Internet and traditional Chinese medicine nursing service talent comprehensive evaluation system based on multi-method fusion

Abstract

The invention relates to the technical field of talent comprehensive evaluation, in particular to an internet and traditional Chinese medicine nursing service talent comprehensive evaluation system based on multi-method fusion. The method comprises the steps that an evaluation index system model constructs a four-dimensional evaluation index system equation based on four dimensions of background, input, process and result; the multi-source index construction unit constructs a dynamic index tree based on a deep reinforcement learning algorithm and a graph convolutional network; the comprehensive evaluation calculation unit calculates an evaluation value weight based on the dynamic index tree, and calculates a comprehensive evaluation value in combination with a four-dimensional evaluation index system equation; and the real-time dynamic display unit dynamically displays the dynamic index tree, the weight of each evaluation value and the comprehensive evaluation value in real time. The Internet and traditional Chinese medicine nursing service talent comprehensive evaluation system based on multi-method fusion is based on multi-method fusion, automatically adjusts evaluation indexes and weights according to real-time data and policy changes, and dynamically evaluates the multi-dimensional performance of traditional Chinese medicine nursing service talents.

Description

Technical Field

[0001] The present invention relates to the technical field of comprehensive talent evaluation, and in particular to a comprehensive talent evaluation system for Internet and traditional Chinese medicine nursing services based on the integration of multiple methods. Background Art

[0002] The comprehensive evaluation system for Internet and traditional Chinese medicine nursing service talents based on the integration of multiple methods aims to improve the evaluation accuracy and dynamic adaptability. By integrating the Delphi method, grey correlation analysis, hidden Markov model and TOPSIS technology, combined with deep reinforcement learning and graph convolutional network, it controls the subjective bias and indicator imbalance in the evaluation process, and realizes the real-time, accurate and comprehensive evaluation of the comprehensive ability, service quality and adaptability of traditional Chinese medicine nursing service talents.

[0003] The existing comprehensive evaluation system for TCM nursing service talents is usually unable to effectively handle the complex relationship between multi-dimensional information. At present, my country's "Internet + TCM nursing" is still in the exploratory stage. Relevant research mainly focuses on the effects and experience after training, the construction of training programs, etc. There is still a lack of a complete "Internet + TCM nursing" talent evaluation system. In addition, the traditional evaluation method is too dependent on expert review, static models and historical data, which cannot fully cover the evaluation indicators of this type of nursing talents, resulting in the evaluation results being too rigid, the talent evaluation is not accurate, and the quality of nursing services cannot be reflected in a timely manner. It is also difficult to reflect the impact of emerging technologies and policy changes on the comprehensive ability of talents. Therefore, a comprehensive evaluation system for Internet and TCM nursing service talents based on the integration of multiple methods is designed. Summary of the invention

[0004] The purpose of the present invention is to provide an Internet and traditional Chinese medicine nursing service talent comprehensive evaluation system based on multi-method integration, so as to solve the problems raised in the above background technology that the traditional evaluation method is too dependent on expert review, static models and historical data, and cannot comprehensively cover the evaluation indicators of such nursing talents, resulting in the evaluation results being too rigid, the talent evaluation accuracy is not high, the changes in nursing service quality cannot be reflected in a timely manner, and it is difficult to reflect the impact of emerging technologies and policy changes on the comprehensive ability of talents.

[0005] To achieve the above objectives, the present invention aims to provide a comprehensive evaluation system for Internet and traditional Chinese medicine nursing service personnel based on multi-method integration, including: An evaluation index system model, wherein the evaluation index system model is based on the four dimensions of background, input, process and result, and combines the Delphi method, grey correlation analysis method, hidden Markov model and TOPSIS technology to construct a four-dimensional evaluation index system equation; It also includes a multi-source indicator construction unit, which constructs a dynamic indicator tree based on a deep reinforcement learning algorithm and a graph convolutional network; It also includes a comprehensive evaluation calculation unit, which calculates the evaluation value weights of the background, input, process and result based on the dynamic indicator tree, and calculates the comprehensive evaluation value including the background, input, process and result in combination with the four-dimensional evaluation indicator system equation; Among them, the comprehensive evaluation value of the background, input, process and result is used as the total cycle score of the comprehensive evaluation of Internet and traditional Chinese medicine nursing service talents; It also includes a real-time dynamic display unit, which is used to dynamically display the dynamic indicator tree, the weights of each evaluation value and the comprehensive evaluation value in real time.

[0006] As a further improvement of the technical solution, the evaluation index system model includes a background evaluation module, an input evaluation module, a process evaluation module and a result evaluation module; Wherein, the background evaluation module dynamically calculates the background evaluation value according to the policy indicator scores of the expert review through the Delphi method combined with the logistic decay function; The input evaluation module dynamically calculates the input evaluation value by combining the ratio of medical technology and health resources through grey relational analysis (GRA); The process evaluation module uses a hidden Markov model to dynamically calculate the process quality score; The result evaluation module uses the TOPSIS method combined with the ReLU function to sort and normalize the results and calculate the comprehensive result evaluation value.

[0007] As a further improvement of this technical solution, the background evaluation module dynamically calculates the background evaluation value based on the policy indicator scores reviewed by experts through the Delphi method combined with the logistic decay function. The specific method is as follows: ; in, For time; is the background evaluation value at time t+1; is the background evaluation value at time t; Index for experts; is the total number of experts; For the Expert review scores; For the The Delphi method weight corresponding to each expert review; The input evaluation module dynamically calculates the input evaluation value by combining the ratio of medical technology and health resources through grey correlation analysis. The specific method is as follows: ; in, for the medical index; is the total number of medical treatments; is the input evaluation value at time t+1; is the technical resource ratio of the 𝑘th medical treatment; is the health resource ratio of the 𝑘th medical treatment; For medical datasets; for the health resource dataset; This is the result of grey relational analysis.

[0008] As a further improvement of the technical solution, the process evaluation module uses a hidden Markov model to dynamically calculate the process quality score value. The specific method is as follows: ; in, Indexing the medical service phase; is the total number of medical service stages; is the process quality score at time t+1; is the process quality score at time t; Score the quality of the sth medical service stage; The hidden Markov model scores the quality of the sth medical service stage The processing result; is the duration of the sth medical service phase; is the total duration of all phases; The result evaluation module uses the TOPSIS method combined with the ReLU function to sort and normalize the results and calculate the result evaluation value. The specific method is as follows: ; in, is the result evaluation value at time t+1; is the multi-dimensional data at time t; is the result evaluation value at time t; It is the result of sorting multi-dimensional data at time t based on TOPSIS method; It is the result of the ReLU function acting on the current evaluation value minus 0.7.

[0009] As a further improvement of the technical solution, the multi-source indicator construction unit includes a state space encoding module, a dual Q network update module, a strategy action generation module and a dynamic indicator tree construction module; The state space encoding module uses the BERT model to encode the policy text to obtain encoded text information, uses a graph convolutional network to extract the relationship information of the service relationship graph, and fuses the encoded text information with the relationship information to form a state space encoding; The dual Q network update module uses the dual Q network structure in the TD3-RL algorithm based on the state space coding and the initial policy network action to calculate the Q value under the current state space coding; The strategy action generation module optimizes the strategy network action based on state space coding and Q value; The dynamic indicator tree construction module constructs a dynamic indicator tree based on the current state space encoding and the optimized strategy network action and Q value.

[0010] As a further improvement of the technical solution, the state space encoding module uses the BERT model to encode the policy text to obtain the encoded text information, uses the graph convolutional network to extract the relationship information of the service relationship graph, and fuses the encoded text information with the relationship information to form the state space encoding. The specific method is as follows: ; in, Encode the state space at time t; is the weight matrix of the BERT model; Enter data for policy texts; [CLS] is the token output of the BERT model; is the weight matrix of the graph convolutional network; The result of processing the service relationship graph by the graph convolutional network; It is the input data of the service relationship diagram; is the LeakyReLU activation function; The dual Q network update module uses the dual Q network structure in the TD3-RL algorithm based on the state space coding and the initial policy network action to calculate the Q value under the state space coding at this time. The specific method is as follows: ; in, Calculate the index for the Q network; Network parameters calculated for the i-th Q network; The target network parameters calculated for the i-th Q network; is the state space encoding and Q value of the policy network action at time t; is the reward value at time t; To encode in the next state space and network strategy The Q value under To encode in the next state space Network Policy The action produced; is the learning rate; Encode the current state space; is the policy network action at time t; The strategy action generation module optimizes the strategy network action based on state space encoding and Q value. The specific method is as follows: ; in, represents a noise generated from a normal distribution with mean 0 and standard deviation 0.1; is the noise attenuation factor; is the optimized strategy network action; The dynamic indicator tree construction module constructs a dynamic indicator tree based on the current state space encoding and the optimized policy network action and Q value. The specific method is as follows: Dynamic indicator tree: ; in, It is a dynamic indicator tree; For network strategy Optimization; is the expectation symbol; is the discount factor; is the deadline.

[0011] As a further improvement of the technical solution, the comprehensive evaluation calculation unit includes a weight distribution module and a comprehensive evaluation calculation module; The weight distribution module calculates the evaluation value weights of the background, input, process and result based on the dynamic indicator tree; The comprehensive evaluation calculation module is used to calculate the comprehensive evaluation value including background, input, process and result according to the evaluation value weights of background, input, process and result in combination with the four-dimensional evaluation index system equation.

[0012] As a further improvement of the technical solution, the weight distribution module calculates the evaluation value weights of the background, input, process and result based on the dynamic indicator tree. The specific method is as follows: Background evaluation value weight: ; Enter the rating weight: ; Process evaluation value weight: ; Result evaluation value weight: ; in, is the background evaluation value weight; Input evaluation value weight; is the process evaluation value weight; is the result evaluation value weight; is the background evaluation constant weight coefficient; is the input evaluation constant weight coefficient; is the process evaluation constant weight coefficient; is the constant weight coefficient for result evaluation; The Q network calculation value corresponding to the background evaluation; The Q network calculates the value corresponding to the input evaluation; The Q network calculation value corresponding to the process evaluation; The Q network calculation value corresponding to the result evaluation.

[0013] As a further improvement of the technical solution, the comprehensive evaluation calculation module is used to calculate the comprehensive evaluation value including background, input, process and result according to the evaluation value weights of background, input, process and result, combined with the four-dimensional evaluation index system equation. The specific method is as follows: S3.2.1. Normalize the evaluation weights of background, input, process and results: ; in, An index representing the context, input, process, and outcome dimensions; ; is the sum of the weights of the four evaluation values; Represents the normalized weight in the z dimension; S3.2.2. Use the weighted method and the output evaluation value of the four-dimensional evaluation index system equation to calculate the comprehensive evaluation value: ; in, is the comprehensive evaluation value; Normalized weight for background evaluation value; Normalize the weights for the input evaluation values; Normalize the weights for process evaluation values; Normalize the weight of the result evaluation value.

[0014] As a further improvement of the technical solution, the real-time dynamic display unit is used to dynamically display the dynamic indicator tree, the weight of each evaluation value and the comprehensive evaluation value in real time, as follows: S4.1. Display the dynamic indicator tree structure and its changes at the current moment, including state space encoding, Q value, and actions of the policy network; S4.2, real-time display of the dynamic calculation values ​​of the four dimensions of background, input, process and result and their corresponding weights; S4.3. Display the comprehensive evaluation value obtained based on the weighted results of the four dimensions calculated dynamically.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. In this comprehensive evaluation system for Internet and traditional Chinese medicine nursing service talents based on the fusion of multiple methods, the fusion of multiple methods based on Delphi method, grey correlation analysis, hidden Markov model and TOPSIS technology can comprehensively consider the multi-dimensional performance of traditional Chinese medicine nursing service talents, dynamically weigh the weights of different dimensions, respond to policy changes and service quality changes in real time, and improve the accuracy and comprehensiveness of the evaluation results.

[0016] 2. In this comprehensive evaluation system for Internet and traditional Chinese medicine nursing service talents based on the integration of multiple methods, the dynamic indicator tree driven by deep reinforcement learning and graph convolutional networks is constructed to dynamically adjust the indicator weights, optimize the structure and content of the talent evaluation system, and automatically adjust the evaluation strategy in complex environmental changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is the overall flow chart of the present invention; The meaning of each number in the figure is: 1. Evaluation index system model; 11. Background evaluation module; 12. Input evaluation module; 13. Process evaluation module; 14. Result evaluation module; 2. Multi-source index construction unit; 21. State space encoding module; 22. Double Q network update module; 23. Strategy action generation module; 24. Dynamic index tree construction module; 3. Comprehensive evaluation calculation unit; 31. Weight allocation module; 32. Comprehensive evaluation calculation module; 4. Real-time dynamic display unit. DETAILED DESCRIPTION

[0018] 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.

[0019] See also Figure 1 As shown, a comprehensive evaluation system for Internet and TCM nursing service talents based on multi-method integration is provided, including: Evaluation index system model 1, based on the four dimensions of background, input, process and result, combined with Delphi method, grey correlation analysis method, hidden Markov model and TOPSIS technology, constructs a four-dimensional evaluation index system equation; In this embodiment, the CIPP model is a decision-oriented evaluation model consisting of four evaluation stages: background evaluation, input evaluation, process evaluation, and result evaluation. It has the advantages of procedural, comprehensive, and feedback evaluation. Based on the CIPP model, this system forms an evaluation system for "Internet + TCM nursing service talents". The background evaluation, input evaluation, process evaluation, and result evaluation in the CIPP model correspond to the four dimensions of Internet + TCM nursing service talent training environment, Internet + TCM nursing service resource allocation capabilities, Internet + TCM nursing service talent training process, and Internet + TCM nursing service talent training results. This is used as the main indicator to develop corresponding sub-indicators, thereby constructing a comprehensive evaluation system for Internet and TCM nursing service talents that integrates multiple methods. Specifically: Background evaluation: mainly evaluates the training environment, focusing on external environmental factors related to the training of TCM nursing service talents, including policy background, educational environment and social needs, etc., and is to judge the external support and environmental conditions for the training of TCM nursing talents; Input evaluation: Focus on evaluating the dispatching capabilities of Internet and TCM nursing service resource allocation, with the core being to evaluate the allocation of resources by nursing talents; Process evaluation: Process evaluation is used to analyze the implementation of the Internet and TCM nursing service talent training process, evaluate the measures, learning methods, and management models taken during the learning process of TCM nursing talents, and examine whether the planned plan is successfully implemented; Result evaluation: To examine the training effect of TCM nursing service personnel, including user feedback on the ability of nursing service personnel and social feedback from patients, etc.; In this embodiment, the evaluation index system model 1 includes a background evaluation module 11, an input evaluation module 12, a process evaluation module 13 and a result evaluation module 14; Among them, the background evaluation module 11 dynamically calculates the background evaluation value based on the policy indicator scores reviewed by experts through the Delphi method combined with the logistic decay function; The input evaluation module 12 dynamically calculates the input evaluation value by combining the ratio of medical technology and health resources through grey relational analysis GRA; The process evaluation module 13 uses a hidden Markov model to dynamically calculate the process quality score value; The result evaluation module 14 uses the TOPSIS method combined with the ReLU function to sort and normalize the results and calculate the comprehensive result evaluation value.

[0020] In this embodiment, the background evaluation module 11 dynamically calculates the background evaluation value based on the policy indicator scores reviewed by experts through the Delphi method combined with the logistic decay function. The specific method is as follows: ; in, For time; is the background evaluation value at time t+1; is the background evaluation value at time t; Index for experts; is the total number of experts; For the Expert review scores; For the The Delphi method weight corresponding to each expert review; In this embodiment, Expert review scores Experts from the Internet and TCM nursing service fields reviewed and scored the background of nursing services based on policy indicators; the Delphi weight corresponding to the 𝑖th expert review The weight of each expert score obtained through the Delphi method; Formula It is used to calculate the background evaluation value based on the policy indicators, weights and time dynamics of the expert review, taking into account the expert scores, the weight of each expert and the decay effect of time to obtain the background evaluation value at the current moment; In this embodiment, the input evaluation module 12 dynamically calculates the input evaluation value by combining the ratio of medical technology and health resources through grey correlation analysis. The specific method is as follows: ; in, for the medical index; is the total number of medical treatments; is the input evaluation value at time t+1; is the technical resource ratio of the 𝑘th medical treatment; is the health resource ratio of the 𝑘th medical treatment; For medical datasets; for the health resource dataset; It is the result of grey correlation analysis, which is used to calculate the relationship between medical technology and health resources.

[0021] In this embodiment, the technical resource ratio of the 𝑘th medical Ratio of health resources to the 𝑘th medical treatment From hospitals, public health systems or medical technology databases, indicating the supply and demand ratio of specific medical technologies and resources; medical data sets and health resource datasets Respectively represent data related to medical data and health resources, which usually come from historical data, databases of existing medical facilities and health services. Grey correlation analysis is used to analyze the relationship between the two and calculate the similarity between the two. Formula It is used to evaluate the relationship between current medical technology and health resources by dynamically calculating the ratio of medical technology and health resources combined with grey correlation analysis. Grey correlation analysis dynamically calculates the input evaluation value by measuring the similarity between medical technology and health resources.

[0022] In this embodiment, the process evaluation module 13 uses a hidden Markov model to dynamically calculate the process quality score value. The specific method is as follows: ; in, Indexing the medical service phase; is the total number of medical service stages; is the process quality score at time t+1; is the process quality score at time t; Score the quality of the sth medical service stage; The hidden Markov model scores the quality of the sth medical service stage The processing result; is the duration of the sth medical service phase; is the total duration of all phases; In this embodiment, the quality score of the sth medical service stage is The hidden Markov model will analyze the transition rules of quality scores between different states based on historical data and state transition matrix to obtain the quality score of the sth medical service stage. The processing result ;formula 1. Use the hidden Markov model to model the quality score of each stage, analyze the state transition and quality change trend in the process, and dynamically calculate the quality score value of the entire process by combining time weighting; In this embodiment, the result evaluation module 14 uses the TOPSIS method combined with the ReLU function to sort and normalize the results and calculate the result evaluation value. The specific method is as follows: ; in, is the result evaluation value at time t+1; is the multi-dimensional data at time t; is the result evaluation value at time t; It is the result of sorting multi-dimensional data at time t based on TOPSIS method; It is the result of the ReLU function acting on the current evaluation value minus 0.7.

[0023] In this embodiment, TOPSIS is a method for multi-attribute decision making, which is sorted according to the distance between the positive ideal solution and the negative ideal solution of each indicator; the multi-dimensional data of time t Derived from multi-dimensional data sets, including comprehensive data on medical technology, health resources, process quality, and other dimensions; The TOPSIS method is used to sort multi-dimensional data. By calculating the distance from the positive ideal solution and the negative ideal solution, the relative advantages and disadvantages of each alternative solution are obtained. Combined with the effect of the ReLU function, nonlinear activation can be introduced into the result, and the evaluation values ​​that are too small can be filtered and the negative numbers can be set to 0, thereby avoiding the influence of outliers on the final result. The final result is standardized by a normalization constant of 1.4 to ensure that the evaluation value is within a reasonable range and avoid extreme results. It also includes a multi-source indicator construction unit 2, which constructs a dynamic indicator tree based on a deep reinforcement learning algorithm and a graph convolutional network; In this embodiment, the multi-source indicator construction unit 2 includes a state space encoding module 21, a dual Q network updating module 22, a strategy action generating module 23 and a dynamic indicator tree construction module 24; In this embodiment, the state space encoding module 21 uses the BERT model to encode the policy text to obtain encoded text information, uses a graph convolutional network to extract the relationship information of the service relationship graph, and fuses the encoded text information with the relationship information to form a state space encoding; The dual Q network update module 22 uses the dual Q network structure in the TD3-RL algorithm based on the state space coding and the initial policy network action to calculate the Q value under the current state space coding; The strategy action generation module 23 optimizes the strategy network action based on the state space encoding and Q value; The dynamic indicator tree construction module 24 constructs a dynamic indicator tree based on the current state space encoding and the optimized strategy network action and Q value.

[0024] In this embodiment, the state space encoding module 21 uses the BERT model to encode the policy text to obtain encoded text information, uses a graph convolutional network to extract the relationship information of the service relationship graph, and fuses the encoded text information with the relationship information to form a state space encoding. The specific method is as follows: ; in, Encode the state space at time t; is the weight matrix of the BERT model, used to encode the policy text; Enter data for policy texts; The [CLS] tag output of the BERT model represents the encoding result of the policy text; It is the weight matrix of the graph convolutional network, which is used to extract the relationship information of the service relationship graph; The result of processing the service relationship graph by the graph convolutional network; It is the input data of the service relationship diagram, which represents the relationship between medical services; is the LeakyReLU activation function; In this embodiment, the dual Q network update module 22 uses the dual Q network structure in the TD3-RL algorithm based on the state space coding and the initial strategy network action to calculate the Q value under the state space coding at this time. The specific method is as follows: ; in, Calculate the index for the Q network; Network parameters calculated for the i-th Q network; The target network parameters calculated for the i-th Q network; is the state space encoding and Q value of the policy network action at time t; is the reward value at time t, which indicates the immediate reward obtained after taking an action in the current state; To encode in the next state space and network strategy The Q value under To encode in the next state space Network Policy The action produced; is the learning rate; Encode the current state space; is the policy network action at time t; In this embodiment, the Q value represents the expected return that can be obtained by performing a certain action in a certain state. The Q value can be understood as the expected evaluation return or effect after taking a certain operation such as nursing measures, decisions, etc. in a specific nursing scenario. For example, in a given state such as the current work performance and service quality of the nursing staff, the Q value represents the final result or score that the comprehensive evaluation system may give after taking a certain action such as improving service quality and adjusting work rhythm. A strategic network action refers to an operation or decision that the system can choose; in the actual comprehensive evaluation of nursing service talents, an action may refer to a specific decision, such as: adjusting nursing strategies: for example, changing nursing plans or adjusting resource allocation according to different needs of patients; service improvement: for example, optimizing a certain nursing link; The state space encoding represents the environment and conditions of the system at the current moment, including information such as the current service quality, patient feedback, and the working status of nursing staff. For example, the current working status of a nursing staff, the patient's satisfaction during the service process, etc. Rewards are feedback given based on the quality of the current action. The reward value usually represents the return given by the system based on the performance of a certain action. Rewards can be based on indicators such as patient satisfaction, nursing quality score or work efficiency. In this embodiment, the strategy action generation module 23 optimizes the strategy network action based on the state space coding and the Q value. The specific method is as follows: ; in, represents a noise generated from a normal distribution with a mean of 0 and a standard deviation of 0.1, which is used to perturb the output of the policy network; is the noise attenuation factor; is the optimized strategy network action; In this embodiment, the dynamic indicator tree construction module 24 constructs a dynamic indicator tree based on the current state space encoding and the optimized policy network action and Q value. The specific method is as follows: Dynamic indicator tree: ; in, It is a dynamic indicator tree; For network strategy Optimization; is the expectation symbol; is the discount factor; is the deadline.

[0025] It also includes a comprehensive evaluation calculation unit 3, which calculates the evaluation value weights of the background, input, process and result based on the dynamic indicator tree, and calculates the comprehensive evaluation value including the background, input, process and result in combination with the four-dimensional evaluation indicator system equation; Among them, the comprehensive evaluation value of background, input, process and result is used as the total cycle score of the comprehensive evaluation of Internet and traditional Chinese medicine nursing service talents; In this embodiment, the comprehensive evaluation calculation unit 3 includes a weight allocation module 31 and a comprehensive evaluation calculation module 32; The weight distribution module 31 calculates the evaluation value weights of the background, input, process and result based on the dynamic indicator tree; The comprehensive evaluation calculation module 32 is used to calculate the comprehensive evaluation value including background, input, process and result according to the evaluation value weights of background, input, process and result in combination with the four-dimensional evaluation index system equation.

[0026] In this embodiment, the weight allocation module 31 calculates the evaluation value weights of the background, input, process and result based on the dynamic indicator tree. The specific method is as follows: Background evaluation value weight: ; Enter the rating weight: ; Process evaluation value weight: ; Result evaluation value weight: ; in, is the background evaluation value weight; Input evaluation value weight; is the process evaluation value weight; is the result evaluation value weight; is the background evaluation constant weight coefficient; is the input evaluation constant weight coefficient; is the process evaluation constant weight coefficient; is the constant weight coefficient for result evaluation; The Q network calculation value corresponding to the background evaluation; The Q network calculates the value corresponding to the input evaluation; The Q network calculation value corresponding to the process evaluation; The Q network calculation value corresponding to the result evaluation.

[0027] In this embodiment, the comprehensive evaluation calculation module 32 is used to calculate the comprehensive evaluation value including background, input, process and result according to the evaluation value weights of background, input, process and result, combined with the four-dimensional evaluation index system equation. The specific method is as follows: S3.2.1. Normalize the evaluation weights of background, input, process and results: ; in, An index representing the context, input, process, and outcome dimensions; ; is the sum of the weights of the four evaluation values; Represents the normalized weight in the z dimension; S3.2.2. Use the weighted method and the output evaluation value of the four-dimensional evaluation index system equation to calculate the comprehensive evaluation value: ; in, is the comprehensive evaluation value; Normalized weight for background evaluation value; Normalize the weights for the input evaluation values; Normalize the weights for process evaluation values; Normalize the weight of the result evaluation value.

[0028] It also includes a real-time dynamic display unit 4, which is used to dynamically display the dynamic indicator tree, the weight of each evaluation value and the comprehensive evaluation value in real time; In this embodiment, the real-time dynamic display unit 4 is used to dynamically display the dynamic indicator tree, the weight of each evaluation value and the comprehensive evaluation value in real time, as follows: S4.1. Display the dynamic indicator tree structure and its changes at the current moment, including state space encoding, Q value, and actions of the policy network; S4.2, real-time display of the dynamic calculation values ​​of the four dimensions of background, input, process and result and their corresponding weights; S4.3. Display the comprehensive evaluation value obtained based on the weighted results of the four dimensions calculated dynamically.

[0029] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and descriptions are only preferred examples of the present invention, and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.

Claims

1. A comprehensive evaluation system for Internet and TCM nursing service talents based on multi-method integration, characterized by: include: An evaluation index system model (1), wherein the evaluation index system model (1) is based on four dimensions of background, input, process and result, and combines the Delphi method, the grey correlation analysis method, the hidden Markov model and the TOPSIS technology to construct a four-dimensional evaluation index system equation; A multi-source indicator construction unit (2), wherein the multi-source indicator construction unit (2) constructs a dynamic indicator tree based on a deep reinforcement learning algorithm and a graph convolutional network; A comprehensive evaluation calculation unit (3), wherein the comprehensive evaluation calculation unit (3) calculates the evaluation value weights of the background, input, process and result based on the dynamic indicator tree, and calculates the comprehensive evaluation value including the background, input, process and result in combination with the four-dimensional evaluation indicator system equation; Among them, the comprehensive evaluation value of the background, input, process and result is used as the total cycle score of the comprehensive evaluation of Internet and traditional Chinese medicine nursing service talents; A real-time dynamic display unit (4), the real-time dynamic display unit (4) is used to dynamically display the dynamic indicator tree, the weights of each evaluation value and the comprehensive evaluation value in real time.

2. The Internet and TCM nursing service talent comprehensive evaluation system based on multi-method integration according to claim 1 is characterized by: The evaluation index system model (1) comprises a background evaluation module (11), an input evaluation module (12), a process evaluation module (13) and a result evaluation module (14); The background evaluation module (11) dynamically calculates the background evaluation value based on the policy indicator scores of the expert review through the Delphi method combined with the logistic decay function; The input evaluation module (12) dynamically calculates the input evaluation value by combining the ratio of medical technology and health resources through grey relational analysis (GRA); The process evaluation module (13) uses a hidden Markov model to dynamically calculate the process quality score value; The result evaluation module (14) uses the TOPSIS method combined with the ReLU function to sort and normalize the results and calculate the comprehensive result evaluation value.

3. The Internet and TCM nursing service talent comprehensive evaluation system based on multi-method integration according to claim 2 is characterized by: The background evaluation module (11) dynamically calculates the background evaluation value based on the policy indicator scores of the expert review through the Delphi method combined with the logistic decay function. The specific method is as follows: ； in, For time; is the background evaluation value at time t+1; is the background evaluation value at time t; Index for experts; is the total number of experts; For the Expert review scores; For the The Delphi method weight corresponding to each expert review; The input evaluation module (12) dynamically calculates the input evaluation value by combining the ratio of medical technology and health resources through grey correlation analysis. The specific method is as follows: ； in, for the medical index; is the total number of medical treatments; is the input evaluation value at time t+1; is the technical resource ratio of the 𝑘th medical treatment; is the health resource ratio of the 𝑘th medical treatment; For medical datasets; for the health resource dataset; This is the result of grey relational analysis.

4. The comprehensive evaluation system for Internet and traditional Chinese medicine nursing service talents based on multi-method integration according to claim 3 is characterized by: The process evaluation module (13) uses a hidden Markov model to dynamically calculate the process quality score value. The specific method is as follows: ； in, Indexing the medical service phase; is the total number of medical service stages; is the process quality score at time t+1; is the process quality score at time t; Score the quality of the sth medical service stage; The hidden Markov model scores the quality of the sth medical service stage The processing result; is the duration of the sth medical service phase; is the total duration of all phases; The result evaluation module (14) uses the TOPSIS method combined with the ReLU function to sort and normalize the results and calculate the result evaluation value. The specific method is as follows: ； in, is the result evaluation value at time t+1; is the multi-dimensional data at time t; is the result evaluation value at time t; It is the result of sorting multi-dimensional data at time t based on TOPSIS method; It is the result of the ReLU function acting on the current evaluation value minus 0.

7.

5. The comprehensive evaluation system for Internet and traditional Chinese medicine nursing service talents based on multi-method integration according to claim 4 is characterized by: The multi-source indicator construction unit (2) comprises a state space encoding module (21), a dual Q network updating module (22), a strategy action generating module (23) and a dynamic indicator tree construction module (24); The state space encoding module (21) uses the BERT model to encode the policy text to obtain encoded text information, uses a graph convolutional network to extract relationship information of the service relationship graph, and fuses the encoded text information with the relationship information to form a state space encoding; The dual Q network update module (22) uses the dual Q network structure in the TD3-RL algorithm based on the state space coding and the initial strategy network action to calculate the Q value under the current state space coding; The strategy action generation module (23) optimizes the strategy network action based on the state space encoding and Q value; The dynamic indicator tree construction module (24) constructs a dynamic indicator tree based on the current state space encoding and the optimized strategy network action and Q value.

6. The Internet and TCM nursing service talent comprehensive evaluation system based on multi-method integration according to claim 5 is characterized by: The state space encoding module (21) uses the BERT model to encode the policy text to obtain encoded text information, uses a graph convolutional network to extract the relationship information of the service relationship graph, and fuses the encoded text information with the relationship information to form a state space encoding. The specific method is as follows: ； in, Encode the state space at time t; is the weight matrix of the BERT model; Enter data for policy texts; [CLS] is the token output of the BERT model; is the weight matrix of the graph convolutional network; The result of processing the service relationship graph by the graph convolutional network; It is the input data of the service relationship diagram; is the LeakyReLU activation function; The dual Q network update module (22) uses the dual Q network structure in the TD3-RL algorithm based on the state space coding and the initial policy network action to calculate the Q value under the state space coding at this time. The specific method is as follows: ； in, Calculate the index for the Q network; Network parameters calculated for the i-th Q network; The target network parameters calculated for the i-th Q network; is the state space encoding and Q value of the policy network action at time t; is the reward value at time t; To encode in the next state space and network strategy The Q value under To encode in the next state space Network Policy The action produced; is the learning rate; Encode the current state space; is the policy network action at time t; The strategy action generation module (23) optimizes the strategy network action based on the state space encoding and Q value. The specific method is as follows: ； in, represents a noise generated from a normal distribution with mean 0 and standard deviation 0.1; is the noise attenuation factor; is the optimized strategy network action; The dynamic indicator tree construction module (24) constructs a dynamic indicator tree based on the current state space encoding and the optimized strategy network action and Q value. The specific method is as follows: Dynamic indicator tree: ； in, It is a dynamic indicator tree; For network strategy Optimization; is the expectation symbol; is the discount factor; is the deadline.

7. The Internet and TCM nursing service talent comprehensive evaluation system based on multi-method integration according to claim 6 is characterized by: The comprehensive evaluation calculation unit (3) comprises a weight distribution module (31) and a comprehensive evaluation calculation module (32); The weight distribution module (31) calculates the evaluation value weights of the background, input, process and result based on the dynamic indicator tree; The comprehensive evaluation calculation module (32) is used to calculate the comprehensive evaluation value including background, input, process and result according to the evaluation value weights of background, input, process and result in combination with the four-dimensional evaluation index system equation.

8. The Internet and TCM nursing service talent comprehensive evaluation system based on multi-method integration according to claim 7 is characterized by: The weight distribution module (31) calculates the evaluation value weights of the background, input, process and result based on the dynamic indicator tree. The specific method is as follows: Background evaluation value weight: ； Enter the rating weight: ； Process evaluation value weight: ； Result evaluation value weight: ； in, is the background evaluation value weight; Input evaluation value weight; is the process evaluation value weight; is the result evaluation value weight; is the background evaluation constant weight coefficient; is the input evaluation constant weight coefficient; is the process evaluation constant weight coefficient; is the constant weight coefficient for result evaluation; The Q network calculation value corresponding to the background evaluation; The Q network calculates the value corresponding to the input evaluation; The Q network calculation value corresponding to the process evaluation; The Q network calculation value corresponding to the result evaluation.

9. The comprehensive evaluation system for Internet and traditional Chinese medicine nursing service talents based on multi-method integration according to claim 8 is characterized by: The comprehensive evaluation calculation module (32) is used to calculate the comprehensive evaluation value including background, input, process and result according to the evaluation value weights of background, input, process and result, combined with the four-dimensional evaluation index system equation. The specific method is as follows: S3.2.

1. Normalize the evaluation weights of background, input, process and results: ； in, An index representing the context, input, process, and outcome dimensions; ; is the sum of the weights of the four evaluation values; Represents the normalized weight in the z dimension; S3.2.

2. Use the weighted method and the output evaluation value of the four-dimensional evaluation index system equation to calculate the comprehensive evaluation value: ； in, is the comprehensive evaluation value; Normalized weight for background evaluation value; Normalize the weights for the input evaluation values; Normalize the weights for process evaluation values; Normalize the weight of the result evaluation value.

10. The Internet and TCM nursing service talent comprehensive evaluation system based on multi-method integration according to claim 9 is characterized by: The real-time dynamic display unit (4) is used to dynamically display the dynamic indicator tree, the weights of each evaluation value and the comprehensive evaluation value in real time, as follows: S4.

1. Display the dynamic indicator tree structure and its changes at the current moment, including state space encoding, Q value, and actions of the policy network; S4.2, real-time display of the dynamic calculation values ​​of the four dimensions of background, input, process and result and their corresponding weights; S4.

3. Display the comprehensive evaluation value obtained based on the weighted results of the four dimensions calculated dynamically.

Images