Analysis and control method and system applied to post-diagnosis service

Abstract

The invention discloses an analysis and control method and system applied to post-diagnosis services, and the method comprises the steps: obtaining historical post-diagnosis service data, carrying out the feature extraction, obtaining the service patient features, post-diagnosis service item features and service scheme features of each historical post-diagnosis service instance, and constructing a post-diagnosis service knowledge graph; establishing a satisfaction degree evaluation rule and constructing a satisfaction degree evaluation model, obtaining user feedback information of each historical post-diagnosis service, and analyzing the satisfaction degree of each historical service scheme to obtain satisfaction degree evaluation information; taking the satisfaction evaluation information as an affiliated feature, updating the post-diagnosis service knowledge graph, constructing a post-diagnosis service recommendation model, and recommending a post-diagnosis service scheme based on the diagnosis and treatment information of the target user; and analyzing the user feedback emotion change of the recommended service scheme, judging the satisfaction condition of the user on the recommended service scheme, and carrying out recommendation priority weight adjustment. And the after-diagnosis service intelligence and the user satisfaction are improved, so that personalized after-diagnosis service adjustment is realized.

Description

Technical Field

[0001] The present invention relates to the field of medical technology, and in particular to an analysis and control method and system for post-diagnosis services. Background Art

[0002] With the rapid development of medical technology, the importance of post-diagnosis services has become increasingly prominent. Post-diagnosis services include follow-up visits, drug delivery, and medical record delivery. In traditional post-diagnosis service management and control, fixed service plans are usually relied on, and there is a lack of understanding of user feedback information. Timely adjustments to service plans cannot be made, and there are problems such as information lag and insufficient data, which makes the task of disease management difficult. For post-diagnosis service plans, patient feedback information is the key to understanding whether a plan is effective. The higher the satisfaction with the post-diagnosis service plan, the better the service effect of the corresponding plan. At the same time, user satisfaction is closely related to the emotional characteristics in the user feedback text. The emotional changes in the feedback text represent the user's satisfaction with the post-diagnosis service plan. Therefore, better intelligent analysis of the feedback on post-diagnosis services, and then realizing personalized post-diagnosis service adjustments and providing better post-diagnosis services is an urgent problem to be solved. Summary of the invention

[0003] The present invention overcomes the defects of the prior art and provides an AI analysis and control method and system for post-diagnosis services, an important purpose of which is to improve the intelligence of post-diagnosis services and provide personalized post-diagnosis services.

[0004] To achieve the above objectives, the present invention provides an AI analysis and control method for post-diagnosis services in a first aspect, comprising:

[0005] Acquire historical post-diagnosis service data, perform feature extraction on the historical post-diagnosis service data to obtain service patient features, post-diagnosis service item features, and service plan features of each historical post-diagnosis service instance, and construct a post-diagnosis service knowledge graph;

[0006] Establish satisfaction evaluation rules and construct a satisfaction evaluation model, obtain user feedback information of each historical post-diagnosis service, analyze the satisfaction of each historical service plan through the constructed satisfaction evaluation model, and obtain satisfaction evaluation information;

[0007] Using the satisfaction evaluation information as an auxiliary feature, updating the post-diagnosis service knowledge graph, building a post-diagnosis service recommendation model, and recommending post-diagnosis service plans based on the target user's diagnosis and treatment information;

[0008] Obtain user feedback information on recommended service plans, analyze the changes in user feedback sentiment for recommended service plans, determine user satisfaction with the recommended service plans based on the changes in user feedback sentiment, and adjust the recommendation priority weights for each service plan.

[0009] In this solution, the acquisition of historical post-diagnosis service data, feature extraction of the historical post-diagnosis service data to obtain the service patient features, post-diagnosis service item features and service solution features of each historical post-diagnosis service instance, and construction of a post-diagnosis service knowledge graph specifically includes:

[0010] Retrieve and obtain historical post-diagnosis service data based on the target diagnosis and treatment service platform, and perform data preprocessing on the historical post-diagnosis service data;

[0011] Extract the service patient characteristics, post-diagnosis service item characteristics and service plan characteristics of each historical post-diagnosis service instance based on the pre-processed historical post-diagnosis service data, and form a feature matrix of each historical service instance;

[0012] The Euclidean distance between each feature matrix is ​​calculated based on the Euclidean distance algorithm, and each historical post-diagnosis service instance is classified according to the calculated Euclidean distance to obtain classification information;

[0013] Obtain different historical post-diagnosis service category sets according to the category classification information, extract service patient characteristics of each historical post-diagnosis service category set, and perform similarity calculation to obtain a similarity value;

[0014] The calculated similarity value is compared with the preset threshold, and based on the judgment result, the similar historical post-diagnosis service category sets are merged to form a post-diagnosis service knowledge graph with service patients, post-diagnosis service items and service plans as the path.

[0015] In this solution, the user feedback information of each historical post-diagnosis service is obtained, and the satisfaction of each historical service solution is analyzed through the constructed satisfaction evaluation model to obtain satisfaction evaluation information, which specifically includes:

[0016] Obtain a post-diagnosis service knowledge graph, extract service plans for different types of patients and corresponding post-diagnosis service items based on the post-diagnosis service knowledge graph, associate the post-diagnosis service items with the corresponding service plans, and form an association path graph with the post-diagnosis service items as a priori conditions;

[0017] The association path diagram is a plurality of historical service solutions associated with the post-diagnosis service item, and user feedback information of each historical service solution is retrieved on the target diagnosis and treatment service platform based on the association path diagram;

[0018] Extract features from user feedback information of each historical service plan, extract historical user scoring features and historical user evaluation features corresponding to each historical post-diagnosis service, and use natural language technology to perform text processing to obtain user evaluation text information;

[0019] The user evaluation text information is segmented, and the segmented user evaluation text information is vectorized using word embedding technology to obtain word vector information; an attention mechanism is introduced to calculate the attention score corresponding to each word vector through the input word vector information as a semantic tendency judgment indicator to obtain semantic tendency analysis information;

[0020] Establish satisfaction evaluation rules and construct a satisfaction evaluation model, input the word vector information and the semantic tendency analysis information into the satisfaction evaluation model for analysis, output the satisfaction score of each historical service plan, and obtain satisfaction evaluation information.

[0021] In this solution, the establishment of satisfaction evaluation rules and the construction of a satisfaction evaluation model specifically include:

[0022] Preset sentiment type labels, obtain a number of positive sentiment words, neutral sentiment words and negative sentiment words as basic sentiment words based on big data retrieval, and calculate the Euclidean distance between them and the sentiment type labels;

[0023] Obtain the Euclidean distance between each sentiment word and each sentiment type label, combine the K-means clustering algorithm to perform category division, and define the sentiment type of each sentiment word according to the category division results;

[0024] A membership threshold is preset, and the Euclidean distance between each sentiment word and the sentiment type label is combined with the membership threshold to make a judgment, analyze whether the target sentiment word belongs to the corresponding sentiment type, and use the Euclidean distance between the target sentiment word and the corresponding sentiment type as the membership degree, and perform membership degree allocation and normalization;

[0025] Calculate the cosine similarity between the sentiment words based on the cosine measurement method, analyze the synonymy relationship between the sentiment words through the calculated cosine similarity, and obtain the first analysis information;

[0026] Based on the synonymy relationship, sentiment type and membership degree, the triples are formed to express the sentiment body of each sentiment word to form a sentiment word library, the satisfaction score of the sentiment type is defined and associated with the sentiment type, the satisfaction evaluation rules are established and the satisfaction evaluation model is constructed.

[0027] In this solution, the satisfaction evaluation information is used as an auxiliary feature to update the post-diagnosis service knowledge graph, build a post-diagnosis service recommendation model, and recommend post-diagnosis service solutions based on the target user's diagnosis and treatment information, specifically including:

[0028] Obtain satisfaction evaluation information and a post-diagnosis service knowledge graph, use the satisfaction evaluation information as a service solution node attached feature, assign priority weights according to corresponding satisfaction scores, and update and represent the post-diagnosis service knowledge graph;

[0029] A post-diagnosis service recommendation model is constructed based on a graph neural network, and the updated post-diagnosis service knowledge graph representation is learned to obtain a post-diagnosis service recommendation model that meets expectations;

[0030] Obtaining the diagnosis and treatment information of the target user, inputting it into the post-diagnosis service recommendation model to perform post-diagnosis service plan analysis, extracting features of the diagnosis and treatment information of the target user, and generating a target node based on the extracted diagnosis and treatment features;

[0031] Performing first-order neighborhood sampling according to the target node to obtain neighbor nodes, constructing an adjacency matrix based on the neighbor nodes and the target node, and calculating the Bhattacharyya coefficient between the adjacency matrix and the target node;

[0032] Preset a Bhattacharyya distance threshold, perform calculation based on the calculated Bhattacharyya coefficient to obtain the Bhattacharyya distance between the adjacency matrix and the target node, and make a judgment with the Bhattacharyya distance threshold;

[0033] If it is less than the Bhattacharyya distance threshold, the corresponding adjacency matrix is ​​merged and updated with the target node, and the characteristic vector of the updated target node is obtained by using the parameter matrix and activation function;

[0034] Feature matching analysis is performed based on the updated feature vector of the target node and the post-diagnosis service knowledge graph to obtain candidate service plans, and the candidate service plans are sorted based on their priority weights to generate a post-diagnosis service recommendation plan for recommendation to the target user.

[0035] In this solution, the user feedback information of the recommended service solution is obtained, the emotional changes of the user feedback of the recommended service solution are analyzed, the user satisfaction with the recommended service solution is judged based on the emotional changes of the user feedback, and the recommendation priority weight of each service solution is adjusted, specifically including:

[0036] Obtain user feedback information on the recommended service solution, input it into the constructed satisfaction evaluation model to analyze user satisfaction with the recommended service solution, and obtain satisfaction analysis information;

[0037] Extracting the user feedback emotional features of the recommended service plan according to the satisfaction analysis information, extracting the feedback text data of the corresponding emotional type, and associating the feedback text data with the user feedback emotional features to form a first data set;

[0038] Generate frequent item sets using the first data set based on the Apriori algorithm, analyze the association rules between sentiment words in the feedback text and calculate support and confidence, and calculate the transition degree between each sentiment type according to the frequency of occurrence of the sentiment type;

[0039] According to the support, confidence and transfer degree, the emotional state transfer matrix is ​​constructed by using the Markov chain, the emotional state transfer probability is obtained through the emotional state transfer matrix, and the user feedback emotional trend of the recommended service plan is analyzed based on the emotional state transfer probability to obtain the user feedback emotional trend information;

[0040] It is determined whether there is any dissatisfaction with the recommended service solution according to the user feedback sentiment trend information. If there is any dissatisfaction, the priority weight of the corresponding recommended service solution is adjusted.

[0041] A second aspect of the present invention provides an AI analysis and control system for post-diagnosis services, the system comprising: a memory, a processor, the memory containing an AI analysis and control method program for post-diagnosis services, the AI ​​analysis and control method program for post-diagnosis services being executed by the processor to implement the following steps:

[0042] Acquire historical post-diagnosis service data, perform feature extraction on the historical post-diagnosis service data to obtain service patient features, post-diagnosis service item features, and service plan features of each historical post-diagnosis service instance, and construct a post-diagnosis service knowledge graph;

[0043] Establish satisfaction evaluation rules and construct a satisfaction evaluation model, obtain user feedback information of each historical post-diagnosis service, analyze the satisfaction of each historical service plan through the constructed satisfaction evaluation model, and obtain satisfaction evaluation information;

[0044] Using the satisfaction evaluation information as an auxiliary feature, updating the post-diagnosis service knowledge graph, building a post-diagnosis service recommendation model, and recommending post-diagnosis service plans based on the target user's diagnosis and treatment information;

[0045] Obtain user feedback information on recommended service plans, analyze the changes in user feedback sentiment for recommended service plans, determine user satisfaction with the recommended service plans based on the changes in user feedback sentiment, and adjust the recommendation priority weights for each service plan.

[0046] The present invention discloses an AI analysis and control method and system for post-diagnosis service, including: obtaining historical post-diagnosis service data, performing feature extraction to obtain service patient features, post-diagnosis service item features and service solution features of each historical post-diagnosis service instance, and constructing a post-diagnosis service knowledge map; establishing satisfaction evaluation rules and constructing a satisfaction evaluation model, obtaining user feedback information of each historical post-diagnosis service, analyzing the satisfaction of each historical service solution, and obtaining satisfaction evaluation information; using the satisfaction evaluation information as an auxiliary feature, updating the post-diagnosis service knowledge map, constructing a post-diagnosis service recommendation model, and recommending post-diagnosis service solutions based on the target user's diagnosis and treatment information; analyzing the emotional changes in user feedback on the recommended service solution, judging the user's satisfaction with the recommended service solution, and adjusting the recommendation priority weight. Improve the intelligence and user satisfaction of post-diagnosis service and thus realize personalized post-diagnosis service adjustment. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments or exemplary embodiments of the present invention, the drawings required for use in the embodiments or exemplary descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained according to the drawings without paying creative work.

[0048] Figure 1 A flowchart of an AI analysis and control method for post-diagnosis services provided by one embodiment of the present invention;

[0049] Figure 2 A flow chart of a post-diagnosis service management and control method provided by an embodiment of the present invention;

[0050] Figure 3 A block diagram of an AI analysis and control system for post-diagnosis services provided by an embodiment of the present invention;

[0051] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0052] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0053] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0054] Figure 1 A flowchart of an AI analysis and control method for post-diagnosis services provided by one embodiment of the present invention;

[0055] like Figure 1 As shown, the present invention provides a flow chart of an AI analysis and control method applied to post-diagnosis services, including:

[0056] S102, obtaining historical post-diagnosis service data, performing feature extraction on the historical post-diagnosis service data to obtain service patient features, post-diagnosis service item features, and service solution features of each historical post-diagnosis service instance, and constructing a post-diagnosis service knowledge graph;

[0057] S104, establishing satisfaction evaluation rules and constructing a satisfaction evaluation model, obtaining user feedback information of each historical post-diagnosis service, analyzing the satisfaction of each historical service plan through the constructed satisfaction evaluation model, and obtaining satisfaction evaluation information;

[0058] S106, using the satisfaction evaluation information as an auxiliary feature, updating the post-diagnosis service knowledge graph, constructing a post-diagnosis service recommendation model, and recommending post-diagnosis service solutions based on the target user's diagnosis and treatment information;

[0059] S108, obtaining user feedback information of the recommended service solutions, analyzing the emotional changes of user feedback on the recommended service solutions, judging the user's satisfaction with the recommended service solutions based on the emotional changes of user feedback, and adjusting the recommendation priority weight of each service solution.

[0060] It should be noted that the present invention provides an AI analysis and control method and system for post-diagnosis service, which extracts features from the historical post-diagnosis service data to obtain the service patient features, post-diagnosis service item features and service plan features of each historical post-diagnosis service instance, and forms a post-diagnosis service knowledge graph with service patients, post-diagnosis service items and service plans as paths. Then, a satisfaction evaluation rule is established and a satisfaction evaluation model is constructed, user feedback information of each historical post-diagnosis service is obtained, the satisfaction of each historical service plan is analyzed through the constructed satisfaction evaluation model, the satisfaction evaluation information is used as an auxiliary feature, the post-diagnosis service knowledge graph is updated, a post-diagnosis service recommendation model is constructed, and post-diagnosis service plans are recommended based on the target user's diagnosis and treatment information, thereby improving the recommendation quality of the post-diagnosis service plan. Then, the emotional changes in user feedback on the recommended service plan are analyzed, and the user's satisfaction with the recommended service plan is judged based on the emotional changes in user feedback, and the recommendation priority weight of each service plan is adjusted to provide personalized post-diagnosis services and realize the intelligence of post-diagnosis service plans.

[0061] Further, in a preferred embodiment of the present invention, the acquisition of historical post-diagnosis service data, feature extraction of the historical post-diagnosis service data to obtain service patient features, post-diagnosis service item features and service solution features of each historical post-diagnosis service instance, and construction of a post-diagnosis service knowledge graph specifically includes:

[0062] Retrieve and obtain historical post-diagnosis service data based on the target diagnosis and treatment service platform, and perform data preprocessing on the historical post-diagnosis service data;

[0063] Extract the service patient characteristics, post-diagnosis service item characteristics and service plan characteristics of each historical post-diagnosis service instance based on the pre-processed historical post-diagnosis service data, and form a feature matrix of each historical service instance;

[0064] The Euclidean distance between each feature matrix is ​​calculated based on the Euclidean distance algorithm, and each historical post-diagnosis service instance is classified according to the calculated Euclidean distance to obtain classification information;

[0065] Obtain different historical post-diagnosis service category sets according to the category classification information, extract service patient characteristics of each historical post-diagnosis service category set, and perform similarity calculation to obtain a similarity value;

[0066] The calculated similarity value is compared with the preset threshold, and based on the judgment result, the similar historical post-diagnosis service category sets are merged to form a post-diagnosis service knowledge graph with service patients, post-diagnosis service items and service plans as the path.

[0067] It should be noted that the target diagnosis and treatment service platform is used to retrieve and obtain historical post-diagnosis service data. Next, based on the pre-processed data, the multi-dimensional features of each historical post-diagnosis service instance are extracted, including the service patient features, post-diagnosis service item features, and service scheme features to form the feature matrix of each historical service instance, and then the historical service instance is quantitatively described. The Euclidean distance between each feature matrix is ​​calculated based on the Euclidean distance algorithm, and the historical post-diagnosis service instance is classified according to the calculated distance information, so as to obtain the category classification information. Subsequently, according to the category classification information, the service patient features in different historical post-diagnosis service category sets are extracted, and the similarity is calculated to obtain the similarity value, revealing the connection between similar service instances. Finally, the calculated similarity value is compared and judged with the preset threshold. Based on the judgment result, the similar historical post-diagnosis service category sets are merged to form a knowledge graph based on service patients, post-diagnosis service items and service schemes. Then, the information of historical post-diagnosis services can be systematized and visualized.

[0068] Furthermore, in a preferred embodiment of the present invention, the user feedback information of each historical post-diagnosis service is obtained, and the satisfaction of each historical service plan is analyzed by the constructed satisfaction evaluation model to obtain the satisfaction evaluation information, which specifically includes:

[0069] Obtain a post-diagnosis service knowledge graph, extract service plans for different types of patients and corresponding post-diagnosis service items based on the post-diagnosis service knowledge graph, associate the post-diagnosis service items with the corresponding service plans, and form an association path graph with the post-diagnosis service items as a priori conditions;

[0070] The association path diagram is a plurality of historical service solutions associated with the post-diagnosis service item, and user feedback information of each historical service solution is retrieved on the target diagnosis and treatment service platform based on the association path diagram;

[0071] Extract features from user feedback information of each historical service plan, extract historical user scoring features and historical user evaluation features corresponding to each historical post-diagnosis service, and use natural language technology to perform text processing to obtain user evaluation text information;

[0072] The user evaluation text information is segmented, and the segmented user evaluation text information is vectorized using word embedding technology to obtain word vector information; an attention mechanism is introduced to calculate the attention score corresponding to each word vector through the input word vector information as a semantic tendency judgment indicator to obtain semantic tendency analysis information;

[0073] Establish satisfaction evaluation rules and construct a satisfaction evaluation model, input the word vector information and the semantic tendency analysis information into the satisfaction evaluation model for analysis, output the satisfaction score of each historical service plan, and obtain satisfaction evaluation information.

[0074] It should be noted that the post-diagnosis service knowledge graph is used to extract service plans for different types of patients and corresponding post-diagnosis service items. The post-diagnosis service items are associated with their corresponding service plans, and an association path diagram with the post-diagnosis service items as a priori conditions is constructed. The association path diagram shows the connection between the post-diagnosis service items and a variety of historical service plans. Based on the association path diagram, the user feedback information of each historical service plan is retrieved on the target diagnosis and treatment service platform. Next, the user feedback information is feature extracted, including the scoring features and evaluation features of historical users. Further processing is performed using natural language processing technology to extract effective information. After processing the user evaluation text, the user evaluation text is segmented, and the sentence can be decomposed into meaningful words. Subsequently, the word embedding technology is applied to vectorize the text information after word segmentation to generate word vector information. After introducing the attention mechanism, the attention score corresponding to each word vector can be calculated by the input word vector, which is used as a judgment indicator of semantic tendency, effectively identifying important information and emotional tendency in the text. Finally, satisfaction evaluation rules are established and a satisfaction evaluation model is constructed. The extracted word vector information and semantic tendency analysis information are input into the model for analysis, and the satisfaction scores of each historical service plan are output to form satisfaction evaluation information.

[0075] Furthermore, in a preferred embodiment of the present invention, the establishment of satisfaction evaluation rules and the construction of a satisfaction evaluation model specifically include:

[0076] Preset sentiment type labels, obtain a number of positive sentiment words, neutral sentiment words and negative sentiment words as basic sentiment words based on big data retrieval, and calculate the Euclidean distance between them and the sentiment type labels;

[0077] Obtain the Euclidean distance between each sentiment word and each sentiment type label, combine the K-means clustering algorithm to perform category division, and define the sentiment type of each sentiment word according to the category division results;

[0078] A membership threshold is preset, and the Euclidean distance between each sentiment word and the sentiment type label is combined with the membership threshold to make a judgment, analyze whether the target sentiment word belongs to the corresponding sentiment type, and use the Euclidean distance between the target sentiment word and the corresponding sentiment type as the membership degree, and perform membership degree allocation and normalization;

[0079] Calculate the cosine similarity between the sentiment words based on the cosine measurement method, analyze the synonymy relationship between the sentiment words through the calculated cosine similarity, and obtain the first analysis information;

[0080] Based on the synonymy relationship, sentiment type and membership degree, the triples are formed to express the sentiment body of each sentiment word to form a sentiment word library, the satisfaction score of the sentiment type is defined and associated with the sentiment type, the satisfaction evaluation rules are established and the satisfaction evaluation model is constructed.

[0081] It should be noted that the establishment of satisfaction evaluation rules and the construction of satisfaction evaluation models are built in a way dominated by feedback sentiment tendency. The sentiment tendency in user feedback information represents the satisfaction tendency for the post-diagnosis service plan. Several positive sentiment words, neutral sentiment words and negative sentiment words are retrieved and screened from big data to form basic sentiment vocabulary. In order to clarify the sentiment tendency of each word, the Euclidean distance between these words and the preset positive, neutral and negative sentiment type labels is calculated. The K-means clustering algorithm is used to classify these sentiment words. This algorithm clusters them into the closest sentiment type according to the Euclidean distance between each sentiment word and the sentiment type label. Then, a membership threshold is preset to determine whether an sentiment word clearly belongs to a certain sentiment type. The membership relationship of the sentiment word is determined by comparing the Euclidean distance and membership threshold between the sentiment word and the sentiment type label. At the same time, this Euclidean distance is used as the membership degree to indicate the degree of membership of the sentiment word to the sentiment type. Subsequently, by calculating the cosine similarity between each sentiment word, it is analyzed which sentiment words are semantically similar. The higher the cosine similarity, the stronger the synonymy relationship between the two sentiment words. Based on the synonymy relationship, sentiment type and membership obtained above, a sentiment body representation is constructed for each sentiment word, which includes the sentiment type, synonymy relationship and membership information of the sentiment word. Finally, a satisfaction score is defined for each sentiment type, and this score is associated with the sentiment type, a satisfaction evaluation rule is established, and a satisfaction evaluation model is constructed. Based on the emotion-based satisfaction evaluation, the user's evaluation of the service is analyzed by corresponding to the different emotion types and emotional tendency levels in the feedback information, that is, the emotional state of the user when evaluating the plan is judged, and then a weighted calculation is performed according to the assignment of the emotional state to output the final satisfaction evaluation.

[0082] The sentiment body is a formalized structure that defines the basic concepts, classifications, and relationships between these concepts in the field of sentiment. It is usually used to organize and represent sentiment knowledge so that computers can understand and process it. The role of the sentiment body is to provide a structured framework for sentiment analysis, helping machines understand the emotional meaning in texts and the complex relationships between emotions.

[0083] Furthermore, in a preferred embodiment of the present invention, the satisfaction evaluation information is used as an auxiliary feature to update the post-diagnosis service knowledge graph, build a post-diagnosis service recommendation model, and recommend post-diagnosis service plans based on the target user's diagnosis and treatment information, specifically including:

[0084] Obtain satisfaction evaluation information and a post-diagnosis service knowledge graph, use the satisfaction evaluation information as a service solution node attached feature, assign priority weights according to corresponding satisfaction scores, and update and represent the post-diagnosis service knowledge graph;

[0085] A post-diagnosis service recommendation model is constructed based on a graph neural network, and the updated post-diagnosis service knowledge graph representation is learned to obtain a post-diagnosis service recommendation model that meets expectations;

[0086] Obtaining the diagnosis and treatment information of the target user, inputting it into the post-diagnosis service recommendation model to perform post-diagnosis service plan analysis, extracting features of the diagnosis and treatment information of the target user, and generating a target node based on the extracted diagnosis and treatment features;

[0087] Performing first-order neighborhood sampling according to the target node to obtain neighbor nodes, constructing an adjacency matrix based on the neighbor nodes and the target node, and calculating the Bhattacharyya coefficient between the adjacency matrix and the target node;

[0088] Preset a Bhattacharyya distance threshold, perform calculation based on the calculated Bhattacharyya coefficient to obtain the Bhattacharyya distance between the adjacency matrix and the target node, and make a judgment with the Bhattacharyya distance threshold;

[0089] If it is less than the Bhattacharyya distance threshold, the corresponding adjacency matrix is ​​merged and updated with the target node, and the characteristic vector of the updated target node is obtained by using the parameter matrix and activation function;

[0090] Feature matching analysis is performed based on the updated feature vector of the target node and the post-diagnosis service knowledge graph to obtain candidate service plans, and the candidate service plans are sorted based on their priority weights to generate a post-diagnosis service recommendation plan for recommendation to the target user.

[0091] It should be noted that after obtaining the satisfaction evaluation of the diagnosis and treatment service plan, it is used as an additional feature to update the original post-diagnosis service knowledge graph, and the priority weight of the corresponding plan is set based on the satisfaction score. The updated post-diagnosis service knowledge graph is represented by a graph, and a post-diagnosis service recommendation model is constructed using a graph neural network. When the target user needs post-diagnosis service, its diagnosis and treatment information is obtained and input into the post-diagnosis service recommendation model. The model will extract features from the diagnosis and treatment information of the target user and generate a target node, representing the current needs and status of the target user. Next, first-order neighborhood sampling is performed to obtain neighbor nodes directly connected to the target node. Then, an adjacency matrix is ​​constructed based on the neighbor nodes and the target node to represent the connection relationship between them. The Bhattacharyya coefficient between the adjacency matrix and the target node is calculated. A Bhattacharyya distance threshold is preset, and the calculated Bhattacharyya distance is compared with this threshold to screen out neighbor nodes that are similar to the target node. For those neighbor nodes that meet the conditions, their corresponding adjacency matrices are merged and updated with the target node, and the feature vectors of the updated target nodes are recalculated using the parameter matrix and activation function. Finally, we perform feature matching analysis based on the updated target node feature vector and the post-diagnosis service knowledge graph to obtain candidate service solutions. Then, we sort the candidate service solutions based on their priority weights, generate post-diagnosis service recommendation solutions that meet the needs of the target users, and recommend them to the target users, providing them with personalized and accurate post-diagnosis service solutions.

[0092] Further, in a preferred embodiment of the present invention, the steps of obtaining user feedback information of the recommended service solutions, analyzing the emotional changes of user feedback of the recommended service solutions, judging the user's satisfaction with the recommended service solutions based on the emotional changes of user feedback, and adjusting the recommendation priority weight of each service solution specifically include:

[0093] Obtain user feedback information on the recommended service solution, input it into the constructed satisfaction evaluation model to analyze user satisfaction with the recommended service solution, and obtain satisfaction analysis information;

[0094] Extracting the user feedback emotional features of the recommended service plan according to the satisfaction analysis information, extracting the feedback text data of the corresponding emotional type, and associating the feedback text data with the user feedback emotional features to form a first data set;

[0095] Generate frequent item sets using the first data set based on the Apriori algorithm, analyze the association rules between sentiment words in the feedback text and calculate support and confidence, and calculate the transition degree between each sentiment type according to the frequency of occurrence of the sentiment type;

[0096] According to the support, confidence and transfer degree, the emotional state transfer matrix is ​​constructed by using the Markov chain, the emotional state transfer probability is obtained through the emotional state transfer matrix, and the user feedback emotional trend of the recommended service plan is analyzed based on the emotional state transfer probability to obtain the user feedback emotional trend information;

[0097] It is determined whether there is any dissatisfaction with the recommended service solution according to the user feedback sentiment trend information. If there is any dissatisfaction, the priority weight of the corresponding recommended service solution is adjusted.

[0098] It should be noted that in order to improve user satisfaction and provide better personalized post-diagnosis services, there is a problem of information lag when simply making recommendations based on the satisfaction of historical solutions. Therefore, the user feedback information of the recommended service solutions is analyzed, and the emotional changes of users towards the recommended service solutions are analyzed from the feedback text data in the recommended service solutions. That is, at a macro level, the emotional trend of user feedback on a recommended service solution after several recommendations and adoptions is analyzed to obtain the emotional change direction of the recommended service solution, thereby predicting the emotional changes of users who will continue to recommend the corresponding service solution for use in the future. If the feedback emotional trend tends to be positive and satisfactory, it means that the solution is a good post-diagnosis service solution. Otherwise, it means that it is not suitable for continued recommendation, and its recommendation priority needs to be lowered to improve user satisfaction with the post-diagnosis service.

[0099] Figure 2 A flow chart of a post-diagnosis service management and control method provided by an embodiment of the present invention;

[0100] like Figure 2 As shown, the present invention provides a flow chart of a post-diagnosis service management and control method, including:

[0101] S202, obtaining diagnosis and treatment information of the target user, inputting it into the post-diagnosis service recommendation model to perform post-diagnosis service plan analysis, and outputting a number of candidate service plans;

[0102] S204, sorting the candidate service solutions based on their priority weights, generating a post-diagnosis service recommendation solution and recommending it to the target user;

[0103] S206, obtaining user feedback information of the recommended service solution, analyzing the satisfaction of the recommended service solution, and extracting the user feedback emotional features of the recommended service solution and the feedback text data of the corresponding emotional type;

[0104] S208, analyzing the sentiment trend of user feedback on the recommended solution using a Markov algorithm based on the feedback text data, and determining whether there is dissatisfaction with the recommended service solution;

[0105] S210: If there is an unsatisfactory situation, the priority weight of the corresponding recommended service solution will be adjusted.

[0106] In addition, in an AI analysis and control method for an application and post-diagnosis service provided by the present invention, the following steps are also included:

[0107] Based on the target diagnosis and treatment platform, user feedback evaluation data of each post-diagnosis service plan is obtained, and satisfaction evaluation is performed to obtain satisfaction evaluation results of each post-diagnosis service plan;

[0108] Extract the user feedback evaluation texts of each post-diagnosis service plan whose satisfaction evaluation results are unsatisfactory, and obtain feedback evaluation text information; perform sentiment analysis on the feedback evaluation text information, extract negative sentiment words in each feedback evaluation text, and form a negative sentiment word data set;

[0109] Based on the negative sentiment word data set, classify the negative sentiment words according to the sentiment type of each negative sentiment word through a clustering algorithm, and output a plurality of negative sentiment type sets;

[0110] Performing statistical operations on the negative emotion type sets, analyzing the emotion degree of each negative emotion type using the ratio between the negative emotion words in each negative emotion type set and the total negative emotion words, and obtaining second analysis information;

[0111] Based on the second analysis information, each negative emotion type set is sorted, and a selection threshold is preset, and a negative emotion type set whose ranking number is greater than the selection threshold is selected for post-diagnosis service operation analysis to obtain post-diagnosis service operation analysis information;

[0112] A post-diagnosis service optimization report is generated based on the post-diagnosis service operation analysis information to provide suggestions for post-diagnosis service operation optimization.

[0113] It should be noted that the analysis and control of post-diagnosis services, if only post-diagnosis service plans with high satisfaction are recommended to improve user satisfaction, will easily lead to the problem of high homogeneity of post-diagnosis services and low personalization. By analyzing the feedback texts of those who are dissatisfied with the post-diagnosis service plans, we can get suggestions and expectations for post-diagnosis services from user feedback, so as to promote better service operations and supplement the insufficient post-diagnosis services, so as to improve user satisfaction and the diversity of post-diagnosis services.

[0114] Figure 3 An AI analysis and control system 3 applied to post-diagnosis services is provided in one embodiment of the present invention. The system includes: a memory 31 and a processor 32. The memory 31 contains an AI analysis and control method program applied to post-diagnosis services. When the AI ​​analysis and control method program applied to post-diagnosis services is executed by the processor 32, the following steps are implemented:

[0115] Acquire historical post-diagnosis service data, perform feature extraction on the historical post-diagnosis service data to obtain service patient features, post-diagnosis service item features, and service plan features of each historical post-diagnosis service instance, and construct a post-diagnosis service knowledge graph;

[0116] Establish satisfaction evaluation rules and construct a satisfaction evaluation model, obtain user feedback information of each historical post-diagnosis service, analyze the satisfaction of each historical service plan through the constructed satisfaction evaluation model, and obtain satisfaction evaluation information;

[0117] Using the satisfaction evaluation information as an auxiliary feature, updating the post-diagnosis service knowledge graph, building a post-diagnosis service recommendation model, and recommending post-diagnosis service plans based on the target user's diagnosis and treatment information;

[0118] Obtain user feedback information on recommended service plans, analyze the changes in user feedback sentiment for recommended service plans, determine user satisfaction with the recommended service plans based on the changes in user feedback sentiment, and adjust the recommendation priority weights for each service plan.

[0119] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0120] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0121] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0122] Those skilled in the art can understand that: all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiments; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), disks or optical disks, and other media that can store program codes.

[0123] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0124] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. An analysis and control method for post-diagnosis services, characterized in that: include: Acquire historical post-diagnosis service data, perform feature extraction on the historical post-diagnosis service data to obtain service patient features, post-diagnosis service item features, and service plan features of each historical post-diagnosis service instance, and construct a post-diagnosis service knowledge graph; Establish satisfaction evaluation rules and construct a satisfaction evaluation model, obtain user feedback information of each historical post-diagnosis service, analyze the satisfaction of each historical service plan through the constructed satisfaction evaluation model, and obtain satisfaction evaluation information; Using the satisfaction evaluation information as an auxiliary feature, updating the post-diagnosis service knowledge graph, building a post-diagnosis service recommendation model, and recommending post-diagnosis service plans based on the target user's diagnosis and treatment information; Obtain user feedback information on recommended service plans, analyze the changes in user feedback sentiment for recommended service plans, determine user satisfaction with the recommended service plans based on the changes in user feedback sentiment, and adjust the recommendation priority weights for each service plan.

2. The analysis and control method for post-diagnosis service according to claim 1, characterized in that: The obtaining of historical post-diagnosis service data, extracting features from the historical post-diagnosis service data to obtain service patient features, post-diagnosis service item features, and service solution features of each historical post-diagnosis service instance, and constructing a post-diagnosis service knowledge graph specifically includes: Retrieve and obtain historical post-diagnosis service data based on the target diagnosis and treatment service platform, and perform data preprocessing on the historical post-diagnosis service data; Extract the service patient characteristics, post-diagnosis service item characteristics and service plan characteristics of each historical post-diagnosis service instance based on the pre-processed historical post-diagnosis service data, and form a feature matrix of each historical service instance; The Euclidean distance between each feature matrix is ​​calculated based on the Euclidean distance algorithm, and each historical post-diagnosis service instance is classified according to the calculated Euclidean distance to obtain classification information; According to the category classification information, different historical post-diagnosis service category sets are obtained, service patient characteristics of each historical post-diagnosis service category set are extracted, and similarity calculation is performed to obtain a similarity value; The calculated similarity value is compared with the preset threshold, and based on the judgment result, the similar historical post-diagnosis service category sets are merged to form a post-diagnosis service knowledge graph with service patients, post-diagnosis service items and service plans as the path.

3. The analysis and control method for post-diagnosis service according to claim 1, characterized in that: The user feedback information of each historical post-diagnosis service is obtained, and the satisfaction of each historical service plan is analyzed through the constructed satisfaction evaluation model to obtain satisfaction evaluation information, which specifically includes: Obtain a post-diagnosis service knowledge graph, extract service plans for different types of patients and corresponding post-diagnosis service items based on the post-diagnosis service knowledge graph, associate the post-diagnosis service items with the corresponding service plans, and form an association path graph with the post-diagnosis service items as a priori conditions; The association path diagram is a plurality of historical service solutions associated with the post-diagnosis service item, and user feedback information of each historical service solution is retrieved on the target diagnosis and treatment service platform based on the association path diagram; Extract features from user feedback information of each historical service plan, extract historical user scoring features and historical user evaluation features corresponding to each historical post-diagnosis service, and use natural language technology to perform text processing to obtain user evaluation text information; The user evaluation text information is segmented, and the segmented user evaluation text information is vectorized using word embedding technology to obtain word vector information; an attention mechanism is introduced to calculate the attention score corresponding to each word vector through the input word vector information as a semantic tendency judgment indicator to obtain semantic tendency analysis information; Establish satisfaction evaluation rules and construct a satisfaction evaluation model, input the word vector information and the semantic tendency analysis information into the satisfaction evaluation model for analysis, output the satisfaction score of each historical service plan, and obtain satisfaction evaluation information.

4. The analysis and control method for post-diagnosis service according to claim 1, characterized in that: The establishment of satisfaction evaluation rules and the construction of satisfaction evaluation models specifically include: Preset sentiment type labels, obtain a number of positive sentiment words, neutral sentiment words and negative sentiment words as basic sentiment words based on big data retrieval, and calculate the Euclidean distance between them and the sentiment type labels; Obtain the Euclidean distance between each sentiment word and each sentiment type label, combine the K-means clustering algorithm to perform category division, and define the sentiment type of each sentiment word according to the category division results; A membership threshold is preset, and the Euclidean distance between each sentiment word and the sentiment type label is combined with the membership threshold to make a judgment, analyze whether the target sentiment word belongs to the corresponding sentiment type, and use the Euclidean distance between the target sentiment word and the corresponding sentiment type as the membership degree, and perform membership degree allocation and normalization; Calculate the cosine similarity between the sentiment words based on the cosine measurement method, analyze the synonymy relationship between the sentiment words through the calculated cosine similarity, and obtain the first analysis information; Based on the synonymy relationship, sentiment type and membership degree, the triples are formed to express the sentiment body of each sentiment word to form a sentiment word library, the satisfaction score of the sentiment type is defined and associated with the sentiment type, the satisfaction evaluation rules are established and the satisfaction evaluation model is constructed.

5. The analysis and control method for post-diagnosis service according to claim 1, characterized in that: The method of using the satisfaction evaluation information as an auxiliary feature, updating the post-diagnosis service knowledge graph, building a post-diagnosis service recommendation model, and recommending post-diagnosis service solutions based on the diagnosis and treatment information of the target user specifically includes: Obtain satisfaction evaluation information and a post-diagnosis service knowledge graph, use the satisfaction evaluation information as a service solution node attached feature, assign priority weights according to corresponding satisfaction scores, and update and represent the post-diagnosis service knowledge graph; A post-diagnosis service recommendation model is constructed based on a graph neural network, and the updated post-diagnosis service knowledge graph representation is learned to obtain a post-diagnosis service recommendation model that meets expectations; Obtaining the diagnosis and treatment information of the target user, inputting it into the post-diagnosis service recommendation model to analyze the post-diagnosis service plan, extracting features of the diagnosis and treatment information of the target user and generating a target node based on the extracted diagnosis and treatment features; Performing first-order neighborhood sampling according to the target node to obtain neighbor nodes, constructing an adjacency matrix based on the neighbor nodes and the target node, and calculating the Bhattacharyya coefficient between the adjacency matrix and the target node; Preset a Bhattacharyya distance threshold, perform calculation based on the calculated Bhattacharyya coefficient to obtain the Bhattacharyya distance between the adjacency matrix and the target node, and make a judgment with the Bhattacharyya distance threshold; If it is less than the Bhattacharyya distance threshold, the corresponding adjacency matrix is ​​merged and updated with the target node, and the characteristic vector of the updated target node is obtained by using the parameter matrix and activation function; Feature matching analysis is performed based on the updated feature vector of the target node and the post-diagnosis service knowledge graph to obtain candidate service plans, and the candidate service plans are sorted based on their priority weights to generate a post-diagnosis service recommendation plan for recommendation to the target user.

6. The analysis and control method for post-diagnosis service according to claim 1, characterized in that: The obtaining of user feedback information of the recommended service solution, analyzing the emotional changes of user feedback of the recommended service solution, judging the user's satisfaction with the recommended service solution based on the emotional changes of user feedback, and adjusting the recommendation priority weight of each service solution specifically includes: Obtain user feedback information on the recommended service solution, input it into the constructed satisfaction evaluation model to analyze user satisfaction with the recommended service solution, and obtain satisfaction analysis information; Extracting the user feedback emotional features of the recommended service plan according to the satisfaction analysis information, extracting the feedback text data of the corresponding emotional type, and associating the feedback text data with the user feedback emotional features to form a first data set; Generate frequent item sets using the first data set based on the Apriori algorithm, analyze the association rules between sentiment words in the feedback text and calculate support and confidence, and calculate the transition degree between each sentiment type according to the frequency of occurrence of the sentiment type; According to the support, confidence and transfer degree, the emotional state transfer matrix is ​​constructed by using the Markov chain, the emotional state transfer probability is obtained through the emotional state transfer matrix, and the user feedback emotional trend of the recommended service plan is analyzed based on the emotional state transfer probability to obtain the user feedback emotional trend information; It is determined whether there is any dissatisfaction with the recommended service solution according to the user feedback sentiment trend information. If there is any dissatisfaction, the priority weight of the corresponding recommended service solution is adjusted.

7. An analysis and control system for post-diagnosis services, characterized in that: The system includes: a memory and a processor, wherein the memory contains an AI analysis and control method program applied to post-diagnosis services, and when the AI ​​analysis and control method program applied to post-diagnosis services is executed by the processor, the following steps are implemented: Acquire historical post-diagnosis service data, perform feature extraction on the historical post-diagnosis service data to obtain service patient features, post-diagnosis service item features, and service plan features of each historical post-diagnosis service instance, and construct a post-diagnosis service knowledge graph; Establish satisfaction evaluation rules and construct a satisfaction evaluation model, obtain user feedback information of each historical post-diagnosis service, analyze the satisfaction of each historical service plan through the constructed satisfaction evaluation model, and obtain satisfaction evaluation information; Using the satisfaction evaluation information as an auxiliary feature, updating the post-diagnosis service knowledge graph, building a post-diagnosis service recommendation model, and recommending post-diagnosis service plans based on the target user's diagnosis and treatment information; Obtain user feedback information on recommended service plans, analyze the changes in user feedback sentiment for recommended service plans, determine user satisfaction with the recommended service plans based on the changes in user feedback sentiment, and adjust the recommendation priority weights for each service plan.

Images