Internet-based medical health information collection method and system
By using preprocessing, fuzzy matching and correlation analysis of time series data and standard reference data in medical health information collection, an optimized medical health information collection model is built, which solves the problem of inefficient medical health information collection in the existing technology, and realizes the Highly accurate and efficient information collection, suitable for the needs of different users.
Patent Information
- Application Number
- CN202411087871.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-09
- Publication Date
- 2024-11-26
AI Technical Summary
The existing medical and health information collection methods are inefficient, which can easily lead to data distortion, and lack unified data standards and collection platforms, resulting in data silos and difficult to effectively integrate and analyze.
By obtaining time-series medical data and standard reference data on the Internet, preprocessing, fuzzy matching, impact analysis, data compensation and correlation analysis are carried out, and an optimized medical and health information collection model is built, using recurrent neural networks, convolutional neural networks, and knowledge graphs. and other algorithms extract keywords and features, perform data correlation and model optimization, and realize automatic and real-time collection of medical and health information.
It improves the accuracy and efficiency of medical and health information collection, saves resources and labor costs, realizes real-time information collection, adapts to the needs of different users, and is universal.
Smart Images

Figure CN119025752A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of collection, and in particular to a method and system for collecting medical and health information based on the Internet. Background Art
[0002] With the rapid development and popularization of Internet technology, people's lives have been closely connected with the Internet. Especially in the field of medical health, the application of Internet technology is gradually changing the traditional medical model and providing patients with more convenient and efficient services. However, how to effectively collect, organize and analyze users' medical and health information in order to provide users with personalized health management and disease prevention suggestions is still a major challenge currently faced.
[0003] Most existing medical and health information collection methods rely on users to actively input or upload data, which is not only inefficient, but also prone to data distortion due to subjective mistakes or forgetfulness of users. In addition, due to the lack of unified data standards and collection platforms, there are data islands between various medical and health information, making it difficult to conduct effective data integration and analysis. Therefore, developing an Internet-based medical and health information collection method and system that can automatically and in real time collect users' medical and health information, and conduct effective integration and analysis, has important practical significance and application value for improving the efficiency and quality of medical services and promoting personal health management. Summary of the invention
[0004] The purpose of the present invention is to provide a method for collecting medical health information based on the Internet.
[0005] To achieve the above object, the present invention is implemented according to the following technical solutions:
[0006] The present invention comprises the following steps:
[0007] Acquire time-series medical data and standard reference data from the Internet, and pre-process the time-series medical data to obtain a user medical data set; the time-series medical data includes user basic information and user medical data; the pre-processing includes merging the medical information data based on the user basic information;
[0008] Performing fuzzy matching on the user medical data set and the standard reference data to obtain comparative data, and performing impact analysis on the user medical data set to obtain hidden health data;
[0009] The contrast data is used to compensate the hidden health data to obtain the expanded medical data, and the expanded medical data is subjected to association analysis according to the knowledge graph to obtain the health data;
[0010] A medical health information collection model is constructed according to the influencing data, the medical health information collection model is optimized, the information to be collected is input into the optimized medical health information collection model, and the collection result is output.
[0011] Furthermore, the method of performing fuzzy matching on the user medical data set and the standard reference data to obtain comparative data includes:
[0012] A recurrent neural network is used to extract keywords from the user medical data set and calculate the semantic similarity between the keywords and the standard reference data:
[0013]
[0014] The semantic vector of the i-th keyword is f i , the semantic vector of the i-th standard reference data is k i , the number of keyword vectors is
[0015] According to the similarity between keywords and standard reference data, standard reference data with semantic similarity greater than 0.572 are used to form fuzzy comparison data;
[0016] Calculate the matching degree between fuzzy comparison data and keywords:
[0017]
[0018] The matching degree between fuzzy comparison data a and keyword k is ε ak , the Euclidean distance between fuzzy comparison data a and keyword k is Weight coefficient of Euclidean distance , the weight coefficient of similarity is χ o , the cosine function is cos(·), and the optimization parameters of fuzzy comparison data a and keyword k are σ fk ;
[0019] Keywords with a matching degree higher than 0.763 are replaced with fuzzy comparison data and output as comparison data.
[0020] Furthermore, the method of performing impact analysis on the user medical data set to obtain hidden health data includes:
[0021] A bidirectional encoder is used to extract the text features of the user's medical data to obtain the first feature, and a convolutional neural network algorithm is used to extract disease image features with reference to standard images;
[0022] According to the expert method, the disease image features are converted into language and text to obtain the disease diagnosis results, and the text features of the diagnosis results are extracted using a deep learning algorithm to obtain the second feature;
[0023] The first feature and the second feature are input into the dynamic adjustment modal fusion model, and the fusion result is input. The expression is:
[0024]
[0025] The fusion features are The first feature is The second feature is The exclusive OR logical operator is The fused features are projected into a low-dimensional space through a linear layer, and the expression is:
[0026]
[0027] The linear rectification function is ReLU(·), and the weight of the fusion feature is The adjustment parameters of the fusion feature are The low-dimensional features are
[0028] Low-dimensional features are input into the pathology diagnosis causal inference model to output hidden health data.
[0029] Furthermore, the method of using the comparison data to compensate for the hidden health data to obtain expanded medical data includes adding the hidden data to the corresponding comparison data to obtain the expanded medical data.
[0030] Furthermore, the method of performing association analysis on the expanded medical data according to the knowledge graph to obtain health data includes:
[0031] Select and expand medical data based on the medical and health knowledge graph to obtain related candidate groups;
[0032] Divide the associated candidate groups into key data sets and general data sets, and calculate the importance of key data:
[0033]
[0034] The fuzzy neighborhood radius of the key data h is The association decision is Q, the key data set is H, and the fuzzy neighborhood radius is The importance of key data h in key data set H under association decision Q is The bias weight is Key data The fuzzy neighborhood granularity of the key data set is Key data The fuzzy neighborhood granularity of the association decision is The amount of expanded medical data is g;
[0035] If the importance of key data is greater than 0, the key data is filtered as healthy data, and the calculation is continued until the key data set is traversed;
[0036] Compute the interaction gain for a general dataset:
[0037]
[0038] The interaction gain of general data a with respect to general data set A is The fuzzy neighborhood particle of general data d with respect to general data a is The fuzzy neighborhood particle size of general data d for association decision is The general data set at the t-1th moment is A t-1 , general data d about general data set A t-1 The fuzzy neighborhood particle is General Dataset A t-1 About the general dataset A t-1 The fuzzy neighborhood particle is
[0039]
[0040] If the interaction gain is greater than 0, the general data has an interaction, and the contrast of the general data with the interaction is calculated:
[0041]
[0042] The first general data with interaction is a ι1 , the second general data with interaction a ι2 , general data d about general data a ι1 The fuzzy neighborhood particle is General data d about general data a ι2 The fuzzy neighborhood particle is
[0043] If the contrast is greater than 0, the general data with interaction is filtered as the influencing data until all the general data with interaction are traversed.
[0044] Furthermore, the method for constructing a medical health information collection model based on the impact data includes:
[0045] The medical and health information collection model includes rolling window partitioning algorithm, anomaly detection algorithm, association rule learning algorithm, recurrent neural network algorithm, and attention mechanism;
[0046] The rolling window partitioning algorithm uses a fixed-size window to roll over the impact data, using the impact data within the window as the training set and the remaining impact data as the test set;
[0047] The anomaly detection algorithm searches for abnormal data based on the radius of the training set data from the hypersphere to obtain deviation data;
[0048] The association rule learning algorithm obtains drug interaction results based on the association of training data;
[0049] The recurrent neural network algorithm learns the patterns of drug interaction results and deviation data;
[0050] The attention mechanism uses test data to optimize the hyperparameters of the recurrent neural network algorithm.
[0051] Furthermore, the method for optimizing the medical health information collection model includes:
[0052] Calculate the convergence factor and search for migration positions:
[0053]
[0054]
[0055] The convergence factor is The maximum number of iterations is t max , the constant is β, the number of iterations is t, the random number from 0 to 1 is r, and the best search position of the zth particle is S bt , the search position of the zth particle in the tth iteration is S z (t), the zth search migration position is Y z , the control coefficient is γ, and the x-axis coordinate value of the particle is u;
[0056] If the spiral turn radius is less than the position update threshold, calculate a new search method:
[0057]
[0058] The first learning factor is c1, the second learning factor is c2, the random numbers from 0 to 1 are r1 and r2, and the historical optimal position of the individual at the tth iteration is S gs , the genetic factor is α, the new search method for the tth iteration is P(t), and the updated position is calculated:
[0059] S z (t)=(Y z ·u·b·q)+S bt
[0060]
[0061] The y coordinate value of the particle is b, the z coordinate value of the particle is q, and the Levy coefficient is The updated position is S nw,z(t), the step size scaling factor is δ, the random number is r3, the position update threshold is υ, and the gamma function is
[0062] If the fitness of the updated position is less than that before updating, continue the iteration; otherwise, keep the updated position.
[0063] In a second aspect, a medical health information collection system based on the Internet includes:
[0064] Data acquisition module: used to obtain time-series medical data and standard reference data on the Internet, and pre-process the time-series medical data to obtain a user medical data set; the time-series medical data includes user basic information and user medical data;
[0065] Matching analysis module: used for performing fuzzy matching on the user medical data set and the standard reference data to obtain comparison data, and performing impact analysis on the user medical data set to obtain hidden health data;
[0066] Compensation screening module: used to use the comparison data to compensate the hidden health data to obtain extended medical data, and to perform association analysis on the extended medical data according to the knowledge graph to obtain health data;
[0067] Modeling module: used to build a medical health information collection model based on the influencing data, optimize the medical health information collection model, input the information to be collected into the optimized medical health information collection model, and output the collection results.
[0068] The beneficial effects of the present invention are:
[0069] The present invention is a method for collecting medical and health information based on the Internet. Compared with the prior art, the present invention has the following technical effects:
[0070] The present invention can improve the accuracy of medical and health information collection on the Internet through data preprocessing, fuzzy matching, influence analysis, data compensation, association analysis, model building and model optimization steps, thereby improving the accuracy of medical and health information collection on the Internet, optimizing the medical and health information collection on the Internet, greatly saving resources and manpower costs, and improving work efficiency. It can realize the collection of medical and health information on the Internet, and improve the collection method of medical and health information on the Internet in real time. It is of great significance to the collection of medical and health information on the Internet, can adapt to different Internet-based medical and health information collection systems and the Internet medical and health information collection needs of different users, and has a certain universality. BRIEF DESCRIPTION OF THE DRAWINGS
[0071] Figure 1 The present invention is a flowchart of the steps of a method for collecting medical health information based on the Internet. DETAILED DESCRIPTION
[0072] The present invention is further described below by means of specific embodiments. The illustrative embodiments and descriptions of the present invention are used to explain the present invention but are not intended to limit the present invention.
[0073] The present invention provides a method for collecting medical health information based on the Internet, comprising the following steps:
[0074] like Figure 1 As shown, in this embodiment, the following steps are included:
[0075] Acquire time-series medical data and standard reference data from the Internet, and pre-process the time-series medical data to obtain a user medical data set; the time-series medical data includes user basic information and user medical data; the pre-processing includes merging the medical information data based on the user basic information;
[0076] In the actual evaluation, the time series medical data of a patient are: heart rate 80-110 beats / minute, blood pressure 120-150 / 80-95 mmHg, respiratory rate 18-24 times / minute, blood oxygen saturation 95%-98%, temperature 36.2-37.0℃; medication: anti-allergic medicine once a day, one tablet each time; taken before bedtime, heart disease medication adjusted according to the condition, three times a day, after meals; liver cancer treatment drugs according to chemotherapy regimen; blood tests showed abnormal liver function indicators, consistent with liver cancer performance; liver medical imaging;
[0077] Standard reference data include heart rate 60-100 beats / minute, blood pressure 90-140 / 60-90 mmHg, respiratory rate 12-20 times / minute, blood oxygen saturation 95%-100%, and temperature 36.1-37.2°C;
[0078] Performing fuzzy matching on the user medical data set and the standard reference data to obtain comparative data, and performing impact analysis on the user medical data set to obtain hidden health data;
[0079] In the actual evaluation, the comparison data are heart rate, blood pressure, and respiratory rate; the hidden health data are: imaging examinations showing space-occupying lesions on liver MRI, and pathological examination results showing liver tissue biopsy confirmed hepatocellular carcinoma;
[0080] The comparison data is used to compensate the hidden health data to obtain the expanded medical data, and the expanded medical data is subjected to association analysis according to the knowledge graph to obtain the health data;
[0081] In the actual evaluation, the expanded medical data included heart rate, blood pressure, respiratory rate, space-occupying lesions on liver MRI, and hepatocellular carcinoma confirmed by liver biopsy; the health data included heart rate, blood pressure, space-occupying lesions on liver MRI, and hepatocellular carcinoma;
[0082] Constructing a medical health information collection model according to the impact data, optimizing the medical health information collection model, inputting the information to be collected into the optimized medical health information collection model, and outputting the collection result;
[0083] In the actual evaluation, the data to be collected are heart rate 75-90 beats / minute, blood pressure 110-130 / 70-85 mmHg, respiratory rate 16-20 times / minute, blood oxygen saturation: 97%-99%, temperature 36.5-37.5℃
[0084] Medication status: cold medicine three times a day, two pills each time, taken after meals; anti-allergy medicine once a day, one pill each time, taken before bedtime; nutritional supplements twice a day, one pack each time, before breakfast and dinner; blood test: white blood cell count slightly elevated, consistent with cold symptoms; chest X-ray showed no abnormalities;
[0085] The collected results showed that the patient had seasonal urticaria plus common cold.
[0086] In this embodiment, the method of performing fuzzy matching on the user medical data set and the standard reference data to obtain comparative data includes:
[0087] A recurrent neural network is used to extract keywords from the user medical data set and calculate the semantic similarity between the keywords and the standard reference data:
[0088]
[0089] The semantic vector of the i-th keyword is f i , the semantic vector of the i-th standard reference data is k i , the number of keyword vectors is
[0090] According to the similarity between keywords and standard reference data, standard reference data with semantic similarity greater than 0.572 are used to form fuzzy comparison data;
[0091] Calculate the matching degree between fuzzy comparison data and keywords:
[0092]
[0093] The matching degree between fuzzy comparison data a and keyword k is ε ak , the Euclidean distance between fuzzy comparison data a and keyword k is Weight coefficient of Euclidean distance The weight coefficient of similarity is χ o , the cosine function is cos(·), and the optimization parameters of fuzzy comparison data a and keyword k are σ fk ;
[0094] Keywords with a matching degree higher than 0.763 are replaced with fuzzy comparison data and output as comparison data.
[0095] In this embodiment, the method of performing impact analysis on the user medical data set to obtain hidden health data includes:
[0096] A bidirectional encoder is used to extract the text features of the user's medical data to obtain the first feature, and a convolutional neural network algorithm is used to extract disease image features with reference to standard images;
[0097] According to the expert method, the disease image features are converted into language and text to obtain the disease diagnosis results, and the text features of the diagnosis results are extracted using a deep learning algorithm to obtain the second feature;
[0098] The first feature and the second feature are input into the dynamic adjustment modal fusion model, and the fusion result is input. The expression is:
[0099]
[0100] The fusion features are The first feature is The second feature is The exclusive OR logical operator is The fused features are projected into a low-dimensional space through a linear layer, and the expression is:
[0101]
[0102] The linear rectification function is ReLU(·), and the weight of the fusion feature is The adjustment parameters of the fusion feature are The low-dimensional features are
[0103] Low-dimensional features are input into the pathology diagnosis causal inference model to output hidden health data.
[0104] In this embodiment, the method for obtaining extended medical data by using the contrast data to compensate for the hidden health data includes adding the hidden data to the corresponding contrast data to obtain the extended medical data.
[0105] In this embodiment, the method of performing association analysis on the expanded medical data according to the knowledge graph to obtain health data includes:
[0106] Select and expand medical data based on the medical and health knowledge graph to obtain related candidate groups;
[0107] Divide the associated candidate groups into key data sets and general data sets, and calculate the importance of key data:
[0108]
[0109] The fuzzy neighborhood radius of the key data h is The association decision is Q, the key data set is H, and the fuzzy neighborhood radius is The importance of key data h in key data set H under association decision Q is The bias weight is Key data The fuzzy neighborhood granularity of the key data set is Key data The fuzzy neighborhood granularity of the association decision is The amount of expanded medical data is g;
[0110] If the importance of key data is greater than 0, the key data is filtered as healthy data, and the calculation is continued until the key data set is traversed;
[0111] Compute the interaction gain for a general dataset:
[0112]
[0113] The interaction gain of general data a with respect to general data set A is The fuzzy neighborhood particle of general data d with respect to general data a is The fuzzy neighborhood particle size of general data d for association decision is The general data set at the t-1th moment is A t-1 , general data d about general data set A t-1 The fuzzy neighborhood particle is General Dataset A t-1 About the general dataset A t-1 The fuzzy neighborhood particle is
[0114]
[0115] If the interaction gain is greater than 0, the general data has an interaction, and the contrast of the general data with the interaction is calculated:
[0116]
[0117] The first general data with interaction is a ι1 , the second general data with interaction a ι2 , general data d about general data a ι1 The fuzzy neighborhood particle is General data d about general data a ι2 The fuzzy neighborhood particle is
[0118] If the contrast is greater than 0, the general data with interaction is filtered as the influencing data until all the general data with interaction are traversed.
[0119] In this embodiment, the method for constructing a medical health information collection model based on the impact data includes:
[0120] The medical and health information collection model includes rolling window partitioning algorithm, anomaly detection algorithm, association rule learning algorithm, recurrent neural network algorithm, and attention mechanism;
[0121] The rolling window partitioning algorithm uses a fixed-size window to roll over the impact data, using the impact data within the window as the training set and the remaining impact data as the test set;
[0122] The anomaly detection algorithm searches for abnormal data based on the radius of the training set data from the hypersphere to obtain deviation data;
[0123] The association rule learning algorithm obtains drug interaction results based on the association of training data;
[0124] The recurrent neural network algorithm learns the patterns of drug interaction results and deviation data;
[0125] The attention mechanism uses test data to optimize the hyperparameters of the recurrent neural network algorithm.
[0126] In this embodiment, the method for optimizing the medical health information collection model includes:
[0127] Calculate the convergence factor and search for migration positions:
[0128]
[0129]
[0130] The convergence factor is The maximum number of iterations is t max , the constant is β, the number of iterations is t, the random number from 0 to 1 is r, and the best search position of the zth particle is S bt , the search position of the zth particle in the tth iteration is S z (t), the zth search migration position is Y z , the control coefficient is γ, and the x-axis coordinate value of the particle is u;
[0131] If the spiral turn radius is less than the position update threshold, calculate a new search method:
[0132]
[0133] The first learning factor is c1, the second learning factor is c2, the random numbers from 0 to 1 are r1 and r2, and the historical optimal position of the individual at the tth iteration is S gs , the genetic factor is α, the new search method for the tth iteration is P(t), and the updated position is calculated:
[0134] S z (t)=(Y z ·u·b·q)+S bt
[0135]
[0136] The y coordinate value of the particle is b, the z coordinate value of the particle is q, and the Levy coefficient is The updated position is S nw,z (t), the step size scaling factor is δ, the random number is r3, the position update threshold is υ, and the gamma function is
[0137] If the fitness of the updated position is less than that before updating, continue the iteration; otherwise, keep the updated position.
[0138] In a second aspect, a medical health information collection system based on the Internet includes:
[0139] Data acquisition module: used to obtain time-series medical data and standard reference data on the Internet, and pre-process the time-series medical data to obtain a user medical data set; the time-series medical data includes user basic information and user medical data;
[0140] Matching analysis module: used for performing fuzzy matching on the user medical data set and the standard reference data to obtain comparison data, and performing impact analysis on the user medical data set to obtain hidden health data;
[0141] Compensation screening module: used to use the comparison data to compensate the hidden health data to obtain extended medical data, and to perform association analysis on the extended medical data according to the knowledge graph to obtain health data;
[0142] Modeling module: used to build a medical health information collection model based on the influencing data, optimize the medical health information collection model, input the information to be collected into the optimized medical health information collection model, and output the collection results.
[0143] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for collecting medical and health information based on the Internet. It is characterized in that The following steps are involved: Acquire time-series medical data and standard reference data from the Internet, and pre-process the time-series medical data to obtain a user medical data set; the time-series medical data includes user basic information and user medical data; Performing fuzzy matching on the user medical data set and the standard reference data to obtain comparative data, and performing impact analysis on the user medical data set to obtain hidden health data; The contrast data is used to compensate the hidden health data to obtain the expanded medical data, and the expanded medical data is subjected to association analysis according to the knowledge graph to obtain the health data; A medical health information collection model is constructed according to the influencing data, the medical health information collection model is optimized, the information to be collected is input into the optimized medical health information collection model, and the collection result is output.
2. According to claim 1, a method for collecting medical and health information based on the Internet, It is characterized in that The method for performing fuzzy matching on the user medical data set and the standard reference data to obtain comparative data comprises: A recurrent neural network is used to extract keywords from the user medical data set and calculate the semantic similarity between the keywords and the standard reference data: The semantic vector of the i-th keyword is f i , the semantic vector of the i-th standard reference data is k i , the number of keyword vectors is According to the similarity between keywords and standard reference data, standard reference data with semantic similarity greater than 0.572 are used to form fuzzy comparison data; Calculate the matching degree between fuzzy comparison data and keywords: The matching degree between fuzzy comparison data a and keyword k is ε ak , the Euclidean distance between fuzzy comparison data a and keyword k is d fk , the weight coefficient of the Euclidean distance χ d , the weight coefficient of similarity is χ o , the cosine function is cos(·), and the optimization parameters of fuzzy comparison data a and keyword k are σ fk ; Keywords with a matching degree higher than 0.763 are replaced with fuzzy comparison data and output as comparison data.
3. According to the method for collecting medical and health information based on the Internet as claimed in claim 1, It is characterized in that The method for performing impact analysis on the user medical data set to obtain hidden health data includes: A bidirectional encoder is used to extract the text features of the user's medical data to obtain the first feature, and a convolutional neural network algorithm is used to extract disease image features with reference to standard images; According to the expert method, the disease image features are converted into language and text to obtain the disease diagnosis results, and the text features of the diagnosis results are extracted using a deep learning algorithm to obtain the second feature; The first feature and the second feature are input into the dynamic adjustment modal fusion model, and the fusion result is input. The expression is: The fusion features are The first feature is The second feature is The XOR logical operator is The fused features are projected into a low-dimensional space through a linear layer, and the expression is: The linear rectification function is ReLU(·), and the weight of the fusion feature is The adjustment parameters of the fusion feature are The low-dimensional features are Low-dimensional features are input into the pathology diagnosis causal inference model to output hidden health data.
4. According to claim 1, a method for collecting medical and health information based on the Internet, It is characterized in that The method for obtaining extended medical data by using the comparison data to compensate for the hidden health data includes adding the hidden data to the corresponding comparison data to obtain the extended medical data.
5. According to the method for collecting medical and health information based on the Internet as claimed in claim 1, It is characterized in that The method of performing association analysis on the expanded medical data according to the knowledge graph to obtain health data includes: Select and expand medical data based on the medical and health knowledge graph to obtain related candidate groups; Divide the associated candidate groups into key data sets and general data sets, and calculate the importance of key data: The fuzzy neighborhood radius of the key data h is The association decision is Q, the key data set is H, and the fuzzy neighborhood radius is The importance of key data h in key data set H under association decision Q is The bias weight is Key data The fuzzy neighborhood granularity of the key data set is Key data The fuzzy neighborhood granularity of the association decision is The amount of expanded medical data is g; If the importance of key data is greater than 0, the key data is filtered as healthy data, and the calculation is continued until the key data set is traversed; Compute the interaction gain for a general dataset: The interaction gain of general data a with respect to general data set A is The fuzzy neighborhood particle of general data d with respect to general data a is The fuzzy neighborhood particle size of general data d for association decision is The general data set at the t-1th moment is A t-1 , general data d about general data set A t-1 The fuzzy neighborhood particle is General Dataset A t-1 About the general dataset A t-1 The fuzzy neighborhood particle is If the interaction gain is greater than 0, the general data has an interaction, and the contrast of the general data with the interaction is calculated: The first general data with interaction is a ι1 , the second general data with interaction a ι2 , general data d about general data a ι1 The fuzzy neighborhood particle is General data d about general data a ι2 The fuzzy neighborhood particle is If the contrast is greater than 0, the general data with interaction is filtered as the influencing data until all the general data with interaction are traversed.
6. The method for collecting medical and health information based on the Internet according to claim 1, It is characterized in that The method for constructing a medical health information collection model according to the impact data includes: The medical and health information collection model includes rolling window partitioning algorithm, anomaly detection algorithm, association rule learning algorithm, recurrent neural network algorithm, and attention mechanism; The rolling window partitioning algorithm uses a fixed-size window to roll over the impact data, using the impact data within the window as the training set and the remaining impact data as the test set; The anomaly detection algorithm searches for abnormal data based on the radius of the training set data from the hypersphere to obtain deviation data; The association rule learning algorithm obtains drug interaction results based on the association of training data; The recurrent neural network algorithm learns the patterns of drug interaction results and deviation data; The attention mechanism uses test data to optimize the hyperparameters of the recurrent neural network algorithm.
7. The method for collecting medical and health information based on the Internet according to claim 1, It is characterized in that The method for optimizing the medical health information collection model comprises: Calculate the convergence factor and search for migration positions: The convergence factor is The maximum number of iterations is t max , the constant is β, the number of iterations is t, the random number from 0 to 1 is r, and the best search position of the zth particle is S bt , the search position of the zth particle in the tth iteration is S z (t), the zth search migration position is Y z , the control coefficient is γ, and the x-axis coordinate value of the particle is u; If the spiral turn radius is less than the position update threshold, calculate a new search method: P(t)=S z (t)+α[c 1 r 1 (S bt -S z (t))+c 2 r 2 (S gs -S z (t))] The first learning factor is c 1 , the second learning factor is c 2 , a random number between 0 and 1 is r 1 、r 2 , the best historical position of the individual in the tth iteration is S gs , the genetic factor is α, the new search method for the tth iteration is P(t), and the updated position is calculated: S z (t)=(Y z ·u·b·q)+S bt The y coordinate value of the particle is b, the z coordinate value of the particle is q, and the Levy coefficient is The updated position is S nw,z (t), the step size scaling factor is δ, and the random number is r 3 , the position update threshold is υ, and the gamma function is If the fitness of the updated position is less than that before updating, continue the iteration; otherwise, keep the updated position.
8. A medical and health information collection system based on the Internet, It is characterized in that include: Data acquisition module: used to obtain time-series medical data and standard reference data on the Internet, and pre-process the time-series medical data to obtain a user medical data set; the time-series medical data includes user basic information and user medical data; Matching analysis module: used for performing fuzzy matching on the user medical data set and the standard reference data to obtain comparison data, and performing impact analysis on the user medical data set to obtain hidden health data; Compensation screening module: used to use the comparison data to compensate the hidden health data to obtain extended medical data, and to perform association analysis on the extended medical data according to the knowledge graph to obtain health data; Modeling module: used to build a medical health information collection model based on the influencing data, optimize the medical health information collection model, input the information to be collected into the optimized medical health information collection model, and output the collection results.
