Internet medical service system based on artificial intelligence

Abstract

The invention relates to the technical field of artificial intelligence, in particular to an internet medical service system based on artificial intelligence, which comprises a data platform used for integrating and managing medical service data; the case generation module is connected with the data platform, obtains examination data of the medical service data, and generates structured case data according to the examination data; the auxiliary classification module is connected with the case generation module, and the auxiliary classification module is provided with a pre-trained deep learning model and is used for outputting a focus classification result according to the case data and the examination data; and the recommendation module is connected with the auxiliary classification module and is used for outputting suggestion information according to the focus classification result and the examination data. According to the invention, a unified data platform is adopted, medical service data and artificial intelligence auxiliary functions are shared and integrated, the application depth and breadth of artificial intelligence to the medical service data are improved, and deep application of artificial intelligence in a medical service system is promoted.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an Internet medical service system based on artificial intelligence. Background Art

[0002] Traditional Internet hospital systems include mobile applications or web pages for patients and doctors, providing functions such as appointment registration, online consultation, and electronic medical record viewing. They also include backend management systems for managing various hospital businesses, including doctor scheduling, patient management, and drug management. Based on the development of AI technology, more and more Internet hospital systems will be connected to AI-assisted systems to generate diagnostic assistance results and treatment assistance suggestions. However, the medical service data and AI-assisted systems of Internet hospitals are often separated. Internet hospital systems and AI-assisted systems are usually developed by different teams and lack unified standards and interfaces, which makes data difficult to communicate with each other. In addition, the degree of intelligence is limited, and only a preliminary description of medical service data can be made. The analysis of medical service data is not thorough enough. The depth and breadth of AI-assisted systems in Internet hospital system applications are different, and the potential of AI-assisted systems cannot be fully tapped. Summary of the invention

[0003] The purpose of the present invention is to provide an Internet medical service system based on artificial intelligence, characterized in that the above technical problems are solved;

[0004] The technical problem solved by the present invention can be achieved by adopting the following technical solutions:

[0005] An Internet medical service system based on artificial intelligence, comprising:

[0006] Data platform for integrating and managing medical service data;

[0007] A case generation module, connected to the data platform, obtains the examination data of the medical service data, and generates structured case data according to the examination data;

[0008] An auxiliary classification module, connected to the case generation module, wherein the auxiliary classification module is provided with a pre-trained deep learning model and is used to output a lesion classification result according to the case data and the examination data;

[0009] A recommendation module is connected to the auxiliary classification module and is used to output recommendation information based on the lesion classification result and the inspection data.

[0010] Preferably, the inspection data includes diagnostic record information, and the case generation module includes:

[0011] A case collection unit, used for collecting the diagnosis record information;

[0012] A case cleaning unit, used for cleaning the diagnosis record information;

[0013] A natural language processing unit, connected to the case cleaning unit, for identifying and extracting case features of the diagnostic record information through a natural language processing model;

[0014] The case generation subunit is connected to the natural language processing unit and is used to fill in the case data according to the preset case template based on the case characteristics and output the case data.

[0015] Preferably, the natural language processing unit includes:

[0016] A model fine-tuning subunit, used to adjust parameters of the natural language processing model according to the diagnosis record information;

[0017] A named entity recognition model, connected to the model fine-tuning subunit, for recognizing named entity features of the diagnostic record information;

[0018] The relationship extraction model is connected to the named entity recognition model, and is used to identify the relationship between the named entity features and output the relationship recognition result of the named entity features.

[0019] Preferably, the model fine-tuning subunit is used to minimize the loss function, and the loss function is a cross entropy loss function, which is expressed as:

[0020]

[0021] Among them, y i represents the target value of the i-th sample, p(y i |x) represents the predicted probability distribution of the natural language processing model for the input diagnosis record information, x represents the diagnosis record information, and N represents the number of samples.

[0022] Preferably, the examination data includes image data, and the auxiliary classification module outputs a lesion classification result according to the case data and the image data, including:

[0023] A classification data collection unit, used for collecting the image data and the case data;

[0024] A classification data preprocessing unit, connected to the classification data collecting unit, for preprocessing the image data and the case data;

[0025] The deep learning model is connected to the classification data preprocessing unit, and is used to extract the text features of the case data and the lesion features in the image data, and fuse the text features and the lesion features to output the lesion classification result.

[0026] Preferably, the classification data collection unit is also used to collect case texts transmitted by an external system, the classification data preprocessing unit receives the case texts, and converts the case texts into the case data through preprocessing, and the classification data preprocessing unit includes:

[0027] An information filling subunit is used to fill in the missing information of the case text to generate a filled case text;

[0028] A text data processing subunit, connected to the information filling subunit, for performing text processing on the filled case text to generate a data-processed case text;

[0029] The structured data processing subunit is connected to the text data processing subunit and is used to normalize the structured data of the case text after the data processing and output the case data.

[0030] Preferably, the recommendation module generates clinical example information corresponding to the lesion classification result through a machine learning model or a rule engine, and the recommendation information includes the clinical example information.

[0031] Preferably, the medical service data also includes doctor schedule data and patient appointment data sent by an external medical system, and also includes a schedule scheduling module for adjusting the schedule according to the doctor schedule data and the patient appointment data, and the schedule scheduling module includes,

[0032] A schedule data collection unit, used for collecting the doctor's schedule data and the patient's appointment data in real time;

[0033] A scheduling unit connected to the schedule data collection unit, the scheduling unit comprising a scheduling model, the scheduling model adjusting the schedule in real time according to preset constraints and outputting schedule data;

[0034] A notification unit is connected to the scheduling unit and is used to push and notify the schedule data to an external terminal.

[0035] Preferably, it also includes a user interaction module, which is connected to the data platform and is used to transmit user information and view the recommendation information. The medical service data includes the saved user information, and the user information is distributedly stored in an anonymized and blockchain manner.

[0036] Preferably, the data platform is connected to the external medical system via a data interface module, and the data interface module integrates a universal calling interface, and receives the medical service data and sends the recommendation information via the calling interface.

[0037] Beneficial effects of the present invention: Due to the adoption of the above technical scheme, the present invention adopts a unified data platform to share and integrate medical service data with artificial intelligence auxiliary functions, and adopts a case generation module, auxiliary classification module and recommendation module adapted to medical service data to improve the depth and breadth of artificial intelligence application for medical service data, and promote the in-depth application of artificial intelligence in the medical service system. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is an architecture diagram of the Internet medical service system in an embodiment of the present invention;

[0039] Figure 2 Schematic diagram of a case generation module in an embodiment of the present invention;

[0040] Figure 3 is an architecture diagram of a natural language processing unit in an embodiment of the present invention;

[0041] Figure 4 is an architecture diagram of an auxiliary classification module in an embodiment of the present invention;

[0042] Figure 5 Schematic diagram of a classification data preprocessing unit in an embodiment of the present invention;

[0043] Figure 6 Schematic diagram of the schedule scheduling module in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0046] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0047] An Internet medical service system based on artificial intelligence, such as Figures 1 to 6 As shown, including

[0048] Data platform 1, used to integrate and manage medical service data;

[0049] The case generation module 2 is connected to the data platform 1, obtains the examination data of the medical service data, and generates structured case data according to the examination data;

[0050] The auxiliary classification module 3 is connected to the case generation module 2, and the auxiliary classification module 3 is provided with a pre-trained deep learning model 33, which is used to output the lesion classification result according to the case data and the examination data;

[0051] The recommendation module 4 is connected to the auxiliary classification module 3 and is used to output recommendation information based on the lesion classification results and examination data.

[0052] Specifically, the main function of the data platform 1 of the present invention is to integrate and manage medical service data, including diagnosis records, examination data, case data, imaging data, doctor schedule information and patient appointment information, and provide a centralized data storage and access interface;

[0053] The data platform 1 aggregates various medical service data into a unified database by connecting with external medical system interfaces, such as hospital information systems and imaging systems. Data interconnection between different systems is achieved through the data interface module 7. The data platform 1 of the present invention adopts distributed and anonymous storage, and uses blockchain technology to ensure that data cannot be tampered with and protect data security.

[0054] In a preferred embodiment, the examination data includes diagnostic record information, and the case generation module 2 includes:

[0055] A case collection unit 21, used for collecting diagnosis record information;

[0056] A case cleaning unit 22, used for cleaning the diagnosis record information;

[0057] A natural language processing unit 23, connected to the case cleaning unit 22, is used to identify and extract case features of diagnostic record information through a natural language processing model;

[0058] The case generation unit 24 is connected to the natural language processing unit 23 and is used to fill in the case data according to the preset case template based on the case characteristics and output the case data.

[0059] Specifically, the case generation module 2 generates structured case data by collecting, cleaning and processing the diagnosis record information. The case generation module 2 will automatically update the electronic medical record data according to the continuous update of the diagnosis record information, thereby reducing the doctor's manual recording workload and improving efficiency.

[0060] Diagnostic record information is stored in text form, including the patient's personal information, medical history, and diagnosis information. If the communication between the doctor and the patient is recorded in voice form, it is converted into text form through voice recognition;

[0061] Since the diagnosis record information usually contains noise information such as spelling errors, inconsistent punctuation, abbreviations, etc., the present invention performs data cleaning through the case cleaning unit 22, including removing irrelevant information, spelling correction and abbreviation expansion.

[0062] In the present invention, manually annotated data is used to train the natural language processing model, and the annotated content includes disease name, symptoms, diagnosis results, drug recommendation information, treatment plan information, etc.

[0063] In a preferred embodiment, the natural language processing unit 23 includes:

[0064] A model fine-tuning subunit 231 is used to adjust parameters of the natural language processing model according to the diagnosis record information;

[0065] A named entity recognition model 232, connected to the model fine-tuning subunit 231, is used to recognize the named entity features of the diagnosis record information;

[0066] The relationship extraction model 233 is connected to the named entity recognition model 232, and is used to identify the relationship between the named entity features and output the relationship recognition result of the named entity features.

[0067] In a preferred embodiment, the model fine-tuning subunit 231 is used to minimize the loss function, and the loss function is a cross entropy loss function, which is expressed as follows:

[0068]

[0069] Among them, y i represents the target value of the i-th sample, p(y i |x) represents the predicted probability distribution of the natural language processing model for the input diagnosis record information, x represents the diagnosis record information, and N represents the number of samples.

[0070] Specifically, the purpose of the cross entropy loss function is to measure the true label y i And the model predicts the probability p(y i |x). Specifically, the smaller the value of the loss function, the closer the model's prediction result is to the true label, and vice versa. By fine-tuning the model, the loss function is minimized, thereby optimizing the model performance.

[0071] Specifically, in order to make the pre-trained model better adapt to the diagnostic record information, it is necessary to fine-tune it on the labeled data of the diagnostic record information. The fine-tuning process adjusts the model parameters so that it can effectively recognize the terms and context of the diagnostic record information. The fine-tuning tasks include named entity recognition and relationship extraction. Named entity recognition is used to identify and mark the named entity features in the diagnostic record, such as diseases, drugs, and symptoms; relationship extraction is used to identify the relationship between different named entity features, such as the causal relationship between symptoms and diseases.

[0072] The present invention identifies key entities in text by training a named entity recognition (NER) model; for example:

[0073] The name of the disease, such as "diabetes" or "pneumonia";

[0074] Symptoms, such as "headache" or "nausea";

[0075] Medicines, such as "amoxicillin" or "aspirin."

[0076] Common NER methods include BiLSTM-CRF, which uses bidirectional LSTM and conditional random fields to simultaneously consider the dependencies between context and labels.

[0077] The task of relation extraction is to identify the relationship between entities. For example, "headache" is a symptom of "cold", or "amoxicillin" is a treatment for "pneumonia". Commonly used techniques include,

[0078] Graph Convolutional Network (GCN), a graph-based model that can effectively extract the relationship between entities;

[0079] BERT+RNN: Combine BERT to extract text features, and then use RNN to process the relationship between entities.

[0080] The trained natural language processing model is used to extract structured information from the diagnostic record information, and the structured information extracted from the diagnostic record information is automatically filled into the medical record template to generate a standardized electronic medical record.

[0081] In a preferred embodiment, the examination data includes image data, and the auxiliary classification module 3 outputs the lesion classification result according to the case data and the image data, including:

[0082] A classification data collection unit 31, used for collecting image data and case data;

[0083] The classification data preprocessing unit 32 is connected to the classification data collecting unit 31 and is used to preprocess the image data and case data;

[0084] The deep learning model 33 is connected to the classification data preprocessing unit 32, and is used to extract the text features of the case data and the lesion features in the image data, and fuse the text features and the lesion features to output the lesion classification results.

[0085] Specifically, the auxiliary classification module 3 transmits the lesion classification results to the user interaction module 6 to provide personalized health advice and warnings. The present invention uses optimization algorithms, such as genetic algorithms and particle swarm optimization, to find the best parameter configuration in multi-module collaboration to improve the overall performance of the system.

[0086] Specifically, the classification data preprocessing unit 32 includes:

[0087] The image data is resized, for example, to a fixed size, and the grayscale values ​​are normalized so that the brightness and contrast of the image meet the input requirements of the deep learning model 33.

[0088] Denoising: Use filters such as Gaussian filters to remove image noise and improve image quality.

[0089] Enhance the data and apply image enhancement techniques such as rotation, flipping, translation, scaling, etc. to expand the training set and help the model generalize better.

[0090] The present invention uses common deep learning architectures such as ResNet, DenseNet, and Inception to process image data. ResNet solves the gradient vanishing problem of deep neural networks through jump connections, and DenseNet further enhances information flow through dense connections, which is suitable for medical image analysis.

[0091] For the text data in the case, use pre-trained language models (such as BERT, GPT, etc.) to perform semantic understanding and analysis of medical records.

[0092] The present invention splices the features extracted from the image data and the case data and inputs them into a common neural network for training. Alternatively, two models are trained independently, one of which processes the image data and the other processes the case data, and then the output results of the two models are fused, for example, using a weighted average, a voting mechanism, etc. After the deep learning model 33 is trained and tuned, the system outputs the lesion classification results based on the image data and the case data.

[0093] In a preferred embodiment, the classification data collection unit 31 is also used to collect case texts transmitted by an external system, and the classification data preprocessing unit 32 receives the case texts and converts the case texts into case data through preprocessing. The classification data preprocessing unit 32 includes:

[0094] The information filling subunit 321 is used to fill in the missing information of the case text and generate a filled case text;

[0095] The text data processing subunit 322 is connected to the information filling subunit 321, and is used to perform text processing on the filled case text to generate a data-processed case text;

[0096] The structured data processing subunit 323 is connected to the text data processing subunit 322 and is used to normalize the structured data of the case text after data processing and output the case data.

[0097] Specifically, since the format of the case data output by the case generation module 2 is a standardized format, the auxiliary classification module 3 does not need to adjust the case data. The present invention also supports receiving non-standardized case files and converting them into standardized forms through the classification data preprocessing unit 32 to facilitate processing by the auxiliary classification model.

[0098] In a preferred embodiment, the recommendation module 4 generates clinical example information corresponding to the lesion classification result through a machine learning model or a rule engine, and the recommended information includes the clinical example information.

[0099] The recommendation module 4 obtains the patient's historical data and current examination results by connecting with the data platform 1 and the user interface module; specifically, the recommendation module 4 presents the generated recommendation information to the doctor or patient through the user interface module. The doctor can make the final treatment decision based on the recommendation results and his own clinical experience.

[0100] Recommendation module 4 uses machine learning models such as random forests, support vector machines, and neural networks to train personalized treatment recommendation algorithms based on historical case and examination data.

[0101] If the amount of data is large enough, deep learning models33 such as convolutional neural networks and recurrent neural networks can be used to process medical imaging data, clinical texts, and lesion classification results to further improve the recommendation accuracy.

[0102] The present invention establishes a rule engine based on clinical guidelines and expert experience, and gives reasonable treatment reference recommendations based on the patient's condition and classification results.

[0103] In a preferred embodiment, the medical service data also includes the doctor's schedule data and the patient's appointment data sent by the external medical system, and also includes a schedule scheduling module 5 for adjusting the schedule according to the doctor's schedule data and the patient's appointment data. The schedule scheduling module 5 includes:

[0104] A schedule data collection unit 51 is used to collect doctor schedule data and patient appointment data in real time;

[0105] The scheduling unit 52 is connected to the schedule data collection unit 51, and the scheduling unit 52 includes a scheduling model, which adjusts the schedule in real time according to preset constraints and outputs the schedule data;

[0106] The notification unit 53 is connected to the scheduling unit 52 and is used to push and notify the schedule data to the external terminal.

[0107] Specifically, the doctor's schedule data includes the doctor's working hours, specialist information, vacation and free time; the patient's appointment data includes appointment requests, condition descriptions, priority information, etc. The present invention optimizes the doctor's schedule, patient appointments and other aspects through a scheduling algorithm.

[0108] The optimization objectives of the scheduling module include the following aspects:

[0109] Minimize patient waiting time;

[0110] Balance the workload of doctors and balance the workload of doctors;

[0111] Improve the utilization rate of medical resources;

[0112] Consider patient urgency.

[0113] In the selection of scheduling models, the present invention adopts learning algorithms such as reinforcement learning, genetic algorithm and linear programming / integer programming;

[0114] Reinforcement learning can continuously learn and optimize scheduling strategies by interacting with the environment. For example, the system can dynamically adjust decisions based on the doctor's workload, patient appointment status, etc., and gradually optimize the scheduling plan. The reinforcement learning model guides learning through a reward mechanism. For example, patients have shorter waiting times and doctors have a balanced workload, which will receive positive feedback.

[0115] Genetic Algorithm can find the optimal solution among multiple possible scheduling solutions by simulating the natural selection process. Each "individual" represents a scheduling solution. The system generates new scheduling solutions through "crossover" and "mutation" operations, and gradually evolves the optimal scheduling solution.

[0116] Linear / Integer Programming can transform the scheduling problem into a mathematical model, set constraints (such as doctor's free time, patient priority, etc.), and then find the optimal solution through optimization algorithm.

[0117] The scheduling model needs to dynamically adjust the schedule based on real-time data. For example, when a new patient appointment request comes in, it will be dynamically scheduled based on the current doctor availability; when a doctor takes a temporary leave, the scheduling model will quickly adjust and arrange for the patient to change doctors or change dates.

[0118] In a preferred embodiment, it also includes a user interaction module 6, which is connected to the data platform 1 and is used to transmit user information and view recommendation information. The medical service data includes the saved user information, and the user information is distributedly stored in an anonymized and blockchain manner.

[0119] In a preferred embodiment, the data platform 1 is connected to the external medical system via the data interface module 7, and the data interface module 7 integrates a universal calling interface, and receives medical service data and sends advice information via the calling interface.

[0120] Specifically, the data platform 1 aggregates various medical service data into a unified database through the connection with the external medical system interface, such as the hospital information system and the imaging system. The data interconnection between different systems is realized through the data interface module 7. The data platform 1 of the present invention adopts distributed and anonymous storage, and uses blockchain technology to ensure that the data cannot be tampered with and protect data security.

[0121] The present invention develops a RESTful API or GraphQL interface to facilitate each module to call and share data, and uses a stream processing framework such as Apache Kafka and Apache Flink to implement real-time data stream processing to ensure that each module can obtain and process data in real time.

[0122] More specifically, the present invention adopts an event-driven architecture, and each module communicates and collaborates through an event notification mechanism. For example, the auxiliary classification module 3 can send an event notification to the user interaction module 6 when an abnormality is found.

[0123] The present invention defines the key performance indicators (KPIs) of each module, such as response time, processing speed, accuracy, etc. Monitoring tools such as Prometheus and Grafana are used to monitor the performance of each module in real time. The present invention also includes a user feedback module, which receives feedback data from users and doctors as the basis for model improvement, and regularly updates and optimizes the artificial intelligence model based on feedback data and new training data.

[0124] The above description is only a preferred embodiment of the present invention, and does not limit the implementation mode and protection scope of the present invention. For those skilled in the art, it should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the description and illustrations of the present invention should be included in the protection scope of the present invention.

Claims

1. An Internet medical service system based on artificial intelligence, characterized in that: include, Data platform (1), used to integrate and manage medical service data; A case generation module (2) is connected to the data platform (1), obtains the examination data of the medical service data, and generates structured case data according to the examination data; An auxiliary classification module (3) connected to the case generation module (2), wherein the auxiliary classification module (3) is provided with a pre-trained deep learning model (33) for outputting a lesion classification result according to the case data and the examination data; A recommendation module (4) is connected to the auxiliary classification module (3) and is used to output recommendation information based on the lesion classification result and the inspection data.

2. The artificial intelligence-based Internet medical service system according to claim 1, characterized in that: The inspection data includes diagnostic record information, and the case generation module (2) includes: A case collection unit (21), used for collecting the diagnosis record information; A case cleaning unit (22), used for cleaning the diagnosis record information; A natural language processing unit (23), connected to the case cleaning unit (22), for identifying and extracting case features of the diagnostic record information through a natural language processing model; A case generation unit (24) is connected to the natural language processing unit (23) and is used to fill in the case data according to the preset case template based on the case characteristics and output the case data.

3. The artificial intelligence-based Internet medical service system according to claim 2, characterized in that: The natural language processing unit (23) comprises: A model fine-tuning subunit (231), used for adjusting parameters of the natural language processing model according to the diagnostic record information; A named entity recognition model (232), connected to the model fine-tuning subunit (231), for recognizing the named entity features of the diagnostic record information; The relationship extraction model (233) is connected to the named entity recognition model (232) and is used to identify the relationship between the named entity features and output the relationship recognition result of the named entity features.

4. The artificial intelligence-based Internet medical service system according to claim 3 is characterized in that: The model fine-tuning subunit (231) is used to minimize the loss function, which is a cross entropy loss function, expressed as: Among them, y i represents the target value of the i-th sample, p(y i |x) represents the predicted probability distribution of the natural language processing model for the input diagnosis record information, x represents the diagnosis record information, and N represents the number of samples.

5. The artificial intelligence-based Internet medical service system according to claim 1, characterized in that: The inspection data includes image data, and the auxiliary classification module (3) outputs a lesion classification result according to the case data and the image data, including: A classification data collection unit (31), used for collecting the image data and the case data; A classification data preprocessing unit (32), connected to the classification data collecting unit (31), for preprocessing the image data and the case data; The deep learning model (33) is connected to the classification data preprocessing unit (32) and is used to extract text features of the case data and lesion features in the image data, and fuse the text features and the lesion features to output the lesion classification result.

6. The artificial intelligence-based Internet medical service system according to claim 5, characterized in that: The classification data collection unit (31) is also used to collect case texts transmitted by an external system. The classification data preprocessing unit (32) receives the case texts and converts the case texts into the case data through preprocessing. The classification data preprocessing unit (32) includes: An information filling subunit (321) is used to fill in the missing information of the case text to generate a filled case text; A text data processing subunit (322), connected to the information filling subunit (321), is used to perform text processing on the filled case text to generate a data-processed case text; The structured data processing subunit (323) is connected to the text data processing subunit (322) and is used to normalize the structured data of the case text after the data processing and output the case data.

7. The artificial intelligence-based Internet medical service system according to claim 1, characterized in that: The recommendation module (4) generates clinical example information corresponding to the lesion classification result through a machine learning model or a rule engine, and the recommendation information includes the clinical example information.

8. The artificial intelligence-based Internet medical service system according to claim 1, characterized in that: The medical service data also includes doctor schedule data and patient appointment data sent by an external medical system, and also includes a schedule scheduling module (5) for adjusting the schedule according to the doctor schedule data and the patient appointment data. The schedule scheduling module (5) includes: A schedule data collection unit (51), used for collecting the doctor's schedule data and the patient's appointment data in real time; A scheduling unit (52) connected to the schedule data collection unit (51), wherein the scheduling unit (52) includes a scheduling model, wherein the scheduling model adjusts the schedule in real time according to preset constraints and outputs schedule data; A notification unit (53) is connected to the scheduling unit (52) and is used to push and notify the schedule data to an external terminal.

9. The artificial intelligence-based Internet medical service system according to claim 1, characterized in that: It also includes a user interaction module (6) connected to the data platform (1) for transmitting user information and viewing the recommendation information, wherein the medical service data includes the stored user information, and the user information is distributedly stored in an anonymized and blockchain manner.

10. The artificial intelligence-based Internet medical service system according to claim 8, characterized in that: The data platform (1) is connected to the external medical system via a data interface module (7), and the data interface module (7) integrates a universal calling interface, through which the medical service data is received and the recommendation information is sent.

Images