Internet-based health information intelligent consultation system and method
Through the Internet-based intelligent health information consultation system, deep learning technology is used to perform semantic analysis of physical examination reports and inquiry texts, and a personalized health management plan is generated, which solves the problem of non-medical professionals in understanding physical examination reports and provides targeted health Management guidance to meet self-care needs.
Patent Information
- Application Number
- CN202411156246.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-22
- Publication Date
- 2024-11-19
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
After the existing health examination, non-medical professionals find it difficult to understand complex medical indicator data, lack in-depth health consulting services, and cannot obtain targeted health management guidance.
The Internet-based intelligent health information consultation system is adopted to generate personalized health management plans, including dietary adjustments and exercise plans through semantic analysis of user physical examination reports and inquiry text information, combined with deep learning technology.
Provide users with scientific and personalized health management advice to meet the growing self-care needs and help users effectively manage their health status.
Smart Images

Figure CN118983100A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of intelligent health management, and more specifically, to an Internet-based health information intelligent consulting system and method. Background Art
[0002] With the improvement of living standards, people pay more and more attention to health care, and the demand for self-care is increasing. However, the current health care service model and channels are far from meeting people's needs. In addition to medical institutions, there are still few in-depth health consultations based on the information obtained from people's health examinations. It is difficult for non-medical professionals to understand the meaning of the complex medical indicator data in the physical examination, let alone obtain targeted, authoritative and scientific health care guidance.
[0003] Therefore, an Internet-based intelligent health information consultation solution is needed. Summary of the invention
[0004] In order to solve the above technical problems, the present application is proposed. The embodiments of the present application provide an Internet-based health information intelligent consultation system and method.
[0005] According to one aspect of the present application, there is provided an Internet-based health information intelligent consultation system, which includes: The user-related data collection module is used to obtain the user's physical examination report and the inquiry text information raised by the user regarding the content of the physical examination report; A user physical examination report semantic feature encoding module, used for performing semantic feature encoding on the user physical examination report to obtain a user physical examination report semantic association understanding feature matrix; A user inquiry data semantic feature encoding module is used to perform semantic feature encoding on the inquiry text information raised by the user regarding the content of the physical examination report to obtain a user health concern semantic understanding feature vector; A user physical examination report-inquiry data feature association module, used for multiplying the user health concern semantic understanding feature vector and the user physical examination report semantic association understanding feature matrix to obtain a user health concern-physical condition association feature vector; The health management program generation module is used to generate a health management program for user reference based on the user's health concern-physical condition association feature vector.
[0006] According to another aspect of the present application, there is provided an Internet-based health information intelligent consultation method, which includes: Obtain the user's physical examination report and the text information of the user's inquiry regarding the content of the physical examination report; Performing semantic feature encoding on the user's physical examination report to obtain a semantic association understanding feature matrix of the user's physical examination report; Performing semantic feature encoding on the inquiry text information raised by the user regarding the content of the physical examination report to obtain a user health concern semantic understanding feature vector; Multiplying the user health concern semantic understanding feature vector and the user physical examination report semantic association understanding feature matrix to obtain a user health concern-physical condition association feature vector; Based on the user's health concern-physical condition association feature vector, a health management plan is generated for the user's reference.
[0007] Compared with the prior art, the Internet-based intelligent health information consultation system and method provided by the present application adopts data analysis technology based on deep learning, and performs semantic analysis on the user's physical examination report to understand the user's physical health status, and at the same time, performs semantic analysis on the inquiry text information raised by the user regarding the content of the physical examination report to understand the user's health concerns, and then considers the user's physical health status and the user's health concerns together, and finally outputs a health management recommendation plan including diet adjustment, exercise plan, etc. for the user's reference. In this way, through intelligent data processing and analysis, it is possible to provide users with personalized health management plans to meet the growing demand for self-care. BRIEF DESCRIPTION OF THE DRAWINGS
[0008] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.
[0009] Figure 1 This is a system block diagram of an Internet-based health information intelligent consultation system according to an embodiment of the present application.
[0010] Figure 2 It is a block diagram of a semantic feature encoding module for a user's physical examination report in an Internet-based intelligent health information consultation system according to an embodiment of the present application.
[0011] Figure 3 This is a block diagram of a user physical examination report semantic association understanding feature generation unit in an Internet-based health information intelligent consultation system according to an embodiment of the present application.
[0012] Figure 4 It is a block diagram of a user inquiry data semantic feature encoding module in an Internet-based health information intelligent consultation system according to an embodiment of the present application.
[0013] Figure 5This is a block diagram of a health management program generation module in an Internet-based health information intelligent consultation system according to an embodiment of the present application.
[0014] Figure 6 The present invention is a flowchart of an Internet-based intelligent health information consultation method according to an embodiment of the present application. DETAILED DESCRIPTION
[0015] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.
[0016] With the rapid development of social economy and the significant improvement of living standards, people are paying more and more attention to their own health, and the emphasis on health care has reached a new height. The demand for self-care is becoming increasingly urgent and increasing. However, the current health care service model and approach have obvious limitations, and are far from fully meeting people's actual needs. Specifically, in the field of health examinations, in addition to professional medical institutions that can provide basic physical examination services, there are very few institutions or platforms that can provide in-depth and comprehensive health consulting services based on the rich information obtained from people's health examinations. Physical examination reports often contain complex medical indicator data. For non-medical professionals, it is undoubtedly difficult to accurately understand the meaning of these data, not to mention obtaining targeted, authoritative and scientific health care guidance suggestions based on them. This makes people often in a state of confusion and confusion when facing physical examination reports, and they cannot effectively convert physical examination results into practical health management actions, which greatly affects people's grasp and improvement of their own health status. Therefore, an Internet-based health information intelligent consultation solution is expected.
[0017] In recent years, deep learning and neural networks have been widely used in computer vision, natural language processing, text signal processing and other fields. In addition, deep learning and neural networks have also demonstrated a level close to or even beyond that of humans in image classification, object detection, semantic segmentation, text translation and other fields. The development of deep learning and neural networks has provided new solutions and solutions for intelligent health information consultation based on the Internet.
[0018] Figure 1 FIG. 1 is a system block diagram of an Internet-based health information intelligent consultation system according to an embodiment of the present application. Figure 1As shown, in the Internet-based health information intelligent consulting system 100, it includes: a user-related data collection module 110, which is used to obtain the user's physical examination report and the inquiry text information raised by the user regarding the content of the physical examination report; a user physical examination report semantic feature encoding module 120, which is used to perform semantic feature encoding on the user's physical examination report to obtain a user physical examination report semantic association understanding feature matrix; a user inquiry data semantic feature encoding module 130, which is used to perform semantic feature encoding on the inquiry text information raised by the user regarding the content of the physical examination report to obtain a user health concern semantic understanding feature vector; a user physical examination report-inquiry data feature association module 140, which is used to multiply the user health concern semantic understanding feature vector and the user physical examination report semantic association understanding feature matrix to obtain a user health concern-physical condition association feature vector; a health management plan generation module 150, which is used to generate a health management plan for user reference based on the user health concern-physical condition association feature vector.
[0019] Specifically, in the technical solution of the present application, the user's physical examination report and the inquiry text information raised by the user for the content of the physical examination report are first obtained. It should be understood that the user's physical examination report generally contains the user's basic personal information, physiological index parameter inspection information, detailed blood, urine, body biochemical index inspection information, imaging examination results and health assessment opinions. The inquiry text information raised by the user for the content of the physical examination report generally includes the user's question information for specific indicators, seeking lifestyle advice, etc. The user's physical examination report provides the user's health status and physiological indicators, and the user's inquiry text reflects the user's concerns and needs. By analyzing the specific data in the physical examination report and combining the user's inquiry situation, the system can identify the user's health risks and concerns, thereby generating a targeted health management plan. The generated scientific and personalized health management plan can effectively guide the user's health management behavior. In particular, in a specific embodiment of the present application, the user's physical examination report can be obtained by applying for the corresponding physical examination data access right to the physical examination center, and the inquiry text information raised by the user for the content of the physical examination report can be obtained from the database of the health consultation system.
[0020] Figure 2 FIG. 1 is a block diagram of a user physical examination report semantic feature encoding module in an Internet-based health information intelligent consultation system according to an embodiment of the present application. Figure 2As shown, the user physical examination report semantic feature encoding module 120 includes: a user physical examination report preprocessing unit 121, used to preprocess the user physical examination report to obtain a preprocessed user physical examination report; a user physical examination report semantic association understanding feature generation unit 122, used to perform feature extraction processing and semantic association processing on the preprocessed user physical examination report to obtain the user physical examination report semantic association understanding feature matrix.
[0021] The user's physical examination report will contain some noise data that is not very important for understanding the user's physical health, such as the physical examination date and physical examination institution information in the report. There are also multiple expressions such as text and numbers in the physical examination report. The diversified data expression is not conducive to subsequent data processing and semantic analysis. Therefore, in the technical solution of this application, it is necessary to preprocess the user's physical examination report to obtain a preprocessed user physical examination report. The noise data in the original physical examination report can be removed through preprocessing, and the key information in the physical examination report can be converted into textual expression, which is convenient for the subsequent semantic understanding of the content in the user's physical examination report, thereby improving the understanding of the user's physical condition.
[0022] Figure 3 FIG. 1 is a block diagram of a user physical examination report semantic association understanding feature generation unit in an Internet-based health information intelligent consultation system according to an embodiment of the present application. Figure 3 As shown, the user physical examination report semantic association understanding feature generation unit 122 includes: a user physical examination report keyword semantic understanding feature generation subunit 1221, which is used to pass the preprocessed user physical examination report through a user physical examination report semantic feature encoder after word segmentation to obtain multiple user physical examination report keyword semantic understanding feature vectors; a user physical examination report keyword semantic understanding feature arrangement subunit 1222, which is used to arrange the multiple user physical examination report keyword semantic understanding feature vectors in two dimensions to obtain a user physical examination report keyword semantic understanding two-dimensional input matrix; a user physical examination report keyword semantic association subunit 1223, which is used to pass the user physical examination report keyword semantic understanding two-dimensional input matrix through a bidirectional user physical examination report association semantic feature encoder to obtain the user physical examination report semantic association understanding feature matrix.
[0023] The pre-processed user physical examination report is a large paragraph of text information. In order to make the content in the user physical examination report easier to learn and understand, it is necessary to first perform word segmentation on the pre-processed user physical examination report. Through the word segmentation operation, the long text of the physical examination report is decomposed into smaller word units, which helps to identify the relationship between different words, and then improves the understanding of the pre-processed user physical examination report text as a whole. Next, the pre-processed user physical examination report after word segmentation is passed through the user physical examination report semantic feature encoder to obtain multiple user physical examination report keyword semantic understanding feature vectors. In this application, the user physical examination report semantic feature encoder is a converter-based semantic encoder including an embedding layer. The embedding layer can convert the user physical examination content keywords after word segmentation into low-dimensional word vector representations, that is, converting high-dimensional sparse vocabulary into low-dimensional dense vectors, which can reduce computational complexity while retaining important information. And the converter uses a self-attention mechanism, which can pay attention to important data in the sequence when processing input sequence data, thereby enhancing the encoder's ability to understand the context of the input sequence data. In short, the transformer can effectively capture long-range dependencies, which helps capture the complex semantics in physical examination reports, such as understanding the relationship between "high blood sugar" and "diabetes risk".
[0024] In an embodiment of the present application, a possible implementation method for obtaining multiple user physical examination report keyword semantic understanding feature vectors after word segmentation of the preprocessed user physical examination report may be: word segmentation processing is performed on the preprocessed user physical examination report to obtain multiple user physical examination report keywords; the multiple user physical examination report keywords are passed through the embedding layer to convert each user physical examination report keyword in the multiple user physical examination report keywords into a user physical examination report keyword embedding vector to obtain a sequence of user physical examination report keyword embedding vectors, wherein the embedding layer uses a learnable embedding matrix to embed-code each user physical examination report keyword; the sequence of user physical examination report keyword embedding vectors is input into a converter-based semantic encoder to obtain the multiple user physical examination report keyword semantic understanding feature vectors.
[0025] In order to combine different feature information in the user's physical examination report, the multiple user's physical examination report keyword semantic understanding feature vectors can be arranged in two dimensions to obtain a user's physical examination report keyword semantic understanding two-dimensional input matrix. After organizing multiple feature vectors into a matrix, multiple vectors can be processed at the same time, which is more efficient than processing each vector one by one, simplifying the data processing process.
[0026] Next, the user physical examination report keyword semantic understanding two-dimensional input matrix is passed through a two-way user physical examination report associated semantic feature encoder to obtain a user physical examination report semantic association understanding feature matrix. In the present application, the two-way user physical examination report associated semantic feature encoder is a convolutional neural network model including a two-way attention mechanism module. Those of ordinary skill in the art should know that the two-way attention mechanism is a technology that enhances the model's ability to understand the context of input data, and its core idea is to dynamically adjust the representation of each element by calculating the relationship between the elements in the input data. Unlike the traditional unidirectional attention mechanism, the two-way attention mechanism simultaneously considers the context information of each element in the input sequence, which means that when processing each feature, the model not only pays attention to the feature itself, but also considers other features related to it, thereby obtaining a more comprehensive semantic understanding. By passing the user physical examination report keyword semantic understanding two-dimensional input matrix through a two-way user physical examination report associated semantic feature encoder, it can take into account the context information in the user physical examination report keyword semantic understanding two-dimensional input matrix, that is, the row direction and column direction of the matrix are considered simultaneously, thereby improving the model's ability to understand complex semantics in the user physical examination report. It is also worth mentioning that through bidirectional processing, the model can better adapt to different types of input data, so that it can still maintain good performance when faced with health data from physical examination reports of different users.
[0027] In an embodiment of the present application, a possible implementation method of passing the user physical examination report keyword semantic understanding two-dimensional input matrix through a bidirectional user physical examination report associated semantic feature encoder to obtain a user physical examination report semantic associated understanding feature matrix may be: pooling the user physical examination report keyword semantic understanding two-dimensional input matrix along the horizontal direction and the vertical direction respectively to obtain a first user physical examination report keyword semantic understanding pooling vector and a second user physical examination report keyword semantic understanding pooling vector; associatively encoding the first user physical examination report keyword semantic understanding pooling vector and the second user physical examination report keyword semantic understanding pooling vector to obtain a user physical examination report semantic understanding bidirectional association matrix; inputting the user physical examination report semantic understanding bidirectional association matrix into a Sigmoid activation function to obtain a bidirectional association weight matrix; calculating the positional point multiplication between the bidirectional association weight matrix and the user physical examination report keyword semantic understanding two-dimensional input matrix to obtain the user physical examination report semantic association understanding feature matrix.
[0028] Figure 4 FIG. 1 is a block diagram of a user query data semantic feature encoding module in an Internet-based health information intelligent consultation system according to an embodiment of the present application. Figure 4As shown, the user inquiry data semantic feature encoding module 130 includes: an inquiry text information word embedding unit 131, which is used to segment the inquiry text information raised by the user regarding the content of the physical examination report and pass it through an embedding layer to obtain a sequence of user health concern inquiry keyword embedding vectors; a user health concern semantic understanding unit 132, which is used to pass the sequence of user health concern inquiry keyword embedding vectors through a bidirectional long short-term memory neural network model to obtain the user health concern semantic understanding feature vector.
[0029] Taking into account that the user's inquiries are usually expressed in natural language and contain a variety of grammatical and semantic information, in order to fully understand the user's inquiry intention, it is necessary to first segment the inquiry text information raised by the user regarding the content of the physical examination report. Through word segmentation, the keywords in the user's inquiry can be extracted. These keywords are usually the health problems or symptoms that the user is most concerned about. Identifying these keywords helps the system respond quickly to user needs. Next, the inquiry text after word segmentation is passed through an embedding layer. The embedding layer can convert the inquiry text after word segmentation into a vector representation. This vector representation can capture the semantic relationship between words, so that the algorithm model can better understand the user's concerns. Specifically, in the technical solution of the present application, it is necessary to segment the inquiry text information raised by the user regarding the content of the physical examination report and then pass it through an embedding layer to obtain a sequence of user health concern inquiry keyword embedding vectors.
[0030] The keyword embedding vector sequence of the user's health concern inquiry is generated by the user's inquiry text through word segmentation and embedding layers. These vectors not only contain the user's inquiry content, but also contain the semantic relationship between words. In order to deeply understand the user's health needs and provide them with a more personalized and effective health management plan, it is necessary to first pass the sequence of the user's health concern inquiry keyword embedding vector through a bidirectional long short-term memory neural network model to obtain the user's health concern semantic understanding feature vector. Long short-term memory neural network (LSTM) is a special recurrent neural network (RNN) that can effectively process and predict time series data. LSTM solves the problem of gradient vanishing or exploding in traditional RNN in long sequence data by introducing a gating mechanism, enabling it to remember long-term dependency information. Bidirectional LSTM adopts a bidirectional structure in the processing of input sequences, encoding the input sequence from the forward and backward directions of the sequence respectively. This structure enables the model to consider the previous and next contextual information at the same time, thereby improving the accuracy of semantic understanding. The generated user health concern semantic understanding feature vector not only includes specific health problems, but also reflects information such as the user's degree of concern.
[0031] The user health concern semantic understanding feature vector reflects the user's concerns about health issues, while the user's physical examination report semantic association understanding feature matrix contains the user's physical health status information. In order to effectively combine the user's concerns with his physical condition and form a comprehensive feature representation, the technical solution of the present application needs to multiply the user health concern semantic understanding feature vector and the user physical examination report semantic association understanding feature matrix to obtain the user health concern-physical condition association feature vector. The multiplication operation can capture the relationship between the user's health concerns and his physical condition. In this way, the system can identify which health concerns are closely related to the user's physical examination results. In short, after multiplying and combining these two feature data, more accurate health management plan content can be generated based on the correlation between the health issues that the user is concerned about and the physical examination report.
[0032] Figure 5 FIG. 1 is a block diagram of a health management program generation module in an Internet-based health information intelligent consultation system according to an embodiment of the present application. Figure 5 As shown, the health management program generation module 150 includes: a feature vector optimization unit 151, which is used to perform back-propagation correction on the user health concern-physical condition association feature vector based on feature space learning to obtain an optimized user health concern-physical condition association feature vector; a feature vector parsing unit 152, which is used to pass the optimized user health concern-physical condition association feature vector through a health management program generator to obtain a generation result, and the generation result is a health management program for user reference.
[0033] In particular, in the technical solution of the present application, it is considered that the preprocessing quality of the physical examination report directly affects the accuracy of subsequent feature extraction. If the preprocessing step fails to effectively remove noise or standardize data, it may cause data inconsistency, thereby affecting the distribution of feature vectors. The word segmentation of the physical examination report and the user inquiry text may have errors due to the diversity and complexity of the terms. In addition, the semantic feature encoder and the embedding layer may not be able to fully capture all the semantic information in the text, resulting in incomplete or biased feature vectors. The bidirectional user physical examination report associated semantic feature encoder and the bidirectional long short-term memory neural network model may have limitations when processing long texts or complex semantics, and cannot fully understand and encode all relevant information in the text. Local abnormal distribution may be caused by error accumulation in the above steps, inaccurate feature extraction, model overfitting or underfitting, etc. The user health concern-physical condition associated feature vector has a local abnormal distribution, which causes it to have a category outlier offset relative to the predetermined generation scheme mode of the generator when generating a scheme through the health management scheme generator, affecting the generation iteration effect. Based on this, it is necessary to perform back-propagation correction based on feature space learning on the user health concern-physical condition association feature vector to obtain an optimized user health concern-physical condition association feature vector.
[0034] Specifically, in an embodiment of the present application, the user health concern-physical condition associated feature vector is subjected to back-propagation correction based on feature space learning to obtain an optimized user health concern-physical condition associated feature vector, including: determining a decoding weight matrix of the health management solution generator; performing matrix decoupling on the decoding weight matrix in units of row vectors to obtain a set of decoding weight vectors; calculating a metric learning loss information item between each decoding weight vector in the set of decoding weight vectors and the user health concern-physical condition associated feature vector to obtain a metric learning compensation representation vector; inputting the metric learning compensation representation vector into an activation function to obtain a probabilistic metric learning compensation weight vector; calculating the positional point multiplication of the probabilistic metric learning compensation weight vector and the user health concern-physical condition associated feature vector to obtain a point product feature vector, and calculating the weighted sum of the point product feature vector and the user health concern-physical condition associated feature vector to obtain the optimized user health concern-physical condition associated feature vector.
[0035] More specifically, in an embodiment of the present application, calculating the metric learning loss information item between each decoding weight vector in the set of decoding weight vectors and the user health concern-physical condition associated feature vector to obtain a metric learning compensation representation vector includes: calculating the metric learning loss information item between each decoding weight vector in the set of decoding weight vectors and the user health concern-physical condition associated feature vector using the following formula; wherein the formula is:
[0036] in, represents the first in the set of decoding weight vectors The decoding weight vector The eigenvalues at the positions, The first character representing the user's health concern-physical condition association feature vector The eigenvalues at the positions, represents the length of the feature vector, represents the metric learning compensation representation vector The eigenvalues at the positions, Represents a logarithmic function.
[0037] That is, the user health concern-physical condition associated feature vector is corrected by back propagation based on feature space learning, which first determines the decoding weight matrix of the health management program generator to understand the health program generator to distinguish different categories of features. Subsequently, the decoding weight matrix is decoupled by row vectors to obtain a set of decoding weight vectors to allow the model to independently analyze the weight of each category and provide more refined control for feature compensation. Next, the metric learning loss information item is calculated. This process involves quantitative analysis of the interaction between each category weight vector and the user health concern-physical condition associated feature vector. Through the calculation of a specific loss function, the inconsistency between the feature vector and the category weight is captured, providing a quantitative indicator for compensation. Next, the metric learning compensation representation vector is input into the activation function, nonlinearly transformed, and converted into a probabilistic form. The application of the activation function introduces nonlinearity, enhances the model's ability to interpret the compensation weight, and makes the compensation weight more consistent with the actual distribution characteristics. Then, the probabilistic metric learning compensation weight vector is calculated using the metric learning compensation representation vector processed by the activation function. This step provides probabilistic guidance for the adjustment of feature vectors in a probabilistic form, making the optimization process more consistent with statistical characteristics, thereby enhancing the accuracy of the model's feature adjustment. Finally, by calculating the dot product of the probabilistic metric learning compensation weight vector and the user's health concern-physical condition associated feature vector, and then calculating the weighted sum with the original user's health concern-physical condition associated feature vector, the optimized user health concern-physical condition associated feature vector is obtained. This step combines the original feature information and compensation information to generate an adjusted feature vector, reducing the impact of abnormal distribution and improving the accuracy of the generation scheme.
[0038] Finally, the optimized user health concern-physical condition associated feature vector is passed through the health management program generator to obtain a generated result, and the generated result is a health management program for user reference. It should be understood that the optimized user health concern-physical condition associated feature vector contains the user's health concerns and rich content semantic information in the physical examination results, but does not directly obtain a health management program that can be used for reference. The health management program generator is a model based on deep learning. Through the physical examination report data, user consultation text data and the corresponding generated health management program used for training, the connection between the content features in the physical examination report data and the user consultation text data and the generated health management program can be learned. By inputting the optimized user health concern-physical condition associated feature vector into the health management program generator, a personalized health management program that meets the specific needs of the user can be generated. This customized service can better meet the user's health management needs. The health management program usually includes specific action suggestions, such as diet adjustment, exercise plan, etc. It provides scientific and reasonable health management suggestions to help users make more informed health decisions.
[0039] In summary, the Internet-based health information intelligent consultation system 100 based on the embodiment of the present application is explained, which adopts data analysis technology based on deep learning, performs semantic analysis on the user's physical examination report to understand the user's physical health status, and performs semantic analysis on the inquiry text information raised by the user regarding the content of the physical examination report to understand the user's health concerns, and then considers the user's physical health status and the user's health concerns together, and finally outputs a health management recommendation plan including diet adjustment, exercise plan, etc. for the user's reference. In this way, through intelligent data processing and analysis, it is possible to provide users with personalized health management plans to meet the growing demand for self-care.
[0040] As described above, the Internet-based health information intelligent consulting system 100 according to the embodiment of the present application can be implemented in various terminal devices, such as a server for Internet-based health information intelligent consulting, etc. In one example, the Internet-based health information intelligent consulting system 100 according to the embodiment of the present application can be integrated into the terminal device as a software module and / or a hardware module. For example, the Internet-based health information intelligent consulting system 100 can be a software module in the operating system of the terminal device, or it can be an application developed for the terminal device; of course, the Internet-based health information intelligent consulting system 100 can also be one of the many hardware modules of the terminal device.
[0041] Alternatively, in another example, the Internet-based health information intelligent consulting system 100 and the terminal device may also be separate devices, and the Internet-based health information intelligent consulting system 100 may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.
[0042] Figure 6 FIG. 1 is a flow chart of an Internet-based intelligent health information consultation method according to an embodiment of the present application. Figure 6As shown, in the Internet-based intelligent consultation method for health information, the following are included: S110, obtaining a user's physical examination report and text information of inquiries raised by the user regarding the contents of the physical examination report; S120, performing semantic feature encoding on the user's physical examination report to obtain a semantic association understanding feature matrix of the user's physical examination report; S130, performing semantic feature encoding on the text information of inquiries raised by the user regarding the contents of the physical examination report to obtain a user health concern semantic understanding feature vector; S140, multiplying the user health concern semantic understanding feature vector and the user physical examination report semantic association understanding feature matrix to obtain a user health concern-physical condition association feature vector; S150, generating a health management plan for user reference based on the user health concern-physical condition association feature vector.
[0043] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned Internet-based health information intelligent consultation method have been referred to above. Figures 1 to 5 The description of the Internet-based health information intelligent consultation system has been introduced in detail, and therefore, its repeated description will be omitted.
[0044] In summary, the Internet-based intelligent health information consultation method based on the embodiment of the present application is explained, which adopts data analysis technology based on deep learning, performs semantic analysis on the user's physical examination report to understand the user's physical health status, and at the same time, performs semantic analysis on the inquiry text information raised by the user regarding the content of the physical examination report to understand the user's health concerns, and then considers the user's physical health status and the user's health concerns together, and finally outputs a health management recommendation plan including diet adjustment, exercise plan, etc. for the user's reference. In this way, through intelligent data processing and analysis, it is possible to provide users with personalized health management plans to meet the growing demand for self-care.
Claims
1. An Internet-based health information intelligent consulting system, characterized in that: include: The user-related data collection module is used to obtain the user's physical examination report and the inquiry text information raised by the user regarding the content of the physical examination report; A user physical examination report semantic feature encoding module, used for performing semantic feature encoding on the user physical examination report to obtain a user physical examination report semantic association understanding feature matrix; A user inquiry data semantic feature encoding module is used to perform semantic feature encoding on the inquiry text information raised by the user regarding the content of the physical examination report to obtain a user health concern semantic understanding feature vector; A user physical examination report-inquiry data feature association module, used for multiplying the user health concern semantic understanding feature vector and the user physical examination report semantic association understanding feature matrix to obtain a user health concern-physical condition association feature vector; The health management program generation module is used to generate a health management program for user reference based on the user's health concern-physical condition association feature vector.
2. The Internet-based health information intelligent consulting system according to claim 1, characterized in that: The user physical examination report semantic feature encoding module includes: A user physical examination report preprocessing unit, used for preprocessing the user physical examination report to obtain a preprocessed user physical examination report; The user physical examination report semantic association understanding feature generation unit is used to perform feature extraction processing and semantic association processing on the pre-processed user physical examination report to obtain the user physical examination report semantic association understanding feature matrix.
3. The Internet-based health information intelligent consulting system according to claim 2, characterized in that: The user physical examination report semantic association understanding feature generation unit includes: A user physical examination report keyword semantic understanding feature generating subunit, used for segmenting the preprocessed user physical examination report and passing it through a user physical examination report semantic feature encoder to obtain a plurality of user physical examination report keyword semantic understanding feature vectors; A user physical examination report keyword semantic understanding feature arrangement subunit, used for two-dimensionally arranging the plurality of user physical examination report keyword semantic understanding feature vectors to obtain a user physical examination report keyword semantic understanding two-dimensional input matrix; The user physical examination report keyword semantic association subunit is used to pass the user physical examination report keyword semantic understanding two-dimensional input matrix through a bidirectional user physical examination report associated semantic feature encoder to obtain the user physical examination report semantic association understanding feature matrix.
4. The Internet-based health information intelligent consulting system according to claim 3 is characterized in that: The user query data semantic feature encoding module includes: The query text information word embedding unit is used to segment the query text information raised by the user regarding the content of the physical examination report and pass it through the embedding layer to obtain a sequence of embedding vectors of the user's health concern query keywords; The user health concern semantic understanding unit is used to pass the sequence of the user health concern query keyword embedding vectors through a bidirectional long short-term memory neural network model to obtain the user health concern semantic understanding feature vector.
5. The Internet-based health information intelligent consulting system according to claim 4, characterized in that: The user physical examination report semantic feature encoder is a converter-based semantic encoder including an embedding layer, and the bidirectional user physical examination report associated semantic feature encoder is a convolutional neural network model including a bidirectional attention mechanism module.
6. The Internet-based health information intelligent consulting system according to claim 5, characterized in that: The health management program generation module includes: A feature vector optimization unit, used for performing back-propagation correction based on feature space learning on the user health concern-physical condition association feature vector to obtain an optimized user health concern-physical condition association feature vector; The feature vector parsing unit is used to pass the optimized user health concern-physical condition association feature vector through a health management program generator to obtain a generation result, and the generation result is a health management program for user reference.
7. The Internet-based health information intelligent consulting system according to claim 6, characterized in that: The feature vector optimization unit is used for: Determining a decoding weight matrix of the health management solution generator; Decoupling the decoding weight matrix in units of row vectors to obtain a set of decoding weight vectors; Calculating a metric learning loss information item between each decoding weight vector in the set of decoding weight vectors and the user health concern-physical condition associated feature vector to obtain a metric learning compensation representation vector; Inputting the metric learning compensation representation vector into an activation function to obtain a probabilistic metric learning compensation weight vector; Calculate the positional point multiplication of the probabilistic metric learning compensation weight vector and the user health concern-physical condition association feature vector to obtain a point product feature vector, and calculate the weighted sum of the point product feature vector and the user health concern-physical condition association feature vector to obtain the optimized user health concern-physical condition association feature vector.
8. An Internet-based health information intelligent consultation method, characterized in that: include: Obtain the user's physical examination report and the text information of the user's inquiry regarding the content of the physical examination report; Performing semantic feature encoding on the user's physical examination report to obtain a semantic association understanding feature matrix of the user's physical examination report; Performing semantic feature encoding on the inquiry text information raised by the user regarding the content of the physical examination report to obtain a user health concern semantic understanding feature vector; Multiplying the user health concern semantic understanding feature vector and the user physical examination report semantic association understanding feature matrix to obtain a user health concern-physical condition association feature vector; Based on the user's health concern-physical condition association feature vector, a health management plan is generated for the user's reference.
9. The Internet-based intelligent health information consultation method according to claim 8, characterized in that: The user's physical examination report is semantically encoded to obtain a semantic association understanding feature matrix of the user's physical examination report, including: Preprocessing the user physical examination report to obtain a preprocessed user physical examination report; The pre-processed user physical examination report is subjected to feature extraction processing and semantic association processing to obtain a semantic association understanding feature matrix of the user physical examination report.
10. The Internet-based intelligent health information consultation method according to claim 9, characterized in that: Performing feature extraction processing and semantic association processing on the pre-processed user physical examination report to obtain a semantic association understanding feature matrix of the user physical examination report, including: The pre-processed user physical examination report is segmented and passed through a user physical examination report semantic feature encoder to obtain a plurality of user physical examination report keyword semantic understanding feature vectors; Arranging the plurality of user physical examination report keyword semantic understanding feature vectors in two dimensions to obtain a user physical examination report keyword semantic understanding two-dimensional input matrix; The user physical examination report keyword semantic understanding two-dimensional input matrix is passed through a bidirectional user physical examination report associated semantic feature encoder to obtain the user physical examination report semantic associated understanding feature matrix.
