Medical quality assessment management system based on artificial intelligence

Abstract

The invention relates to the technical field of medical management, in particular to a medical quality assessment management system based on artificial intelligence. The system is based on a Donabedan structure-process-result model and comprises a data acquisition module, a preprocessing module, a feature extraction module, a medical quality evaluation module, an evaluation result display and feedback module, a management decision module and the like. The data acquisition module collects multi-source related data, after the data is preprocessed by the preprocessing module, the feature extraction module extracts features according to three dimensions of supportability guarantee, initiative behaviors and result output, the medical quality evaluation model outputs evaluation scores and details of each dimension according to the features, and the display and feedback module visually displays results and collects and feeds back. And the management decision module generates targeted suggestions accordingly. The artificial intelligence technology is utilized to overcome the defects of a traditional evaluation mode, comprehensive, objective and accurate evaluation and effective management of the medical quality are achieved, and a hospital is assisted to improve the medical service level.

Description

Technical Field

[0001] The present invention relates to the technical field of medical management, and particularly to a medical quality assessment and management system based on artificial intelligence. Background Art

[0002] In the modern medical system, as an important component, the medical technology departments cover multiple professional fields such as inspection, imaging, pathology, electrocardiogram, etc. The various examination and diagnosis results provided by them provide key basis for clinical diagnosis and treatment work and have a crucial impact on the overall medical quality.

[0003] However, there are still many defects in the current assessment methods for the medical quality of medical technology departments. Traditional assessment methods mostly rely on manual collection of partial index data, simple statistics, and subjective judgment based on expert experience, making it difficult to comprehensively and objectively consider the quality performance of medical technology departments in all aspects. For example, when manually counting data such as equipment maintenance status and personnel qualification information, not only is the workload large, the efficiency low, but also omissions are likely to occur; when evaluating the standardization of diagnostic and treatment operation processes and report quality, the judgment criteria of different experts may vary, resulting in the lack of consistency and objectivity of assessment results; at the same time, traditional methods often focus on single or a few aspects and cannot systematically analyze medical quality from multiple dimensions such as overall structure, process, and results.

[0004] The structure-process-result model proposed by Donabedian provides a scientific and comprehensive theoretical framework for medical quality assessment, which emphasizes comprehensively measuring medical quality from three dimensions: the supportive, active behaviors, and outcome outputs of medical services. However, when actually applied to medical technology departments, in the face of massive and complex medical data, it is difficult to effectively implement the assessment work based on this model by relying on manual means.

[0005] With the booming development of artificial intelligence technology in various fields, it demonstrates powerful capabilities in data processing, feature extraction, pattern recognition, and predictive analysis, bringing new opportunities for constructing an efficient, objective, and comprehensive medical quality assessment and management system for medical technology departments based on the Donabedian model. Summary of the Invention

[0006] The present invention provides a medical quality assessment and management system based on artificial intelligence, which makes full use of the advantages of artificial intelligence technology, accurately, comprehensively, and objectively assesses the medical quality of medical technology departments according to Donabedian's structure-process-result model, and promotes the continuous optimization of services in medical technology departments through effective management decision-making suggestions, improves medical quality, and further enhances the overall medical service level and management efficiency of the hospital.

[0007] A medical quality assessment and management system based on artificial intelligence includes:

[0008] Data collection module: This module is responsible for collecting multi-dimensional data related to the medical quality of medical technology departments. The data sources cover multiple aspects such as the equipment management system, medical technology information system, clinical feedback system, and personnel qualification management system. Information on equipment procurement, maintenance, calibration, etc. is obtained from the equipment management system, and these data reflect the supportive guarantee situation of the medical technology department in terms of hardware facilities; business operation records such as inspection reports and imaging diagnosis reports are collected from the medical technology information system, which are key data for analyzing the initiative behaviors of medical technology personnel; the clinical feedback system provides feedback information from clinical departments on the application of the results of medical technology departments, which is crucial for evaluating the dimensionality of outcome outputs; information such as personnel education, training, and professional titles in the personnel qualification management system further improves the data in the supportive guarantee dimension. Through standardized data interfaces and appropriate data extraction protocols, this module can comprehensively and stably collect complete data covering the three dimensions of supportive guarantee, initiative behavior, and outcome output, laying a foundation for subsequent evaluation work.

[0009] Data preprocessing module: Since the collected original data comes from different systems with uneven formats and qualities, preprocessing operations are required. For text data, this module removes noise information such as extra spaces and punctuation errors, corrects typos, and unifies the expression of medical terms with reference to the medical term standard library to ensure the standardization and consistency of the data; for numerical data, normalization processing is performed according to its physical meaning and industry standards to make it within an appropriate value range for subsequent analysis and model calculations; formatting and alignment operations are performed on time series data to facilitate the analysis of relevant laws in chronological order. Through these preprocessing means, the quality of the data is improved, providing a good data foundation for subsequent feature extraction and model analysis.

[0010] Feature extraction module:

[0011] Based on Donabedian's structure-process-outcome model, this module uses corresponding artificial intelligence technologies to extract valuable features for three different dimensions respectively.

[0012] Supporting guarantee dimension: In terms of equipment-related aspects, extract equipment performance index features. For example, by screening, sorting, and normalizing the original technical parameter data of the equipment, obtain the feature vectors of key indicators such as detection accuracy and range to characterize the basic performance level of the equipment. At the same time, use time series analysis algorithms to model and analyze the equipment maintenance records, extract features such as equipment failure interval time and maintenance cycle regularity to reflect the reliability and stability of the equipment. For personnel qualification data, use the word vector model in natural language processing technology to convert personnel qualification-related text information into vector representations, and then use clustering algorithms to extract features of personnel professional skill levels. For example, analyze the distribution characteristics of personnel at different skill levels to evaluate the overall professional quality and ability improvement of the department.

[0013] Initiative behavior dimension: From the records of the diagnosis and treatment operation process, use the sequence model of deep learning to extract the standardization features of the diagnosis and treatment operation process in chronological order, including calculating the deviation degree between the operation step sequence and the standard process, and whether the time consumption of key operation nodes is reasonable; for the report writing situation, use natural language processing technology to parse the report text and extract the integrity features of the report content to comprehensively measure the standardization of medical technicians in the diagnosis and treatment operation process and the accuracy of report writing.

[0014] Resultant output dimension: Combine the diagnostic result data of the medical technology department and the final clinical diagnosis results. First, clean and standardize the terms of the data, then calculate the diagnostic compliance rate feature, and further calculate the diagnostic compliance rate by subdividing different disease types and examination item types, etc., to more accurately reflect the diagnostic accuracy of the medical technology department in different business areas; in addition, collect patient satisfaction evaluation data on the services of the medical technology department from channels such as the hospital's patient satisfaction survey system, and use technologies such as sentiment analysis to extract patient satisfaction features, including the statistics of satisfaction scores and the proportion of emotional classification of text evaluations, etc., to comprehensively reflect the quality of the resultant output of the services of the medical technology department.

[0015] Medical quality assessment model module for the medical technology department:

[0016] This module is a model specifically designed for evaluating the medical quality of medical technology departments, constructed based on artificial intelligence algorithms such as deep learning and ensemble learning. The model internally contains a supportive guarantee evaluation unit, an active behavior evaluation unit, and a resultant output evaluation unit, which independently evaluate three dimensions respectively. Each evaluation unit analyzes and processes the input feature vectors of the corresponding dimension using specific algorithms, and outputs the preliminary evaluation results of each dimension. For example, the supportive guarantee evaluation unit, based on the feature vectors related to equipment and personnel, captures the temporal relationships and complex associations between features through an architecture that combines convolutional neural networks and bidirectional LSTM, and outputs the evaluation score of the supportive guarantee dimension; the active behavior evaluation unit takes the feature vectors related to diagnostic operations and report writing as input, and outputs the evaluation score of this dimension after corresponding training and analysis processes; the resultant output evaluation unit uses ensemble learning algorithms to output the evaluation score of the resultant output dimension according to feature vectors such as diagnostic compliance rate and patient satisfaction.

[0017] On this basis, there is also a medical quality evaluation unit, which uses a specific fusion algorithm to fuse the evaluation results of the three dimensions. During the fusion process, it is first necessary to determine the weights of each dimension in the final evaluation result. The determination of this weight can be achieved through a combination of expert scoring and analytic hierarchy process, ensuring that the importance of each dimension is reasonably reflected in the final evaluation. The medical quality evaluation unit performs operations such as weighted summation on the evaluation scores of the three dimensions according to the determined weights, and finally outputs the medical quality evaluation score of the medical technology department and the detailed evaluation of each dimension, providing comprehensive data support for subsequent result display and management decision-making.

[0018] Evaluation result display and feedback module: This module is responsible for presenting the evaluation results output by the medical quality evaluation model module to relevant personnel in an intuitive and visual form, including medical technology department personnel, hospital management personnel, and relevant clinical department personnel, etc. The display forms include intuitively presenting the overall medical quality score of the medical technology department and the proportion of scores of each dimension through a dashboard, using bar charts to compare the quality differences of different medical technology departments in the same dimension, and using line charts to show the medical quality trend of a certain medical technology department over time, etc., facilitating all parties to quickly understand and grasp the medical quality status of the medical technology department. At the same time, this module also sets up a feedback channel, which can collect feedback opinions from all parties on the evaluation results, such as the explanations of medical technicians on specific evaluation situations, the suggestions put forward by clinicians based on actual applications, and the views of management personnel, etc. These feedback information will provide important references for the further optimization of the system and the verification of the rationality of the evaluation results.

[0019] Management Decision Module: Based on the medical quality assessment results provided by the Evaluation Result Display and Feedback Module and the feedback opinions collected, this module generates targeted management decision suggestions for the hospital management. For example, in the dimension of supportive guarantee, if it is found that the interval time between equipment failures is shortened and the professional skill level of personnel is low, it will be recommended to arrange equipment repair and update plans in a timely manner and organize relevant personnel for training and further education; for problems such as low operation standardization and poor report quality in the dimension of initiative behavior, it is prompted to strengthen the supervision of internal operation processes and carry out training on report writing norms; if the diagnostic compliance rate is not high in the dimension of outcome output, it is proposed to organize case discussions and strengthen communication and cooperation with clinical departments. Through these decision suggestions, the medical technology departments are guided to take improvement measures from different aspects, continuously optimize the medical service process, improve the medical quality, and further improve the overall operation management level and medical service efficiency of the hospital.

[0020] Compared with the prior art, the advantages of the present invention are as follows:

[0021] Comprehensive improvement: Based on Donabedian's structure-process-outcome model, the present invention comprehensively considers the medical quality of medical technology departments from three dimensions: supportive guarantee, initiative behavior, and outcome output, avoiding the limitations of traditional evaluation methods that only focus on local indicators or single links, and being able to systematically and deeply explore various factors affecting medical quality, making the evaluation results more comprehensive and valuable for reference;

[0022] Enhanced objectivity and accuracy: With the powerful data processing and analysis capabilities of artificial intelligence technology, a large amount of complex medical technology department data is automatically processed, feature extracted, and model-based evaluation and analysis are carried out, reducing subjectivity and errors in the manual evaluation process, and outputting objective and accurate evaluation results based on data-driven models, more truly reflecting the actual medical quality status of medical technology departments, and providing a reliable basis for subsequent management decisions;

[0023] Improved decision-making support effectiveness: Through the management decision module, targeted management decision suggestions are generated based on detailed evaluation results, enabling the hospital management to take precise measures, effectively guiding medical technology departments to take improvement measures in aspects such as equipment management, personnel training, business process optimization, and cooperation with clinical departments, effectively improving the medical service quality, enhancing the overall operation efficiency of the hospital, and promoting the sustainable development of the hospital. Specific Implementation Manner

[0024] In one embodiment, a specific implementation manner of applying an artificial intelligence-based medical quality assessment and management system to a hospital is provided, including:

[0025] System Deployment: Deploy a medical quality assessment and management system based on artificial intelligence of the present invention in the information center of the hospital. The system adopts a distributed architecture, including a data acquisition server, a data preprocessing server, a feature extraction server, a model training and evaluation server, and a result display and feedback server, etc., to ensure that the system can efficiently process a large amount of data.

[0026] Data Acquisition: The data acquisition module is connected to the hospital's equipment management system, medical technology information system, clinical feedback system, and personnel qualification management system through standardized data interfaces and data extraction protocols. Adopt a combination of timed extraction and real-time push to collect data, specifically as follows:

[0027] Equipment Management System: Regularly extract the procurement information, maintenance records, calibration data, etc. of the equipment every day. For example, for the biochemical analyzer in the laboratory department, collect information such as the purchase time, the most recent maintenance time, maintenance content, calibration parameters, etc.

[0028] Medical Technology Information System: Real-time push business operation records such as inspection reports and imaging diagnosis reports. When the laboratory department completes a blood test report, the system immediately pushes the report content to the data acquisition module.

[0029] Clinical Feedback System: Regularly extract the feedback information of the clinical departments on the application of the results of the medical technology departments, such as once a week. The feedback information includes the accuracy evaluation of the test results, the degree of help of the imaging diagnosis to the clinical decision-making, etc.

[0030] Personnel Qualification Management System: Extract information such as the education background, training records, and professional titles of medical technology personnel every month. Ensure the timeliness and accuracy of personnel qualification data.

[0031] The collected data is stored in the database of the data acquisition server, and at the same time, a data caching mechanism is set up to ensure the stability and integrity of data acquisition.

[0032] Data Preprocessing: The data preprocessing module performs operations such as cleaning, normalization, and encoding on the collected raw data. The specific steps are as follows:

[0033] Text Data Cleaning: For text data such as inspection reports and personnel qualification descriptions, use regular expressions to remove format problems such as extra spaces and punctuation errors. At the same time, refer to the medical terminology standard library for term replacement, and unify non-standard medical terms into standard terms.

[0034] Numerical Data Normalization: For numerical data such as the performance parameters of equipment and the values of inspection indicators, perform normalization processing according to their physical meanings and industry standards. For example, for the detection accuracy index of the biochemical analyzer, use the min-max normalization method to map it to the interval [0,1].

[0035] Time series data formatting: For time series data such as equipment maintenance time and inspection report generation time, uniformly convert them into the standard timestamp format to facilitate subsequent time series analysis.

[0036] Feature extraction: The feature extraction module extracts features according to Donabedian's Structure-Process-Outcome model for three dimensions: supportive guarantee, proactive behavior, and outcome output. The specific methods are as follows:

[0037] Supportive guarantee dimension:

[0038] Feature extraction of equipment performance indicators: Obtain the original technical parameter data of the equipment from the equipment management system, such as the spatial resolution and density resolution of CT scanners in the imaging department. For different types of equipment, screen and organize the obtained original parameters according to the preset industry standards and equipment performance evaluation specifications, and retain the key performance indicator data. Then, use the data standardization method to normalize each key performance indicator data to form a normalized performance indicator feature vector.

[0039] Feature extraction of equipment mean time between failures: Extract the maintenance record data of the equipment from the equipment management system to form a set of equipment maintenance time series data. Use the seasonal autoregressive integrated moving average model to model and analyze the equipment maintenance time interval data, and extract the feature of the equipment mean time between failures.

[0040] Feature extraction of personnel professional skill levels: Obtain the qualification-related text information of medical technicians from the personnel qualification management system, including academic degree certificate information, professional qualification certificate acquisition situation, records of participating in various professional training courses, and academic achievement information such as past participation in scientific research projects and publication of professional papers. Use the Word2Vec word vector model to convert the words in the above text information into corresponding vector representations. Based on the converted vector representations, use the K-Means clustering algorithm to cluster and group medical technicians according to the similarity of the vectors, and extract the distribution characteristics of personnel at different skill levels.

[0041] Proactive behavior dimension:

[0042] Feature extraction of the standardization of operation step sequences:

[0043] Obtain the detailed process records of diagnostic and therapeutic operations from the medical technology information system, and match the actual operation records in chronological order according to the pre-set standard diagnostic and therapeutic operation process template. For each operation step, calculate the deviation degree between its actual execution order and the standard order, and conduct statistical analysis on the deviation values of all steps to calculate the average deviation value, maximum deviation value, etc. At the same time, mark the situations of missing steps or additional steps in the actual operation, count the number of missing steps and additional steps, and combine the above statistical quantities as the feature vector of the order normality of operation steps.

[0044] Extraction of the rationality feature of the time consumption of key operation nodes: Determine the key operation nodes in the diagnostic and therapeutic operation process, such as sample collection, test analysis, report review, etc. in the laboratory department. Obtain the start time and end time of each key operation node from the operation process record, and calculate its actual time consumption. Refer to historical data and industry standards to set a reasonable time consumption range for each key operation node, compare the actual time consumption with the reasonable time consumption range, count the number of key operation nodes with abnormal time consumption and the proportion of these abnormal nodes in the total number of key operation nodes, and combine them as the feature vector of the rationality of the time consumption of key operation nodes.

[0045] Extraction of the accuracy feature of report writing: Define the standard content items that the report should contain, such as patient basic information, examination item name, examination result description, diagnosis opinion, etc. Analyze the report text generated by the medical technology department, use natural language processing technology to identify the content items actually contained in the report, and calculate the proportion of the number of content items actually contained to the number of standard content items. Mark the missing important content items, count the number of reports with missing important content items and their proportion in the total number of reports.

[0046] Result output dimension:

[0047] Extraction of the diagnostic coincidence rate: Obtain the diagnostic result data of the medical technology department from the medical technology information system, and collect the clinical final diagnosis results from the clinical feedback system and the electronic medical record system at the same time. Clean the obtained diagnostic result data, remove duplicate, incorrect or incomplete records, and unify the expression of diagnostic terms. For each case, compare the diagnostic result of the medical technology department with the clinical final diagnosis result, count the number of cases with diagnostic coincidence and the total number of cases respectively, and calculate the diagnostic coincidence rate. Classify the cases according to different disease types, examination item types, etc., calculate the diagnostic coincidence rate for each classification respectively, and combine the diagnostic coincidence rate of each classification with the total diagnostic coincidence rate as the feature vector of the diagnostic coincidence rate.

[0048] Patient satisfaction feature extraction: Collect data on patients' satisfaction evaluations of the medical technology department's services from the hospital's patient satisfaction survey system or through online questionnaires. Preprocess the collected text evaluation data, removing irrelevant information, performing word segmentation and part-of-speech tagging, etc. For the scoring data, ensure it is under a unified scoring standard (such as a 1-5 scale). For the scoring data, calculate statistical measures such as the mean, median, and standard deviation of the patient satisfaction scores. For the text evaluation data, use sentiment analysis techniques to classify the text evaluations into three categories: positive, negative, and neutral, and count the number and proportion of positive evaluations, negative evaluations, and neutral evaluations. Combine the above-mentioned mean, median, and standard deviation of the patient satisfaction scores, as well as the number and proportion of positive evaluations, negative evaluations, and neutral evaluations into a patient satisfaction feature vector.

[0049] Medical quality assessment model training and evaluation:

[0050] Model construction: The medical quality assessment model module adopts an architecture that combines a multi-layer convolutional neural network and a bidirectional LSTM. The model internally contains a supportive guarantee assessment unit, an active behavior assessment unit, and a resultant output assessment unit, which independently evaluate the three dimensions respectively. Each assessment unit analyzes and processes the input feature vectors based on the corresponding dimension using specific algorithms. It also includes a medical quality assessment unit that fuses the evaluation results of the three dimensions into the final medical quality assessment result.

[0051] Model training: Use the extracted feature vectors of the three dimensions as input, randomly select 80% of the data as the training set and 20% of the data as the test set. Train the model using the backpropagation algorithm combined with the mean squared error loss function and the cross-entropy loss function, and adjust the model's parameters to make the model's prediction results as close as possible to the actual medical quality situation.

[0052] Model evaluation: Use the test set to evaluate the trained model, calculate indicators such as the accuracy rate, recall rate, and F1 value of the model, and evaluate the performance of the model. According to the evaluation results, further optimize and adjust the model.

[0053] Determination of dimension weights: Use a combination of expert scoring and analytic hierarchy process to determine the weights of the three dimensions of supportive guarantee, active behavior, and resultant output in the final evaluation result. The specific steps are as follows:

[0054] Expert scoring: Select 10 experts in related fields such as hospital management, medical technology department specialty, and medical quality assessment to form an expert team. Design a questionnaire and require the experts to score the relative importance between the three dimensions of supportive guarantee, active behavior, and resultant output using the 1-9 scale method. Collect the scoring results of the experts, calculate the average score, and obtain the average scoring matrix for pairwise comparison of the dimensions.

[0055] Hierarchical analysis: Calculate the maximum eigenvalue of the average scoring matrix, calculate the consistency index through the formula, find the average random consistency index, and calculate the consistency ratio. If the consistency ratio is less than 0.1, it is considered that the average scoring matrix has satisfactory consistency. Use the sum-product method to calculate the eigenvector of the matrix, and this eigenvector is the weight vector of the three dimensions.

[0056] Final evaluation: The fusion layer of the medical quality evaluation unit weights and sums the evaluation scores output by the supportive guarantee evaluation unit, the active behavior evaluation unit, and the outcome output evaluation unit according to the determined weights to obtain the final medical quality evaluation score and the detailed evaluation of each dimension.

[0057] Display and feedback of evaluation results:

[0058] The evaluation result display and feedback module displays the evaluation results through the web page of the hospital internal management platform. The overall medical quality score of the medical technology department and the proportion of scores of each dimension are presented in the form of a dashboard. The bar chart is used to compare the quality differences of different medical technology departments in the same dimension, and the line chart is used to present the quality trend of a certain medical technology department over time. At the same time, an interactive interface is provided for medical technology department personnel, hospital management personnel, and relevant personnel in clinical departments to input feedback opinions.

[0059] Generation and implementation of management decisions:

[0060] The management decision-making module provides management decision-making suggestions for the hospital management regarding the medical technology department based on the evaluation results and feedback opinions. For example, for the laboratory department, if the regularity of the equipment maintenance cycle in the supportive guarantee dimension is poor, it is recommended to increase the training of equipment maintenance personnel and optimize the equipment maintenance plan; if the accuracy of report writing in the active behavior dimension is low, organize training on report writing specifications; if the diagnostic coincidence rate in the outcome output dimension is not high, propose to organize case discussions and strengthen communication and cooperation with clinical departments. The hospital management formulates specific implementation plans based on these suggestions and supervises the implementation effects.

[0061] In another embodiment, the outcome output dimension further includes professional quality control index characteristics, specifically including:

[0062] Define and scope the quality control indicators based on historical data and expert opinions. Apply statistical and data mining techniques to identify key indicators that have a significant impact on quality control. Establish an effective feature extraction process, including steps such as data preprocessing, feature selection, and feature transformation. Data preprocessing is the key to ensuring data quality and involves operations such as cleaning data, filling in missing values, removing noise and outliers. Feature selection is to select the features that best represent the data quality from the preprocessed data, which is achieved through filtering, wrapper, or embedding methods. Through feature transformation, the original data is transformed into a form that is more conducive to model analysis, which may include standardization, normalization, or specific mathematical transformations. Through these steps, a set of optimized quality control indicator features are obtained, providing data support for subsequent quality control decisions. After the feature extraction process is completed, verify and evaluate the quality control indicators to ensure that the features truly reflect the data quality. Through the comparison of historical data between hospitals and within hospitals, evaluation indicators such as year-on-year, month-on-month, mean, and percentile are formed to form the quality control indicator features for each specialty.

[0063] In summary, an artificial intelligence-based medical quality assessment and management system of the present invention can effectively evaluate and manage the medical quality of hospitals, providing strong support for hospitals to improve the level of medical services.

[0064] In the description of this specification, the descriptions referring to terms such as "one embodiment", "example", "specific example", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0065] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and utilize the present invention well. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A medical quality assessment and management system based on artificial intelligence, characterized in that, Including: Data acquisition module: used to collect data from the device management system, medical technology information system, clinical feedback system, and personnel qualification management system; Data preprocessing module: used to perform preprocessing operations on the collected raw data, including removing noise information in text data, correcting typos, unifying the expression of medical terms, normalizing numerical data, and formatting and aligning time series data; Feature extraction module: used to extract features according to Donabedian's structure-process-result model for three dimensions of supportive guarantee, active behavior, and resultant output; Medical quality assessment model module: used to be constructed based on deep learning and ensemble learning artificial intelligence algorithms, taking the feature vectors of the three dimensions extracted by the feature extraction module as input, and outputting the medical quality assessment scores of the medical technology department and the detailed assessment of each dimension; Assessment result display and feedback module: used to display the assessment results output by the medical quality assessment model module in a visual form to the medical technology department personnel, hospital management personnel, and relevant clinical department personnel, and collect feedback from all parties on the assessment results; Management decision-making module: used to provide management decision-making suggestions for the hospital management regarding the medical technology department based on the assessment results and feedback; 2. The medical quality evaluation and management system based on artificial intelligence according to claim 1, wherein The feature extraction module includes a supportive guarantee feature extraction unit, an active behavior feature extraction unit, and a resultant output feature extraction unit. Among them, the supportive guarantee feature extraction unit extracts device performance index features, mean time between failures (MTBF) features, and personnel professional skill level features. The active behavior feature extraction unit extracts the feature of the standardization of the operation step sequence, the feature of the rationality of the time consumption of key operation nodes, and the feature of the integrity of the report content. The resultant output feature extraction unit extracts the diagnostic compliance rate feature and the patient satisfaction feature.

3. The medical quality assessment and management system based on artificial intelligence according to claim 2, wherein, The supportive guarantee feature extraction unit extracts device performance index features, equipment maintenance rule features, and personnel professional skill level features, specifically including: Extraction of device performance index features: Obtain the original technical parameter data of the device from the device management system, and the data contains various physical quantity indicators of the device; For different types of devices, screen and sort the obtained original parameters according to the preset industry standards and device performance evaluation specifications to obtain key performance index data; Normalize each key performance index data and combine it into a device performance index feature vector; Extraction of mean time between failures (MTBF) features: Extract the maintenance record data of the device from the device management system to form a device maintenance time series data set; Use time series analysis algorithms to perform modeling analysis on the device maintenance time interval data and extract the mean time between failures (MTBF) feature vector; Extraction of personnel professional skill level features: Obtain the qualification-related text information of medical technicians from the personnel qualification management system; Use the word vector model in natural language processing technology to convert the vocabulary in the above text information into corresponding vector representations; Based on the transformed vector representation, a clustering algorithm is used to cluster and group medical technicians according to the similarity of vectors, extract the distribution characteristics of personnel at different skill levels, divide the personnel into high skill level, medium skill level, and low skill level, count the number and proportion of personnel in each category, and combine them as the characteristic vector of the professional skill level of personnel.

4. The medical quality evaluation and management system based on artificial intelligence according to claim 2, characterized in that, The extraction unit of the initiative behavior characteristics extracts the characteristics of the standardization of the operation step sequence, the rationality of the time consumption of key operation nodes, and the integrity of the report content, specifically including: Extraction of the characteristics of the standardization of the operation step sequence: Obtain the detailed process records of the diagnostic operations from the medical technician information system, which include the execution time, operation content, and operator information of each step during the operation; Match the actual operation records in chronological order according to the pre-set standard diagnostic operation process template; For each operation step, calculate the deviation degree between its actual execution order and the standard order, conduct statistical analysis on the deviation values of all steps, calculate the average deviation value and the maximum deviation value, and at the same time count the number of missing steps and extra steps in the actual operation. Combine the above statistical quantities as the characteristic vector of the standardization of the operation step sequence; Extraction of the characteristics of the rationality of the time consumption of key operation nodes: Determine the key operation nodes in the diagnostic operation process, and set a reasonable time consumption range for each key operation node according to historical data and industry standards; Obtain the start time and end time of each key operation node from the operation process records, and calculate its actual time consumption; Compare the actual time consumption with the reasonable time consumption range. If the actual time consumption exceeds the reasonable range, mark it as abnormal; Count the number of key operation nodes with abnormal time consumption and the proportion of these abnormal nodes in the total number of key operation nodes, and combine them as the characteristic vector of the rationality of the time consumption of key operation nodes; Extraction of the characteristics of the integrity of the report content: Define the standard content items that the report should contain; Parse the generated report text, and use natural language processing technology to identify the content items actually included in the report; Calculate the ratio of the number of content items actually included to the number of standard content items as the characteristic vector of the integrity of the report content.

5. The medical quality assessment and management system based on artificial intelligence according to claim 2, characterized in that, The extraction unit of the result output characteristics extracts the diagnostic compliance rate characteristics and patient satisfaction characteristics, specifically including: Extraction of the diagnostic compliance rate characteristics: Obtain the diagnostic result data of the medical technician department from the medical technician information system. At the same time, collect the clinical final diagnostic results from the clinical feedback system and the electronic medical record system; Clean the obtained diagnostic result data and unify the expression of diagnostic terms; For each case, compare the diagnostic result of the medical technician department with the clinical final diagnostic result. If the two are consistent, it is determined that the diagnosis of this case is in line; if they are inconsistent, it is determined that the diagnosis is not in line; Count the number of cases with consistent diagnoses and the total number of cases respectively, and calculate the total diagnostic compliance rate, specifically the ratio of the number of cases with consistent diagnoses to the total number of cases; Classify the cases according to different disease types and examination item types, and calculate the diagnostic compliance rate for each classification; Combine the diagnostic compliance rate of each classification with the total diagnostic compliance rate as the characteristic vector of the diagnostic compliance rate; Extraction of patient satisfaction characteristics: Collect the satisfaction evaluation data of patients on the services of medical technology departments from the patient satisfaction survey system; For the scoring data, calculate the mean, median, and standard deviation of the patient satisfaction scores; For the text evaluation data, use sentiment analysis technology to classify the text evaluations into three categories: positive, negative, and neutral, and count the number and proportion of positive evaluations, negative evaluations, and neutral evaluations; Combine the mean, median, and standard deviation of the above patient satisfaction scores, as well as the number and proportion of positive evaluations, negative evaluations, and neutral evaluations, into a patient satisfaction feature vector.

6. The medical quality evaluation and management system based on artificial intelligence according to claim 1, wherein The medical quality assessment model module includes a supportive guarantee quality assessment unit, an initiative behavior quality assessment unit, and a resultant output quality assessment unit, which respectively conduct quality assessments on the three dimensions of supportive guarantee, initiative behavior, and resultant output. It also includes a medical quality assessment unit that fuses the assessment results of the three dimensions into the final medical quality assessment result. The specific fusion method includes: Determination of dimension weights: Determine the weights of the three dimensions of supportive guarantee, initiative behavior, and resultant output in the final assessment result; Result fusion: The medical quality assessment unit performs weighted summation on the assessment scores of the three dimensions according to the determined dimension weights to obtain the final medical quality assessment score.

7. The medical quality evaluation and management system based on artificial intelligence according to claim 6, characterized in that The determination of the weights of the three dimensions of supportive guarantee, initiative behavior, and resultant output in the final assessment result specifically includes: Expert scoring: Select an expert team: Select experts in the fields of hospital management, medical technology department specialties, and medical quality assessment to form an expert team; Collect expert scores: Collect the scores of experts on the relative importance between any two of the three dimensions of supportive guarantee, initiative behavior, and resultant output; Calculate the average score: Summarize the scores of all experts for each pair of dimension comparisons, calculate the average value, and obtain the average score matrix A for pairwise dimension comparisons; Analytic hierarchy process: Consistency test: Calculate the consistency index where n = 3 is the matrix dimension, and λ max is the maximum eigenvalue of the matrix. Then, find the average random consistency index RI and calculate the consistency ratio If CR < 0.1, the average scoring matrix A has satisfactory consistency, and the weight calculation can continue; if CR ≥ 0.1, return to the expert scoring stage, collect expert opinions again or adjust the scoring Calculate the weight vector: If the average score matrix A passes the consistency test, use the sum-product method to calculate the eigenvector of the matrix, and obtain the eigenvector W=(w1, w2, w3), which is the weight vector of the three dimensions of supportive guarantee, initiative behavior, and resultant output.

8. An artificial intelligence-based medical quality assessment and management system according to claim 1, characterized in that, Based on the assessment result of the supportive guarantee dimension, if the equipment failure interval is shortened and the professional skill level of personnel is low, the management decision module generates decision suggestions to arrange equipment maintenance and renewal plans in a timely manner and organize relevant personnel for training and further education; Based on the assessment result of the initiative behavior dimension, if the operation standard degree is low and the report quality is poor, it prompts to strengthen the supervision of internal operation processes and carry out training on report writing norms; In response to the situation of low diagnostic compliance rate in the resultant output dimension, it is proposed to organize case discussions and strengthen communication and cooperation among relevant departments.

Images