Medical level assessment method and system based on clustering analysis and regression analysis
Through cluster analysis and regression analysis methods, the problems of insufficient data analysis depth and feedback lag in the existing medical level evaluation methods are solved, accurate assessment and personalized feedback of medical level are achieved, and the efficiency and quality of medical training are improved.
Patent Information
- Application Number
- CN202411043769.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2024-11-22
AI Technical Summary
The existing medical level evaluation methods have shortcomings in the depth of data analysis and feedback mechanism, which leads to inaccurate evaluation results and lag in feedback, which affects learning efficiency and effectiveness.
Using clustering analysis and regression analysis methods, we collect and preprocess the evaluation index data, perform feature extraction, clustering processing and regression processing, and generate personalized feedback suggestions to improve the objectivity and accuracy of the evaluation.
It has achieved a comprehensive and objective assessment of the medical level, provided immediate personalized feedback, and significantly improved the efficiency and quality of medical training.
Smart Images

Figure CN119004150A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of medical training technology, and in particular to a medical level evaluation method and system based on cluster analysis and regression analysis. Background Art
[0002] In medical training, cultivating medical talents with superb equipment operation skills and accurate diagnosis capabilities is the key to improving the quality of medical services. For a long time, the assessment of trainees' skill levels in medical training has relied on the traditional direct evaluation model of instructors. This model is not only limited by the instructor's subjective experience and observation perspective, but also difficult to achieve a comprehensive and objective assessment of the trainees' skill levels. In addition, because the assessment process is time-consuming and inefficient, it often fails to cover all trainees and every subtle aspect of their skills, resulting in a certain degree of one-sidedness and inaccuracy in the assessment results.
[0003] With the development of virtual reality (VR) and augmented reality (AR) technologies, more and more medical training uses these technologies to simulate real medical environments, providing trainees with a highly immersive operation experience while accurately recording every detail of the trainees' operation, including motion trajectory, reaction time, decision-making process and other operation data, providing a rich data source for subsequent evaluation and analysis.
[0004] However, the existing medical level assessment methods based on VR / AR technology have the following shortcomings: First, the data analysis is not deep enough. Most systems only stay at the simple statistical level of the data and fail to deeply explore the information and patterns hidden behind the data, resulting in inaccurate assessment results; second, the feedback mechanism is lagging, and it is difficult for students to obtain personalized feedback and improvement suggestions based on data analysis in a timely manner during the operation, which affects learning efficiency and effectiveness. Summary of the invention
[0005] In order to solve the above technical problems, the present invention provides a medical level evaluation method and system based on cluster analysis and regression analysis.
[0006] The technical problem solved by the present invention can be achieved by adopting the following technical solutions:
[0007] A medical level evaluation method based on cluster analysis and regression analysis, with at least one evaluation index related to medical level pre-set;
[0008] The medical level assessment method includes:
[0009] Step S1, collecting data related to each evaluation indicator of the student to be evaluated, and preprocessing, to obtain the data to be evaluated corresponding to each evaluation indicator;
[0010] Step S2, extracting features from the data to be evaluated corresponding to each of the evaluation indicators to obtain feature information corresponding to each of the evaluation indicators;
[0011] Step S3, clustering the feature information corresponding to each evaluation indicator to obtain a clustering result corresponding to each evaluation indicator;
[0012] Step S4, performing regression processing on the feature information corresponding to each evaluation indicator to obtain a regression result corresponding to each evaluation indicator;
[0013] Step S5, performing a comprehensive evaluation based on the clustering results and regression results corresponding to the evaluation indicators of all items, and generating personalized feedback suggestions.
[0014] Preferably, the evaluation indicators include medical equipment operation proficiency and / or diagnostic accuracy.
[0015] Preferably, the evaluation index includes proficiency in operating medical equipment, and the data related to the evaluation index is operation data, and the operation data includes at least one of the trainee's hand motion data or equipment interaction records;
[0016] The evaluation index includes diagnostic accuracy, and the data related to the evaluation index is diagnostic data, which includes medical image data and diagnostic records generated by the trainee during the diagnosis process.
[0017] Preferably, in step S2, the evaluation index includes the proficiency of medical device operation, and the characteristic information corresponding to the evaluation index is an operation feature, and the operation feature includes at least one of a hand motion feature or a device interaction feature;
[0018] The evaluation index includes diagnostic accuracy, and the feature information corresponding to the evaluation index is a diagnostic feature, and the diagnostic feature includes at least one of an image feature or a diagnostic record feature.
[0019] Preferably, in step S3, the clustering process is a K-means clustering process, comprising:
[0020] Step S31A, randomly selecting K initial centroids from the feature information corresponding to the evaluation indicators of all items; wherein K is the number of clusters;
[0021] Step S32A, assigning each feature sample in the feature information to the cluster with the nearest centroid, to form K clusters;
[0022] Step S33A, calculating the centroid of each cluster and updating the centroid position;
[0023] Step S34A, repeating steps S32A-S33A until the centroid position no longer changes or a preset number of iterations is reached.
[0024] Preferably, in step S3, the clustering process is a hierarchical clustering process, comprising:
[0025] Step S31B, treating each feature sample in the feature information corresponding to the evaluation indicators of all items as an independent cluster;
[0026] Step S32B, calculating the distances between all current clusters and constructing a distance matrix;
[0027] Step S33B, merging the two closest clusters into a new cluster;
[0028] Step S34B, recalculating the distance between the new cluster and other clusters, and updating the distance matrix;
[0029] Step S35B, repeating steps S32B-S34B until all the feature samples are merged into one cluster.
[0030] Preferably, in step S4, a linear regression model is used for regression processing;
[0031] Among them, the linear regression model uses the feature information corresponding to each evaluation indicator in the training data set as the model input, and the scoring result corresponding to the evaluation indicator as the model output. The regression coefficient and bias term are determined after training, the mean square error is used as the loss function, and the gradient descent method is used to update the regression coefficient and bias term; finally, the test data set is used to test the model performance.
[0032] Preferably, in step S4, a multivariate regression model is used for regression processing;
[0033] Among them, the multivariate regression model uses the feature information corresponding to the evaluation indicators of all items in the training data set as the model input, and the scoring results corresponding to the evaluation indicators as the model output. The regression coefficient and bias term are determined after training, the mean square error is used as the loss function, and the gradient descent method is used to update the regression coefficient and bias term; finally, the test data set is used to test the model performance.
[0034] Preferably, the step S5 comprises:
[0035] Step S51, performing weighted average processing on the clustering result and the regression result to obtain a comprehensive score of the student to be evaluated, and generating a comprehensive evaluation report;
[0036] Step S52, comparing the feature information corresponding to each of the evaluation indicators with the corresponding preset standard features, and generating error description information according to the comparison result;
[0037] Step S53: matching the misalignment description information with a preset suggestion template library to generate the personalized feedback suggestion.
[0038] The present invention also provides a medical level evaluation system based on cluster analysis and regression analysis, which is used to implement the medical level evaluation method based on cluster analysis and regression analysis as described above, and the system also includes:
[0039] The medical level assessment method is pre-set with at least one assessment index related to the medical level; and further includes:
[0040] The data collection and preprocessing module is used to collect the data related to each evaluation indicator of the students to be evaluated, and perform preprocessing to obtain the data to be evaluated corresponding to each evaluation indicator;
[0041] A feature extraction module, connected to the data collection and preprocessing module, is used to extract features from the data to be evaluated corresponding to each evaluation indicator, and obtain feature information corresponding to each evaluation indicator;
[0042] A cluster analysis module, connected to the feature extraction module, is used to perform clustering processing on the feature information corresponding to each of the evaluation indicators to obtain a clustering result corresponding to each of the evaluation indicators;
[0043] A regression analysis module, connected to the feature extraction module, for performing regression processing on the feature information corresponding to each evaluation indicator to obtain a regression result corresponding to each evaluation indicator;
[0044] The comprehensive evaluation module is connected to the cluster analysis module and the regression analysis module respectively, and is used to perform a comprehensive evaluation based on the clustering results and regression results corresponding to all the evaluation indicators, and generate personalized feedback suggestions.
[0045] The advantages or beneficial effects of the technical solution of the present invention are:
[0046] The present invention accurately evaluates the skill level of trainees based on cluster analysis and regression analysis algorithms, improves the objectivity and accuracy of the evaluation, realizes the instant generation of personalized feedback suggestions, and significantly improves the efficiency and quality of medical training. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] Figure 1 A schematic diagram of a process flow of a medical level assessment method based on cluster analysis and regression analysis in a preferred embodiment of the present invention;
[0048] Figure 2 A schematic diagram of a process flow of K-means clustering processing in a preferred embodiment of the present invention;
[0049] Figure 3 A schematic diagram of the process of hierarchical clustering processing in a preferred embodiment of the present invention;
[0050] Figure 4 A schematic diagram of a comprehensive evaluation process in a preferred embodiment of the present invention;
[0051] Figure 5 This is a structural block diagram of a medical level evaluation system based on cluster analysis and regression analysis in a preferred embodiment of the present invention. DETAILED DESCRIPTION
[0052] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0053] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.
[0054] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.
[0055] In a preferred embodiment of the present invention, based on the above problems existing in the prior art, a medical level evaluation method based on cluster analysis and regression analysis is provided, which is pre-set with at least one evaluation index related to medical level, including but not limited to medical equipment operation proficiency and / or diagnostic accuracy. Figure 1 As shown, the medical level assessment methods include:
[0056] Step S1, collecting data related to each evaluation indicator of the student to be evaluated, and preprocessing, to obtain the data to be evaluated corresponding to each evaluation indicator;
[0057] Step S2, extracting features from the data to be evaluated corresponding to each evaluation indicator to obtain feature information corresponding to each evaluation indicator;
[0058] Step S3, clustering the feature information corresponding to each evaluation indicator to obtain a clustering result corresponding to each evaluation indicator;
[0059] Step S4, performing regression processing on the feature information corresponding to each evaluation indicator to obtain a regression result corresponding to each evaluation indicator;
[0060] Step S5, perform a comprehensive evaluation based on the clustering results and regression results corresponding to all evaluation indicators, and generate personalized feedback suggestions.
[0061] Specifically, in this embodiment, based on data analysis algorithms such as cluster analysis and regression analysis, the skill level of trainees is accurately evaluated, and a comprehensive and objective evaluation of the medical level of trainees is achieved, and personalized feedback suggestions are generated for each trainee based on the evaluation results. Specific improvement measures and training suggestions are proposed for the deficiencies of trainees in specific evaluation indicators to help trainees improve their medical skills in a targeted manner and improve the efficiency and quality of medical training.
[0062] As a preferred embodiment, the evaluation index includes medical equipment operation proficiency and / or diagnostic accuracy.
[0063] Specifically, medical equipment operation proficiency is used to measure the trainees' proficiency in operating (including actual or virtual operation) various medical equipment, including but not limited to surgical instruments, imaging diagnostic equipment, therapeutic instruments, etc. The trainees' operation proficiency is comprehensively evaluated by recording data such as hand movement data and equipment interaction records during the operation.
[0064] Diagnostic accuracy is used to evaluate the accuracy and judgment ability of students in the medical diagnosis process. By simulating real cases or using historical case image data, students are required to analyze the condition, make diagnostic judgments, and compare them with their actual diagnostic results to calculate the diagnostic accuracy.
[0065] Evaluation indicators can be flexibly set according to actual needs. Specifically, other related indicators can be added as needed to evaluate the medical level, such as case analysis ability, patient communication ability, etc.
[0066] As a preferred embodiment, the evaluation index includes the proficiency of medical device operation, and the data related to the evaluation index is operation data, and the operation data includes at least one of the trainee's hand movement data or device interaction records;
[0067] The evaluation indicators include diagnostic accuracy, and the data related to the evaluation indicators are diagnostic data, which include medical image data and diagnostic records generated by the trainees during the diagnosis process.
[0068] Specifically, for each student to be evaluated, the evaluation indicator can be the proficiency in operating medical equipment, the diagnostic accuracy, or a comprehensive evaluation of the combination of proficiency in operating medical equipment and diagnostic accuracy.
[0069] If the evaluation indicator is the proficiency in operating medical equipment, the data related to the evaluation indicator is the operation data, and the operation data includes at least one of the trainee's hand movement data or the equipment interaction record.
[0070] Specifically, hand motion data includes but is not limited to operation trajectory, speed and force data, which can be captured and recorded using high-precision sensors. Device interaction records are the interaction data between trainees and virtual medical devices, including but not limited to key operations, probe selection and parameter adjustment.
[0071] If the evaluation indicator is diagnostic accuracy, the data related to the evaluation indicator is diagnostic data, which includes medical image data (such as ultrasound images, CT scan images) and diagnostic records generated by the trainees during the diagnosis process.
[0072] After completing data collection, data preprocessing steps are performed. Preprocessing steps include but are not limited to data cleaning and data normalization. Data cleaning removes noise data, outliers, and invalid data to ensure the purity and accuracy of the data used for analysis. Considering the possible dimensional differences between different evaluation indicators, the operation data and diagnostic data are normalized to convert all data to the same measurement to ensure data consistency and comparability.
[0073] As a preferred embodiment, in step S2, the evaluation index includes the proficiency of medical device operation, and the characteristic information corresponding to the evaluation index is the operation feature, and the operation feature includes at least one of a hand movement feature or a device interaction feature;
[0074] The evaluation index includes diagnostic accuracy, and the feature information corresponding to the evaluation index is a diagnostic feature, and the diagnostic feature includes at least one of an image feature or a diagnostic record feature.
[0075] Specifically, through the feature extraction step, key features are extracted from the operation data related to the proficiency of medical equipment operation and the diagnosis data related to the diagnosis accuracy to reveal the key information and patterns hidden in the data.
[0076] For operation data related to the proficiency of medical device operation, the corresponding feature information is the operation feature, which includes but is not limited to hand movement features and device interaction features. Among them, hand movement features include but are not limited to trajectory features, force features and time features; trajectory features include but are not limited to the curvature, path length, average speed, acceleration and other features of the operation trajectory; force features include but are not limited to the mean and variance of the force applied during the operation; time features include but are not limited to operation time distribution features such as operation time and reaction time.
[0077] The device interaction features include, but are not limited to, key operation features, probe selection features, and parameter adjustment features. Among them, key operation features include, but are not limited to key frequency, key sequence, wrong key rate, response time, etc.; probe selection features include, but are not limited to probe switching times, probe selection accuracy, and probe usage time, etc.; parameter adjustment features include, but are not limited to parameter adjustment times, etc.
[0078] For diagnostic data related to diagnostic accuracy, the corresponding feature information is the diagnostic feature, which includes but is not limited to image features and diagnostic record features. Among them, image features can use deep learning models, such as Residual Neural Network (ResNet) and U-Net models, to extract features of medical images (such as ultrasound images, CT scan images) and identify key features such as lesion areas, boundaries, and shapes in the images. Diagnostic record features can be derived from key features in the trainees' diagnostic records, including but not limited to the marked position of the lesion (coordinates or area in the image), measurement data (such as the diameter and area of the lesion), and diagnostic conclusions.
[0079] Each evaluation indicator generates a corresponding set of feature information, which will be used as input data for subsequent model training and evaluation to assess trainees' abilities in medical equipment operation proficiency and diagnostic accuracy.
[0080] Furthermore, the features corresponding to the operation data and the diagnosis data are clustered to identify the different proficiency and accuracy levels of the trainees. The clustering may be performed using K-means clustering or hierarchical clustering.
[0081] As a preferred embodiment, wherein Figure 2 As shown, in step S3, the clustering process is a K-means clustering process, including:
[0082] Step S31A, randomly selecting K initial centroids from the feature information corresponding to all evaluation indicators; where K is the number of clusters;
[0083] Step S32A, assigning each feature sample in the feature information to the cluster with the nearest centroid to form K clusters;
[0084] Step S33A, calculating the centroid of each cluster and updating the centroid position;
[0085] Step S34A, repeating steps S32A-S33A until the centroid position no longer changes or a preset number of iterations is reached.
[0086] Specifically, K-means is a commonly used unsupervised learning algorithm that assigns data features to K clusters and minimizes the square errors of data points within the clusters through iterative optimization, so that the sum of the square errors between the data points in each cluster and their centroids is minimized.
[0087] In this embodiment, trainees are divided into groups with different levels of equipment operation proficiency and diagnostic accuracy through K-means cluster analysis.
[0088] Assume that there are data related to N evaluation indicators, each evaluation indicator has M features, where N and M are both positive integers greater than 0. In the specific application process, first, randomly select K initial centroids, and the initial centroid matrix C is:
[0089] C = {c1, c2, ..., c k , ..., c K}
[0090] Wherein, the subscript k represents the centroid number, i.e., the cluster number, k = 1, 2, 3, ..., K; c K represents the Kth centroid.
[0091] Then, the distance between the feature sample and the centroid is calculated, and each feature sample is assigned to the cluster with the nearest centroid. The distance can be calculated using the Euclidean distance.
[0092]
[0093] Among them, the subscript i represents the data serial number of the evaluation index related data; the subscript j represents the feature serial number; the subscript j represents the feature serial number; x i represents the characteristics of the i-th data point, x ij represents the jth feature of the i-th data point.
[0094] Next, calculate the centroid of each cluster and update the centroid position;
[0095]
[0096] Among them, C k represents the data set of the kth cluster, |C k | represents the number of data in the kth cluster.
[0097] Finally, repeat the above steps of distance calculation, sample allocation, and centroid update until the centroid position no longer changes or the maximum number of iterations is reached and the clustering result is output.
[0098] Each cluster represents a level of equipment operation proficiency or diagnostic accuracy. For example, for equipment operation proficiency, three clusters can be formed, representing low, medium, and high levels of equipment operation proficiency. For diagnostic accuracy, it can also be divided into three levels: low, medium, and high. Specific applications can also be subdivided according to actual needs.
[0099] As a preferred embodiment, wherein Figure 3 As shown, in step S3, the clustering process is a hierarchical clustering process, including:
[0100] Step S31B, treating each feature sample in the feature information corresponding to all evaluation indicators as an independent cluster;
[0101] Step S32B, calculating the distances between all current clusters and constructing a distance matrix;
[0102] Step S33B, merging the two closest clusters into a new cluster;
[0103] Step S34B, recalculate the distance between the new cluster and other clusters, and update the distance matrix;
[0104] Step S35B, repeat steps S32B-S34B until all feature samples are merged into one cluster.
[0105] Specifically, hierarchical clustering is a clustering method based on a tree structure, which represents the clustering relationship of data points by constructing a hierarchical tree (dendrogram). Hierarchical clustering is divided into two methods, namely bottom-up (agglomerative) and top-down (divisive).
[0106] In this embodiment, the trainees are divided into groups with different levels of equipment operation proficiency and diagnostic accuracy through hierarchical cluster analysis, and each cluster represents a level of equipment operation proficiency or diagnostic accuracy.
[0107] Assume that there are data related to N evaluation indicators, each evaluation indicator has M features, where N and M are both positive integers greater than 0. In the specific application process, first, initialize: treat each data point as an independent cluster, and the initial cluster set C' is:
[0108] C'={{X1},{X2},...,{X N}}
[0109] Wherein, the subscript N represents the data sequence number; N represents the Nth cluster.
[0110] Then, the distance between the two clusters is calculated and the distance matrix is constructed; the two clusters with the closest distance are merged into a new cluster. The distance can be measured using Euclidean distance or other distance metrics.
[0111] Next, recalculate the distance between the new cluster and other clusters and update the distance matrix; the distance can be measured by minimum distance (single link), maximum distance (complete link), average distance, etc.
[0112] Then, the above steps of calculating the inter-cluster distance and merging clusters are repeated until all data points are merged into one cluster.
[0113] Furthermore, when batch evaluation is performed on the levels of multiple trainees, data of different trainees may be collected and processed using a clustering algorithm, and the clustering results may include the number of trainees corresponding to each cluster ID, as well as the equipment operation proficiency level and diagnostic accuracy level corresponding to the cluster ID.
[0114] Through the cluster analysis results, we can identify the group of trainees with high equipment operation proficiency but medium diagnostic accuracy, and the group of trainees with low equipment operation proficiency but high diagnostic accuracy, and provide personalized improvement suggestions for different groups.
[0115] In regression analysis, mathematical models are built to quantify the relationship between operational characteristics and diagnostic characteristics and student ratings. Regression analysis includes linear regression and multiple regression.
[0116] As a preferred embodiment, in step S4, a linear regression model is used for regression processing;
[0117] Among them, the linear regression model uses the feature information corresponding to each evaluation indicator in the training data set as the model input, and the scoring result corresponding to the evaluation indicator as the model output. After training, the regression coefficient and bias term are determined, the mean square error is used as the loss function, and the gradient descent method is used to update the regression coefficient and bias term; finally, the test data set is used to test the model performance.
[0118] Specifically, linear regression is a supervised learning algorithm. In this embodiment, the linear relationship between the input features and the output variables is found by minimizing the sum of squared errors.
[0119] The linear regression model is:
[0120]
[0121] in, is the predicted value output by the model, i.e., the scoring result; x is the model input, i.e., the above-mentioned feature; w represents the regression coefficient; b represents the bias term; and the superscript T represents the transposition operation.
[0122] Assume that there are data related to N evaluation indicators, and each evaluation indicator has M features, where N and M are both positive integers greater than 0. In the specific application process, the features corresponding to the operation data and the diagnosis data are used as the input variables of the linear regression model, and the scoring results are used as the output variables of the model. Through model training, the regression coefficient w and the bias term b are determined so that the predicted value The error between the actual value y is the smallest.
[0123] The loss function is the mean square error (MSE):
[0124]
[0125] Where, L(w,b) represents the mean square error function; the subscript i represents the data sequence number;
[0126] The loss function is optimized by the gradient descent method to minimize the loss function and update the regression coefficient w and the bias term b:
[0127]
[0128] Among them, η represents the learning rate.
[0129] Then the model is evaluated. Specifically, the prediction performance of the model is evaluated using the test data set. Common evaluation indicators include mean square error (MSE) and determination coefficient (R). 2 :
[0130]
[0131] in, Represents the mean of the actual values y.
[0132] In the specific application process, it is assumed that the data features include trajectory curvature, path length, average speed, acceleration, lesion marking position, measured diameter, measured area, and diagnostic conclusion, that is, the linear regression model Among them, w1, w2, w3, w4, w5, w6, w7, and w8 are the regression coefficients of the corresponding feature items, and these coefficients and the bias term b are obtained through training.
[0133] As a preferred embodiment, in step S4, a multivariate regression model is used for regression processing;
[0134] Among them, the multivariate regression model uses the feature information corresponding to the evaluation indicators of all items in the training data set as the model input, and the scoring results corresponding to the evaluation indicators as the model output. After training, the regression coefficient and bias term are determined, the mean square error is used as the loss function, and the gradient descent method is used to update the regression coefficient and bias term; finally, the test data set is used to test the model performance.
[0135] Specifically, multiple regression is an extended linear regression model. In this embodiment, the multiple regression model can be used to process multiple input variables, and the regression coefficient w and the bias term b are determined through model training so that the predicted value The error between the actual value y is the smallest.
[0136] The model training process is the same as the above linear regression model and will not be repeated here.
[0137] As a preferred embodiment, wherein Figure 4 As shown, step S5 includes:
[0138] Step S51, performing weighted average processing on the clustering results and the regression results to obtain a comprehensive score of the student to be evaluated and generate a comprehensive evaluation report;
[0139] Step S52, comparing the feature information corresponding to each evaluation indicator with the corresponding preset standard feature, and generating error description information according to the comparison result;
[0140] Step S53: matching the misalignment description information with a preset suggestion template library to generate personalized feedback suggestions.
[0141] Specifically, in this embodiment, by integrating the results of cluster analysis and regression analysis, a comprehensive evaluation of the trainee's equipment operation proficiency and diagnostic accuracy is formed, and personalized feedback suggestions are generated.
[0142] The trainees were divided into groups with different levels of equipment operation proficiency and diagnostic accuracy through K-means clustering. For example, the clustering results are shown in Table 1 below:
[0143] Table 1 Clustering result example 1
[0144] Student ID Cluster ID Average operation proficiency Average diagnostic accuracy 1 1 high middle 2 2 Low high 3 1 high middle 4 2 Low high
[0145] For example, the clustering results may also be shown in Table 2 below:
[0146] Table 2 Clustering result example 2
[0147] Cluster ID Number of students Average operation proficiency Average diagnostic accuracy 1 10 high middle 2 15 Low high
[0148] The student's score is predicted by linear regression or multivariate regression model. For example, the regression results are shown in Table 3 below:
[0149] Table 3 Regression results
[0150]
[0151]
[0152] The clustering and regression results are combined to calculate the comprehensive score and generate a comprehensive evaluation report, including the trainees' operational proficiency, diagnostic accuracy, prediction score, and comprehensive score.
[0153] The comprehensive score can be calculated by weighted average or other methods. For example, if weighted average is used, the calculation formula is as follows:
[0154] S 综合 =α*S 聚类 +β*S 回归
[0155] Among them, S 综合 Indicates the comprehensive score; S 聚类 represents the clustering score; S 回归 represents the regression score; α and β represent weight coefficients, satisfying α+β=1.
[0156] Furthermore, by analyzing the trainee's operation and diagnostic data, comparing the trainee's operation characteristics and diagnostic characteristics with the standard characteristics, identifying the parts with large deviations, and thus identifying wrong actions and irregular operations. For example, for trajectory curvature, path length, average speed and acceleration, thresholds can be set, and the parts exceeding the threshold are considered errors.
[0157] Based on the identified errors, a corresponding error description is generated.
[0158] For example, the error description is as follows:
[0159] The trajectory curvature is too large, resulting in unstable operation.
[0160] The path length is too long, which affects the operation efficiency.
[0161] An average speed that is too fast may cause erroneous operation.
[0162] The acceleration changes too much, affecting the operation accuracy.
[0163] Furthermore, a suggestion template library is pre-set, a group of suggestion templates are pre-defined and stored in the suggestion template library. The suggestion template library stores suggestion templates for each feature of each evaluation indicator at different levels.
[0164] By way of example, the suggested template library is as follows:
[0165] For students whose trajectory curvature is relatively large, it is recommended to strengthen hand stability training.
[0166] For students whose path length is too long, it is recommended to optimize the operation path.
[0167] For students whose average speed is too fast, it is recommended to control the operation speed.
[0168] For students who experience large changes in acceleration, it is recommended to practice smooth operation.
[0169] Using natural language generation (NLG) technology, according to different error types and scoring results, the suggestion template is applied to generate corresponding personalized improvement suggestions, such as strengthening hand stability training, optimizing operation paths, and improving lesion marking accuracy.
[0170] For example, assume that student A's error description is as follows:
[0171] The trajectory curvature is too large
[0172] Average speed is too fast
[0173] The generated personalized recommendations are as follows:
[0174] It is recommended to strengthen hand stability training to reduce trajectory curvature deviation.
[0175] It is recommended to control the operation speed to avoid too fast average speed.
[0176] Finally, a feedback report is generated based on the student's error description and personalized improvement suggestions. The generated feedback report can be included in the above comprehensive evaluation report and fed back to the corresponding student or instructor.
[0177] The embodiment of the present invention can use Python and Pandas to process and integrate data to generate the above report.
[0178] For example, the cluster analysis results
[0179]
[0180]
[0181] For example, the regression analysis results
[0182]
[0183] For example, data integration
[0184] merged_results=pd.merge(cluster_results,regression_results,on='student ID')
[0185] #Calculate the comprehensive score
[0186] alpha = 0.5
[0187] beta = 0.5
[0188] merged_results['comprehensive score'] = alpha*merged_results['actual score'] + beta*merged_results['predicted score']
[0189] print(merged_results)
[0190] For example, using Python and the template library to generate personalized recommendations:
[0191]
[0192] Exemplary, error identification and suggestion generation
[0193]
[0194] Through the above steps, a comprehensive assessment of trainees' operational proficiency and diagnostic accuracy can be achieved, and personalized feedback suggestions can be generated.
[0195] Through cluster analysis and regression analysis, more accurate and comprehensive evaluation results can be provided to help students understand their own operational proficiency and diagnostic accuracy. At the same time, through the personalized feedback mechanism, targeted improvement suggestions can be provided to help students improve their skills. Compared with existing technologies, this system has significant advantages in data analysis accuracy and personalized feedback.
[0196] See also Figure 5 The present invention also provides a medical level evaluation system based on cluster analysis and regression analysis, which is used to implement the medical level evaluation method based on cluster analysis and regression analysis as described above, and the system also includes:
[0197] The medical level assessment method is pre-set with at least one assessment indicator related to the medical level; and also includes:
[0198] The data collection and preprocessing module 1 is used to collect data related to each evaluation indicator of the students to be evaluated, and perform preprocessing to obtain the data to be evaluated corresponding to each evaluation indicator;
[0199] The feature extraction module 2 is connected to the data collection and preprocessing module 1 and is used to extract features from the data to be evaluated corresponding to each evaluation indicator to obtain feature information corresponding to each evaluation indicator;
[0200] The cluster analysis module 3 is connected to the feature extraction module 2 and is used to perform clustering processing on the feature information corresponding to each evaluation indicator to obtain a clustering result corresponding to each evaluation indicator;
[0201] The regression analysis module 4 is connected to the feature extraction module 2 and is used to perform regression processing on the feature information corresponding to each evaluation indicator to obtain the regression result corresponding to each evaluation indicator;
[0202] The comprehensive evaluation module 5 is connected to the cluster analysis module 3 and the regression analysis module 4 respectively, and is used to perform a comprehensive evaluation based on the clustering results and regression results corresponding to all evaluation indicators, and generate personalized feedback suggestions.
[0203] The advantages or beneficial effects of adopting the above technical solution are: the present invention accurately evaluates the skill level of trainees based on cluster analysis and regression analysis algorithms, improves the objectivity and accuracy of the evaluation, realizes the instant generation of personalized feedback suggestions, and significantly improves the efficiency and quality of medical training.
[0204] The above are only preferred embodiments of the present invention, and are not intended to limit the implementation methods and protection scope of the present invention. Those skilled in the art should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the contents of this specification and illustrations should be included in the protection scope of the present invention.
Claims
1. A medical level assessment method based on cluster analysis and regression analysis, It is characterized in that At least one evaluation indicator related to medical level is preset; The medical level assessment method includes: Step S1, collecting data related to each evaluation indicator of the student to be evaluated, and preprocessing, to obtain the data to be evaluated corresponding to each evaluation indicator; Step S2, extracting features from the data to be evaluated corresponding to each of the evaluation indicators to obtain feature information corresponding to each of the evaluation indicators; Step S3, clustering the feature information corresponding to each evaluation indicator to obtain a clustering result corresponding to each evaluation indicator; Step S4, performing regression processing on the feature information corresponding to each evaluation indicator to obtain a regression result corresponding to each evaluation indicator; Step S5, performing a comprehensive evaluation based on the clustering results and regression results corresponding to the evaluation indicators of all items, and generating personalized feedback suggestions.
2. The medical level evaluation method based on cluster analysis and regression analysis according to claim 1, It is characterized in that The evaluation indicators include medical equipment operation proficiency and / or diagnostic accuracy.
3. The medical level evaluation method based on cluster analysis and regression analysis according to claim 2, It is characterized in that The evaluation index includes proficiency in operating medical equipment, and the data related to the evaluation index is operation data, and the operation data includes at least one of the trainee's hand movement data or equipment interaction records; The evaluation index includes diagnostic accuracy, and the data related to the evaluation index is diagnostic data, which includes medical image data and diagnostic records generated by the trainee during the diagnosis process.
4. The medical level evaluation method based on cluster analysis and regression analysis according to claim 2, It is characterized in that In step S2, the evaluation index includes the proficiency of medical device operation, and the characteristic information corresponding to the evaluation index is an operation feature, and the operation feature includes at least one of a hand movement feature or a device interaction feature; The evaluation index includes diagnostic accuracy, and the feature information corresponding to the evaluation index is a diagnostic feature, and the diagnostic feature includes at least one of an image feature or a diagnostic record feature.
5. The medical level evaluation method based on cluster analysis and regression analysis according to claim 1, It is characterized in that In step S3, the clustering process is a K-means clustering process, including: Step S31A, randomly selecting K initial centroids from the feature information corresponding to the evaluation indicators of all items; wherein K is the number of clusters; Step S32A, assigning each feature sample in the feature information to the cluster with the nearest centroid, to form K clusters; Step S33A, calculating the centroid of each cluster and updating the centroid position; Step S34A, repeating steps S32A-S33A until the centroid position no longer changes or a preset number of iterations is reached.
6. The medical level evaluation method based on cluster analysis and regression analysis according to claim 1, It is characterized in that In step S3, the clustering process is a hierarchical clustering process, including: Step S31B, treating each feature sample in the feature information corresponding to the evaluation indicators of all items as an independent cluster; Step S32B, calculating the distances between all current clusters and constructing a distance matrix; Step S33B, merging the two closest clusters into a new cluster; Step S34B, recalculating the distance between the new cluster and other clusters, and updating the distance matrix; Step S35B, repeating steps S32B-S34B until all the feature samples are merged into one cluster.
7. The medical level evaluation method based on cluster analysis and regression analysis according to claim 1, It is characterized in that In step S4, a linear regression model is used for regression processing; Among them, the linear regression model uses the feature information corresponding to each evaluation indicator in the training data set as the model input, and the scoring result corresponding to the evaluation indicator as the model output. The regression coefficient and bias term are determined after training, the mean square error is used as the loss function, and the gradient descent method is used to update the regression coefficient and bias term; finally, the test data set is used to test the model performance.
8. The medical level evaluation method based on cluster analysis and regression analysis according to claim 1, It is characterized in that In step S4, a multivariate regression model is used for regression processing; Among them, the multivariate regression model uses the feature information corresponding to the evaluation indicators of all items in the training data set as the model input, and the scoring results corresponding to the evaluation indicators as the model output. The regression coefficient and bias term are determined after training, the mean square error is used as the loss function, and the gradient descent method is used to update the regression coefficient and bias term; finally, the test data set is used to test the model performance.
9. The medical level evaluation method based on cluster analysis and regression analysis according to claim 1, It is characterized in that The step S5 comprises: Step S51, performing weighted average processing on the clustering result and the regression result to obtain a comprehensive score of the student to be evaluated, and generating a comprehensive evaluation report; Step S52, comparing the feature information corresponding to each of the evaluation indicators with the corresponding preset standard features, and generating error description information according to the comparison result; Step S53: matching the misalignment description information with a preset suggestion template library to generate the personalized feedback suggestion.
10. A medical level evaluation system based on cluster analysis and regression analysis. It is characterized in that For implementing the medical level assessment method based on cluster analysis and regression analysis as described in any one of claims 1 to 9, the system further comprises: The medical level assessment method is pre-set with at least one assessment index related to the medical level; and further includes: The data collection and preprocessing module is used to collect the data related to each evaluation indicator of the students to be evaluated, and perform preprocessing to obtain the data to be evaluated corresponding to each evaluation indicator; A feature extraction module, connected to the data collection and preprocessing module, is used to extract features from the data to be evaluated corresponding to each evaluation indicator, and obtain feature information corresponding to each evaluation indicator; A cluster analysis module, connected to the feature extraction module, is used to perform clustering processing on the feature information corresponding to each of the evaluation indicators to obtain a clustering result corresponding to each of the evaluation indicators; A regression analysis module, connected to the feature extraction module, for performing regression processing on the feature information corresponding to each evaluation indicator to obtain a regression result corresponding to each evaluation indicator; The comprehensive evaluation module is connected to the cluster analysis module and the regression analysis module respectively, and is used to perform a comprehensive evaluation based on the clustering results and regression results corresponding to all the evaluation indicators, and generate personalized feedback suggestions.
Citation Information
Cited By
Water supply system energy consumption evaluation monitoring system and method based on data analysis
CN119784233A
Practical training evaluation system for old-age care based on man-machine cooperation
CN121120338A
