via TCP / IP protocol, data such as the resident physician's operational trajectory, intravenous puncture depth, auscultation duration, and ECG monitoring data reading speed during the consultation of simulated patients are collected. This accurately records key parameters of the resident physician's diagnostic and treatment operations, providing a realistic and detailed data source for quantifying operational skills and ensuring that evaluations are based on actual diagnostic and treatment details. Furthermore, by connecting to the hospital's electronic medical record system via the HL7FHIR interface, textual data such as the simulated case's real medical history, preliminary diagnosis report, and proposed treatment plan are collected. This provides a complete picture of the resident physician's analysis and processing of case information, offering clinical reasoning evaluations that closely reflect real-world diagnostic and treatment scenarios. Textual support; One high-definition camera is deployed in each of the internal medicine treatment room and the doctor-patient communication area to collect video data of the resident physician's diagnostic and treatment actions and body language during communication with simulated patients. Simultaneously, audio data of their conversations regarding patient inquiries, treatment explanations, and prognosis is collected via microphones. This comprehensively recreates the diagnostic and communication process, providing a direct basis for assessing the standardization of communication and the rigor of operations in professional competence evaluation. Through an API interface connected to the hospital's residency training log platform, team collaboration data is collected to fully understand the rationality of division of labor, emergency response efficiency, and accuracy of information transmission in multidisciplinary collaboration. This ensures that professional competence evaluation covers the core elements of team collaboration, such as... Figure 2 As shown.

[0054] Data standardization preprocessing module:

[0055] The collected diagnostic and treatment operation videos and doctor-patient communication videos are converted into continuous frame sequences, and the dialogue audio is transcribed into text format. This transforms unstructured audio and video data into structured data suitable for subsequent quantitative analysis, laying the foundation for the algorithmic processing of the data intelligent quantitative calculation module. For equipment data such as puncture depth, auscultation duration, and data reading speed obtained from internal medicine simulation diagnostic and treatment equipment, the Z-score method is used to identify outliers. If a data value deviates from the mean by 3 standard deviations, it is marked as an anomaly. The correction method is to replace it with the arithmetic mean of the five sampling points before and after the anomaly, eliminating invalid data caused by temporary equipment fluctuations or operational errors, ensuring the reliability of equipment data, and avoiding interference from anomaly data with the quantitative results of operational skills. For text data in the electronic medical record system, if the medical history record lacks "blood pressure control,"... Key fields such as "treatment history" and "drug allergy history" were marked as missing. If there were duplicate entries for the treatment plan, the first entry was retained to ensure the integrity and accuracy of the medical record text data, providing high-quality text materials for the quantification of clinical thinking. All processed data were stored and encrypted using the AES-256 algorithm. This algorithm uses a 256-bit key length, a 128-bit block length, and operates in CBC mode. The initial vector is a 128-bit random value. The key is generated by a CTR-DRBG-based random number generator. The generation process incorporates the hardware fingerprint of the hospital's standardized training equipment as an entropy source. At the same time, the names of resident physicians, employee numbers, and standardized patient identity information are anonymized and desensitized to strictly comply with the hospital's data security management regulations, ensure that personnel privacy information is not leaked, and ensure the security of data storage.

[0056] Data intelligent quantification calculation module:

[0057] Based on standardized data, a dynamic risk-weighted algorithm is used to transform resident physicians' diagnostic and treatment procedures, puncture depths, auscultation durations, and other operation-related data into quantitative results of operational skills. The expression for the dynamic risk-weighted algorithm is as follows: This algorithm can objectively assess the accuracy, proficiency, and control over the pace of internal medicine diagnosis and treatment procedures by combining risk levels and efficiency indicators. It employs a fuzzy information association algorithm to transform medical history records, diagnostic reports, treatment plans, and transcribed doctor-patient communications from electronic medical records into quantitative results of clinical reasoning. The expression for the fuzzy information association algorithm is as follows: By logically linking key information in medical history, diagnostic conclusions, and treatment plans, this study accurately measures the rigor of resident physicians' clinical thinking and its alignment with the standard diagnostic and treatment pathways for coronary heart disease. Employing an interactive collaborative entropy algorithm, it transforms collaborative log data—including audio and video recordings of doctor-patient communication and data collected from the residency training log platform such as team division of labor, emergency response time, and information transmission records—into quantitative results of professional competence. The expression for the interactive collaborative entropy algorithm is: The evaluation comprehensively assesses the effectiveness of resident physicians' communication with patients, their level of attention to patients' emotions, and their cooperation in teamwork, thus fully reflecting the actual level of each dimension of professional competence.

[0058] Multi-dimensional dynamic evaluation module:

[0059] A three-dimensional evaluation model was constructed, encompassing operational skills, clinical reasoning, and professional ethics. The operational skills dimension includes indicators of accuracy, efficiency, and stability; the clinical reasoning dimension includes indicators of history taking, diagnosis, and emergency management; and the professional ethics dimension includes indicators of doctor-patient communication, empathy, and teamwork. This comprehensive model covers all core competencies required for internal medicine patient care, ensuring no dimension is overlooked. A three-tiered evaluation cycle—real-time, phased, and final—was established to track the entire residency training process, enabling real-time correction of operational deviations, phased assessment of skill improvement, and a comprehensive evaluation of training effectiveness. AI-based pre-scoring was employed, covering 70% of the evaluation. The system automatically calculates scores based on quantitative data, efficiently processing large amounts of standardized data, reducing the workload of the assessment expert group, and improving evaluation efficiency. Manual review covers 30% of complex scenarios; the hospital's residency training assessment expert group reviews original treatment videos, doctor-patient communication audio and video, electronic medical records, and team logs through the system to correct scores and record the reasons for corrections. This relies on the experts' rich clinical experience to solve complex problems that AI struggles to accurately assess, enhancing the professionalism and credibility of the evaluation results. Finally, an adaptive dimension fusion algorithm is used to output the comprehensive score and grade of the resident physicians. The expression for the adaptive dimension fusion algorithm is: ; ≥90 points is excellent; 80 ≤ <90 points is good; 60 ≤ A score of 80 or below is considered passing. A score of 60 or below is considered unqualified. The weights of each dimension are adjusted according to the specific characteristics of internal medicine diagnosis and treatment scenarios, so that the comprehensive evaluation is more in line with actual clinical needs and provides a scientific and objective basis for the residency training assessment.

[0060] Personalized feedback and scenario-based reporting modules:

[0061] Based on the comprehensive evaluation results, a personal evaluation report is generated for each resident physician, identifying their problems in internal medicine patient reception procedures, clinical reasoning, and professional ethics, and providing personalized improvement plans to help resident physicians accurately pinpoint their weaknesses, identify targeted improvement measures, and accelerate their professional development. A summary report adapted to the hospital's standardized residency training management scenario is also output, including the overall pass rate of internal medicine patient reception skills for that batch of resident physicians, average scores for each dimension, and statistics on common collaborative issues. This provides the hospital's standardized residency training management department with batch-specific competency data, helping to optimize the standardized residency training plan, adjust faculty allocation, and training focus. Furthermore, the system supports integration with the hospital's HIS system, synchronizing resident physicians' standardized residency training evaluation data to the HIS system, enabling linked management of standardized residency training results and professional qualification assessment and recommendation, simplifying the connection process between standardized residency training assessment and qualification recommendation, and improving the standardization and efficiency of the hospital's standardized residency training work.

[0062] In summary, in the evaluation of residency training in internal medicine at tertiary hospitals, the system's modules precisely meet the needs: the multi-source data acquisition module collects data on diagnostic and treatment procedures, medical records, audio and video recordings, and collaborative data, aligning with actual clinical training practices; the data standardization and preprocessing module processes data and ensures security, guaranteeing data reliability; the intelligent data quantification module quantifies capabilities across three dimensions using specified algorithms; the multi-dimensional dynamic evaluation module achieves accurate assessment through a three-dimensional model, a three-level cycle, and AI + human review; and the personalized feedback and scenario-based reporting module outputs customized solutions and hospital-wide summary reports, linking with the HIS system. This system not only helps resident physicians identify their weaknesses but also provides objective support for hospital training management and assessment.

[0063] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A multi-dimensional evaluation and analysis system for clinical medical practice skills, characterized in that, The system includes: Multi-source data acquisition module: By connecting to clinical skills simulation equipment and electronic medical record system, deploying high-definition cameras and microphones to collect operation action videos and doctor-patient communication audio and video, and linking with training log platform to collect team collaboration data, obtain operation trajectory, medical record text, audio and video streams and collaboration logs; Data standardization preprocessing module: Converts video into frame sequences, transcribes audio into text, filters and corrects outliers in device data and text data, standardizes data format and units, and encrypts and desensitizes all data; Data intelligence quantification and calculation module: Based on standardized data, it transforms operation-related data into operation skill quantification results through dynamic risk weighting algorithm, transforms medical records and dialogue texts into clinical thinking quantification results through fuzzy information association algorithm, and transforms communication audio and video and team logs into professional quality quantification results through interactive collaborative entropy algorithm; Multi-dimensional dynamic evaluation module: Constructs a three-dimensional evaluation model that includes operational skills, clinical thinking, and professional qualities, sets up three-level evaluation cycles of real-time, stage, and final evaluation, combines AI pre-scoring and manual review mechanisms, and uses an adaptive dimension fusion algorithm to output a comprehensive score and level; Personalized feedback and scenario-based reporting module: Generates learner's personal evaluation report and personalized improvement plan based on comprehensive evaluation results, outputs summary report adapted to medical colleges, hospital standardized training, and health administration scenarios, and supports data linkage with academic affairs system and HIS system.

2. The multi-dimensional evaluation and analysis system for clinical medical practice skills according to claim 1, characterized in that, The multi-source data acquisition module includes the following acquisition methods and data: It connects to the clinical skills simulation equipment via USB or TCP / IP protocol to acquire data on operation trajectory, clamping force, puncture depth, suture distance, and operation speed; it connects to the electronic medical record system via the HL7FHIR interface to acquire text data of medical history records, diagnostic reports, and treatment plans; a high-definition camera is deployed in the operation area and communication area to acquire video data of operation actions and body language; a microphone simultaneously acquires audio data of doctor-patient dialogue; and it connects to the training log platform via API interface to acquire collaborative log data on team division of labor, emergency response time, and information transmission records.

3. The multi-dimensional evaluation and analysis system for clinical medical practice skills according to claim 1, characterized in that, The data standardization preprocessing module includes outlier handling and encryption / desensitization: The Z-score method is used to determine outlier values ​​in equipment data. When a data value deviates from the mean by 3 standard deviations, it is marked as an outlier. The correction method is to take the arithmetic mean of the five sampling points before and after the mean. Missing key fields in text data are marked as missing, and duplicate diagnostic records retain the content of the first entry; Storage encryption uses the AES-256 algorithm, and the key is rotated periodically; The desensitization rule involves anonymizing learner identity information and patient information.

4. The multi-dimensional evaluation and analysis system for clinical medical practice skills according to claim 1, characterized in that, In the data intelligent quantification calculation module, the expression for the dynamic risk weighting algorithm is: ,in, It is a quantitative score of operational skills. for Real-time operational precision This is the proficiency decay coefficient. For decay rate parameters, This is the start time of the operation. It is the end time of the operation. As for the operational risk level, for Time of the first Operational efficiency indicators This is the standard value for the indicator. For the accuracy-efficiency balance factor, This indicates the number of operational efficiency indicators.

5. The multi-dimensional evaluation and analysis system for clinical medical practice skills according to claim 1, characterized in that, In the data intelligent quantification calculation module, the expression for the fuzzy information association algorithm is: ,in, It is a quantitative score of clinical reasoning. For medical history information For diagnosis Information gain For diagnosis Compared with standard treatment pathways Fuzzy membership degree, This refers to the actual processing time. Standard processing time, For information-execution balance factor, It is the ratio of the relative deviation between the actual time and the standard time. This refers to the actual processing time. This is the standard processing time.

6. The multi-dimensional evaluation and analysis system for clinical medical practice skills according to claim 1, characterized in that, In the data intelligent quantification calculation module, the expression for the interactive collaborative entropy algorithm is: ,in, Professional competence quantification score, For learners of language Feedback from patients mutual information, For team collaboration entropy, For the maximum possible entropy, To ensure compliance with professional ethics, It serves as a communication-collaboration balancing factor.

7. The multi-dimensional evaluation and analysis system for clinical medical practice skills according to claim 1, characterized in that, The specific content of the three-dimensional evaluation model in the multi-dimensional dynamic evaluation module is as follows: The operational skills dimension includes indicators of precision, efficiency, and stability. Precision involves needle distance error and puncture angle deviation; efficiency involves operation time and amount of operation per unit time; and stability involves the amplitude and force fluctuation of hand tremors. Clinical thinking dimensions include indicators for medical history taking, diagnosis, and emergency management. Among them, medical history taking involves the coverage rate of core keywords, diagnosis involves the degree of matching with standard pathways, and emergency management involves the response speed to critical values. The professional competence dimension includes indicators of doctor-patient communication, empathy, and teamwork. Among them, doctor-patient communication involves the proportion of empathetic language, empathy involves the proportion of positive emotions in patients, and teamwork involves the degree of division of labor and matching.

8. The multi-dimensional evaluation and analysis system for clinical medical practice skills according to claim 1, characterized in that, The multi-dimensional dynamic evaluation module includes an AI pre-scoring and manual review mechanism, which comprises: AI pre-scoring covers 70% of the indicators, including the accuracy, efficiency, and stability of operational skills; the completeness of medical history collection and the degree of matching between diagnosis and standard pathway in clinical thinking; the proportion of empathetic language and the degree of compliance with professional norms in professional ethics; and the scores are automatically calculated based on quantitative data. Manual review covers 30% of complex scenarios, including the degree of tacit understanding in multidisciplinary collaboration, the innovativeness of clinical decision-making, and the rationality of handling rare cases. After reviewing the original audio and video, medical record text, and team logs through the system, examiners will correct the scores, and the specific reasons for the corrections must be recorded.

9. The multi-dimensional evaluation and analysis system for clinical medical practice skills according to claim 1, characterized in that, In the multi-dimensional dynamic evaluation module, the expression for the adaptive dimension fusion algorithm is: ,in, It is a three-dimensional comprehensive evaluation score. For the first Coefficient of variation of dimensional scores Quantify the score for the corresponding dimension. This is a scene correction factor. It is the sum of the reciprocals of the coefficients of variation of all dimensions. It is a dimension identifier. It is a skill of operation. It's clinical thinking. It is professional ethics; ≥90 points is excellent; 80 ≤ <90 points is good; 60 ≤ A score of 80 or below is considered passing. A score of 60 or below is considered unqualified.

Images