An internet-based health care service system

Abstract

The application relates to an internet-based health care service system, relates to the technical field of health care, acquires user voice data, performs voice recognition in combination with a private semantic library, generates a guide diagnosis information flow, uploads the guide diagnosis information flow to a hybrid cloud architecture network for pathological screening, and acquires a current pathological state, wherein the hybrid cloud architecture network comprises a public cloud architecture and a private cloud architecture; based on the current pathological state, pathological state development prediction is performed on a historical load response level, and if a prediction result exceeds an early warning threshold, continuous alarm is performed, nursing rescue is performed based on alarm signal linkage positioning, and the technical problems of insufficient health care service data processing capability, low pathological screening precision, lack of personalized service and poor nursing rescue linkage are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of health care, and particularly relates to an Internet-based health care service system. BACKGROUND

[0002] With the rapid development and popularization of the Internet, people's health awareness is also increasing. Traditional health care service mode often relies on offline medical institutions and medical staff, and there are problems such as uneven distribution of resources, low service efficiency, and poor patient experience. Especially in the face of chronic disease management and emergency medical assistance, the traditional mode often cannot meet the needs of patients for timely, accurate and personalized health care services. In recent years, with the continuous development and application of cloud computing, big data, artificial intelligence and other technologies, new technical means and solutions have been provided for health care services. By using these advanced technologies, the rapid processing and analysis of medical health data can be realized, and more scientific and accurate basis can be provided for disease prevention, diagnosis and treatment. At the same time, the introduction of Internet technology also breaks the time and space limitations of traditional health care services, so that patients can obtain the required health care services anytime and anywhere. However, the health care service systems on the market at present mostly have problems such as insufficient data processing capacity, low pathological screening accuracy, lack of personalized services, and poor nursing assistance linkage. Therefore, how to fully utilize the Internet technology and medical health data to build an efficient, convenient and personalized health care service system has become a technical problem to be solved. SUMMARY

[0003] The present application provides an Internet-based health care service system to solve the technical problems of insufficient data processing capacity, low pathological screening accuracy, lack of personalized services and poor nursing assistance linkage in the prior art.

[0004] The technical solution of the present application to solve the above technical problems is as follows:

[0005] The present application provides an Internet-based health care service system, and the execution steps include: obtaining user voice data, performing voice recognition in combination with a private semantic library, and generating a guide information stream; uploading the guide information stream to a hybrid cloud architecture network for pathological screening to obtain a current pathological state, wherein the hybrid cloud architecture network includes a public cloud architecture and a private cloud architecture; based on the current pathological state, performing pathological state development prediction on a historical load response level, and if the prediction result exceeds an early warning threshold, performing continuous alarm, and based on the alarm signal linkage positioning for nursing assistance.

[0006] The beneficial effect of the present application is that: through obtaining user voice data to generate a guide diagnosis information flow, then uploading to a hybrid cloud architecture network for pathological screening, then developing prediction based on the pathological state and alarming when necessary, and finally linking positioning for nursing and rescue, the efficiency and accuracy of health care services can be improved, and timely and accurate personalized health care services are realized. BRIEF DESCRIPTION OF DRAWINGS

[0007] Figure 1 An execution step flow diagram of the health care service system based on the Internet is provided.

[0008] Figure 2 An execution step flow diagram of the health care service system based on the Internet is provided. DETAILED DESCRIPTION

[0009] The technical solutions in the embodiments of the present application will be clearly and completely described in the present application 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. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the present application.

[0010] In the description of the present application, the terms "first", "second" are used only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0011] In the description of the present application, the term "for example" is used to indicate "as an example, illustration or explanation". Any embodiment described as "for example" in the present application is not necessarily interpreted as more preferred or more advantageous than other embodiments. The following description is given in order to enable any person skilled in the art to implement and use the present application. In the following description, details are listed for the purpose of explanation. It should be understood that those skilled in the art can realize the present application without using these specific details. In other examples, well-known structures and processes will not be described in detail to avoid unnecessary details making the description of the present application obscure. Therefore, the present application is not intended to be limited to the shown embodiments, but is consistent with the broadest scope in accordance with the principles and characteristics disclosed.

[0012] Embodiment:

[0013] As Figure 1As shown, the embodiment of the present application provides an internet-based health care service system, and the execution steps include:

[0014] S10: Acquire user voice data, combine private semantic library for voice recognition, and generate guide information flow.

[0015] S20: Upload the guide information flow to a hybrid cloud architecture network for pathological screening and acquire current pathological state, wherein the hybrid cloud architecture network includes a public cloud architecture and a private cloud architecture.

[0016] S30: Based on the current pathological state, predict the development of pathological state on the historical load response level, if the prediction result exceeds the warning threshold, perform continuous alarm, and based on the alarm signal linkage positioning, perform nursing rescue.

[0017] For example, the user inputs voice data through a microphone to describe his / her symptoms, and this interaction process triggers the voice recognition function of the system. The system first acquires these voice data and uses the locally trained dialect database (i.e. private semantic library) for voice recognition. This private database is optimized and trained for specific dialects, so it can more accurately understand the user's voice input, especially those dialect words and expressions that may be difficult for general voice recognition systems to understand. Through efficient and accurate semantic recognition combined with the private semantic library, the system can convert the user's voice description into text data and further generate a guide information flow containing preliminary judgments of diseases, related causes, and recommended examination items. This process not only improves the user experience, but also enables the system to provide subsequent health care services more quickly. For example, if the user describes his / her headache and vomiting symptoms in dialect, the system can accurately recognize and generate the corresponding guide information flow, providing strong support for subsequent medical diagnosis. Especially for the elderly with unclear pronunciation, the private semantic library based on targeted and simple training for interactive recognition can increase the recognition efficiency and user experience, and improve their self-care ability.

[0018] Further, after obtaining the user's symptom guidance information flow, the system will upload it to a hybrid cloud architecture network for pathological screening to determine the current pathological state. This hybrid cloud architecture network combines public cloud architecture and private cloud architecture. Public cloud architecture has strong computing power and rich cross-regional and cross-population pathological data resources, and can process large-scale data and give comprehensive pathological analysis; while private cloud architecture provides higher data security and privacy protection, especially suitable for storing and processing sensitive health information of users. The system will intelligently select to upload the guidance information flow to the public cloud architecture or the private cloud architecture for pathological screening according to the data size, urgency and current network load of the guidance information flow. For example, for non-urgent and small data volume guidance information flow, the system may prefer to select the private cloud architecture for processing to utilize the localized pathological recognition model to quickly give a preliminary diagnosis. For urgent or large data volume cases, the system will upload the guidance information flow to the public cloud architecture to utilize its strong computing power for more in-depth analysis, so as to more accurately determine the current pathological state. This flexible selection mechanism ensures that the system can efficiently handle various complex health care needs while ensuring data security.

[0019] Finally, after determining the user's current pathological state, the system will further predict the development of the pathological state. This step is based on the user's historical load response level, i.e. the body's reaction and treatment under similar pathological conditions in the past. The system will collect and analyze these historical data, using the collected historical pathological data as the horizontal coordinate and the pathological state development trend as the vertical coordinate, and perform curve fitting on the pathological state development trend. In this way, a prediction function curve is obtained, which is actually a visual representation of a mathematical model that reflects the disease development law adapted to the user's own condition based on historical data. For example, in the case of diabetes, the curve shows the trend of blood sugar levels over time and with different treatment methods, allowing the system to clearly see the development pattern of the disease under different conditions. However, due to the different adaptability of different individuals, individualized analysis is required when performing curve fitting. Then, with the prediction function curve as a constraint, the possible trend of the user's current state is predicted under the curve level of the original body mechanism response fluctuation, and the prediction result is obtained. If the prediction result shows that the user's pathological state may worsen or exceed the pre-set warning threshold, the system will immediately start a continuous alarm mechanism. This alarm mechanism sends an emergency signal to the user through various means (such as sound, vibration, pop-up window, etc.), reminding the user to pay attention and take appropriate measures. At the same time, the system will automatically match nearby medical institutions based on the user's geographic location information, and recommend the most suitable medical institution for the user based on the current pathological state development prediction result. For example, if the user is diagnosed with subsequent shock or the body enters an emergency stage, and the system predicts that the body may rapidly deteriorate in a short period of time, the system will immediately start the alarm and recommend a nearby hospital with strong emergency department to the user. At the same time, the system will automatically generate navigation path information to help the user quickly reach the hospital, and send a request for nursing assistance to the hospital in advance if necessary, so that the hospital can prepare for admission. In this way, the user can obtain timely and effective medical assistance in a short period of time, thereby greatly improving the effectiveness of assistance and survival rate.

[0020] In a preferred embodiment, as shown in Figure 2 obtaining user voice data, performing voice recognition in combination with a private semantic library, and generating a guide diagnosis information stream, including: collecting voice data through a user terminal device, preprocessing the collected user voice data to obtain converted voice data; inputting the converted voice data into a semantic recognition model, combining a pre-constructed private semantic library, performing semantic recognition and conversion on the converted voice data to obtain semantic request information; matching a symptom list based on the semantic request information, and generating a guide diagnosis information stream according to the matching result, wherein the guide diagnosis information stream includes a preliminary judgment of a disease, a related cause, and a recommended examination item.

[0021] Alternatively, when a user needs health care services, they can interact through their own terminal device (such as a smartphone, tablet, or smart speaker) by voice. The system first starts the voice data collection function, accurately capturing the user's description of the symptoms through the device's microphone. In order to ensure the accuracy and clarity of the voice data, the system will preprocess the raw voice data collected, including noise reduction, noise removal, and voice signal enhancement operations, to obtain the voice data to be converted. Then, these voice data to be converted are input into the semantic recognition model. This model is trained through deep learning and can efficiently convert voice into text. At the same time, in order to improve the accuracy and adaptability of recognition, the system also combines a pre-constructed private semantic library. This private semantic library is pre-trained through the user's local language, such as making the local dialect easier to be recognized after communicating through guided dialogue in the early stage, and its training contains a large number of medical health-related vocabulary and local expression methods, especially for common diseases, causes and examination items. By combining the private semantic library, the semantic recognition model can more accurately understand the user's voice description and convert it into detailed text data, i.e. semantic request information. With the semantic request information, the system will link a symptom list corresponding to its own record information according to the user's care record and information to match. This symptom list contains the typical symptoms of various diseases, possible causes, and recommended examination items. By matching the user's semantic request information with the symptom list, the system can quickly generate a guide information stream. This guide information stream not only includes the preliminary judgment of the disease, but also involves possible causes and recommended examination items for diagnosis. For example, if the user describes through voice during the postoperative care stage: "I have been feeling a headache and a little nauseated recently." After voice data collection, preprocessing, semantic recognition, and symptom list matching, the system will generate a guide information stream like this: "Preliminary judgment: postoperative concurrent headache with nausea; Possible causes: migraine, intracranial hypertension, etc.; Recommended examination items: head CT, electroencephalogram, etc." In this way, the user can have a preliminary understanding of their own health status according to the guide information stream, and further interact or decide whether to need further medical examination.

[0022] In a preferred embodiment, the guide information stream is uploaded to a hybrid cloud architecture network for pathological screening to obtain the current pathological state, wherein the hybrid cloud architecture network includes a public cloud architecture and a private cloud architecture, comprising: encrypting the guide information stream, dynamically classifying according to a preset shunting strategy, and uploading to the hybrid cloud architecture network to complete pathological identification screening to obtain the identification result; the identification result is taken as the current pathological state.

[0023] Further, after the user generates a guidance information stream containing a preliminary diagnosis of the illness, related causes, and recommended examination items through voice interaction, the system will immediately upload this guidance information stream to a high-efficiency and secure hybrid cloud architecture network for pathological screening. To ensure the security and privacy of user data, the system will use a hash function to encrypt the guidance information stream before uploading. This encryption technology can ensure the confidentiality and integrity of data during transmission, preventing unauthorized access or tampering. Subsequently, the guidance information stream is dynamically classified according to a pre-set shunting strategy. This shunting strategy takes into account multiple factors, including the data size of the guidance information stream, the urgency level, and the current load of the hybrid cloud architecture network. Through comprehensive evaluation of these factors, the system can intelligently decide whether to upload the guidance information stream to a public cloud architecture or a private cloud architecture for processing. For example, if the guidance information stream contains a description of an urgent and serious illness, the system may prefer to upload it to a public cloud architecture. This is because public cloud architectures usually have more powerful computing capabilities and richer medical resource databases, allowing for faster pathological recognition screening and more accurate recognition results. Conversely, if the guidance information stream describes a relatively minor or common illness, and the current load of the public cloud architecture is high, the system may choose to upload it to a private cloud architecture for processing to reduce the burden on the public cloud architecture and improve processing efficiency. Once the guidance information stream is uploaded to the corresponding cloud architecture, the system will immediately start the pathological recognition screening program. This program will use advanced algorithms and models to analyze the guidance information stream in depth and compare and match it with information in the medical resource database. Finally, the system will give an identification result containing the current pathological state. This result not only helps users better understand their health status, but also provides strong support for subsequent medical services.

[0024] In a preferred embodiment, the pre-set shunting strategy for dynamic classification includes: dynamically selecting to upload the guidance information stream to a public cloud architecture or a private cloud architecture according to the data size, urgency level, and current network load of the guidance information stream; when the data size, urgency level, and current network load are all below their respective thresholds, uploading the guidance information stream to a private cloud architecture for pathological recognition screening; when any of the data size, urgency level, and current network load is above its threshold, uploading the guidance information stream to a public cloud architecture for pathological recognition screening.

[0025] Specifically, the preset shunting strategy considers the data size, urgency, and current network load of the triage information flow when performing dynamic classification. Specifically, the system evaluates these three factors in real-time and dynamically determines the upload destination of the triage information flow based on their comparison with respective preset thresholds. If the data size of the triage information flow is not large, the urgency is not high, and the current network load is also at a low level, i.e., all three factors are below their respective thresholds, the system will choose to upload the triage information flow to the private cloud architecture for pathological recognition screening. This is because the private cloud architecture can usually provide faster and more personalized services while ensuring data security and privacy when handling small-scale, non-urgent, and lightly loaded tasks. However, if the data size of the triage information flow is large, the urgency is high, or the current network load is heavy, i.e., any of the three factors is higher than its threshold, the system will choose to upload the triage information flow to the public cloud architecture for pathological recognition screening. The public cloud architecture has strong computing power and rich resources, which can handle large-scale and high-urgency tasks and maintain stable performance when the network load is heavy, thereby ensuring the accuracy and efficiency of pathological recognition screening. For example, during a busy medical consultation period, if a large number of users upload triage information flows simultaneously, causing a sharp increase in network load, the system will automatically upload some or all of the triage information flows to the public cloud architecture for processing according to the preset shunting strategy, to avoid overloading the private cloud architecture and affecting service quality. Similarly, if a user uploads a triage information flow containing a large amount of data or describing an urgent medical condition, the system will also preferentially choose the public cloud architecture for fast processing.

[0026] In a preferred embodiment, when uploaded to the private cloud architecture for pathological recognition screening, it includes: combining the private cloud architecture, performing localized processing on the triage information flow to obtain an identification result; wherein the private cloud architecture stores local user data, and based on the local user data, the triage information flow is verified and analyzed to obtain a pathological feature vector; input the pathological feature vector into a pre-trained local pathological recognition model, the local pathological recognition model is trained based on local user pathological case data, and output a first pathological recognition result; quality assessment is performed on the first pathological recognition result, if the assessment result meets the preset standard, the identification result is taken as the final identification result; if the assessment result does not meet the preset standard, the triage information flow is re-uploaded to the public cloud architecture for secondary recognition screening.

[0027] Preferably, when the triage information stream is uploaded to the private cloud architecture for pathological identification screening, the system will make full use of the local user data stored on the private cloud architecture to perform localization processing on the triage information stream. First, the system will perform verification analysis on the triage information stream based on these local user data, extracting key pathological feature vectors that can accurately reflect the characteristics of the disease described in the triage information stream, the cause and possible examination items, etc. Next, the system will input the extracted pathological feature vectors into the pre-trained local pathological identification model. This local pathological identification model is trained based on a large number of local user pathological case data, so it can accurately identify specific pathological features of local users. Specifically, a neural network structure based on deep learning is used to build the local pathological identification model, which includes an input layer, a hidden layer, and an output layer. The input layer receives the pathological feature vectors extracted from the processed triage information stream; the hidden layer is composed of multiple neurons and performs deep feature extraction and nonlinear transformation on the input data through complex weight connections; the output layer outputs the first pathological identification result based on the processing result of the hidden layer, which contains the system's preliminary judgment of the disease, possible causes, and recommended examination items, etc. In the specific training process, a large number of local user pathological cases are collected, covering various disease types, symptom manifestations, examination results, and diagnosis and treatment information, etc. These data are cleaned and preprocessed to remove errors or incomplete data, and are divided into training set, validation set and test set according to a certain proportion. In the training phase, the pathological feature vectors in the training set are input into the model, and the model gradually approaches the true diagnosis result by continuously adjusting the weights between the hidden layer neurons. In the training process, the loss function is used to measure the difference between the model output and the true result, and the optimization algorithm is used to minimize the loss function, thereby continuously optimizing the performance of the model. During the training process, the validation set is used to periodically evaluate the model and monitor the accuracy, recall rate and other indicators of the model to prevent overfitting. When the performance of the model on the validation set no longer improves, it is considered that the model training has reached a good state. Finally, the test set is used to comprehensively evaluate the trained model to ensure that the model has good generalization ability on unseen data and can accurately identify various pathological conditions.

[0028] However, to further ensure the accuracy and reliability of the identification result, the system will also perform a quality assessment on the first pathology identification result. The assessment process will consider multiple factors, such as the confidence level of the identification result, the degree of matching with local user data, etc. If the assessment result meets the preset standard, the system will take the identification result as the final identification result and provide the user with the corresponding health care service accordingly. However, if the assessment result does not meet the preset standard, it means that the first pathology identification result may have errors or uncertainties. In this case, the system will re-upload the triage information stream to the public cloud architecture for secondary identification screening. The public cloud architecture can perform more in-depth and comprehensive analysis on the triage information stream, thus giving a more accurate identification result. For example, suppose a user uploads a triage information stream describing a rare disease, and the local pathology recognition model on the private cloud architecture has limited recognition ability for this disease. In the initial identification, the model may give an inaccurate identification result. However, through quality assessment, the system finds that the confidence level of this result is low, and the degree of matching with local user data is also not high. Therefore, the system will re-upload the triage information stream to the public cloud architecture for secondary identification screening. The more advanced pathology recognition model on the public cloud architecture can accurately identify this rare disease and give the corresponding treatment recommendations, thus providing more accurate health care services for the user.

[0029] In a preferred embodiment, when uploading to the public cloud architecture for pathology identification screening, it includes: uploading the triage information stream to the public cloud architecture, standardizing the triage information stream to meet the input requirements of the universal pathology recognition model on the public cloud architecture; inputting the standardized triage information stream into the universal pathology recognition model on the public cloud architecture, the universal pathology recognition model is trained based on large-scale cross-regional and cross-population pathology data, and outputs the second pathology identification result; performing a security review on the second pathology identification result returned by the public cloud architecture, and taking the identification result as the current pathology state after the review is passed.

[0030] In detail, when the triage information stream is uploaded to the public cloud architecture for pathological recognition screening, it will first undergo a standardization process. Since the general pathological recognition model on the public cloud architecture needs to process pathological data from different regions and different populations, the triage information stream must meet the input requirements of the model. The system will perform format conversion, data cleaning, and other operations on the triage information stream to ensure that it can be accurately recognized and processed by the general pathological recognition model. Next, the standardized triage information stream will be input into the general pathological recognition model on the public cloud architecture. This model is trained based on large-scale cross-regional and cross-population pathological data, so it has very high accuracy and generalization ability. It can quickly analyze the symptom description, cause, and recommended examination items in the triage information stream and output a second pathological recognition result. This result usually includes detailed classification of the symptoms, possible cause analysis, and recommended treatment plan for the symptoms. However, before the second pathological recognition result is used as the current pathological state, the system will also conduct a security review. This is because the public cloud architecture, although it has strong computing power and rich resources, also has security risks such as data leakage and tampering. The system will check the integrity of the recognition result, the reliability of the source, and whether there is abnormal data to ensure its authenticity and accuracy. Only after the security review and confirmation of no errors, the system will use the recognition result as the current pathological state and provide further health care services for the user based on it. For example, suppose a user uploads a triage information stream describing a common cold symptom, but the symptoms are more special. The general pathological recognition model on the public cloud architecture will quickly analyze the symptom description and cause in the standardized triage information stream and output a second pathological recognition result, indicating that the user may have a special type of cold and giving corresponding treatment recommendations. Before returning this result to the user, the system will conduct a security review to confirm that the recognition result has not been tampered with or contains abnormal data, and then use it as the current pathological state and inform the user of the specific symptom and treatment plan.

[0031] In a preferred embodiment, based on the current pathological state, the pathological state development prediction is made on the historical load response level, and if the prediction result exceeds the early warning threshold, a persistent alarm is given, comprising: collecting historical pathological data related to the current pathological state, including pathological index changes and treatment effect data of the same disease at different time periods; taking the historical pathological data of the same disease at different time periods as the horizontal coordinates and the pathological state development trend as the vertical coordinates, curve fitting is performed on the pathological state development trend to obtain a prediction function curve graph; taking the prediction function curve graph as a constraint, the result of the current pathological state is predicted to obtain the pathological state development prediction result in a preset time period; setting an early warning threshold, when the prediction result shows that the pathological state development exceeds the early warning threshold, starting a persistent alarm mechanism, and sending an alarm signal to the user through the user terminal device in the form of sound, vibration and pop-up window.

[0032] For example, based on the current determined pathological state, the system can further make a pathological state development prediction to assess the possible trend of the disease and take timely measures. First, the system can collect the user's historical pathological data closely related to the current pathological state, which covers the changes in pathological indicators of users with the same disease at different time periods and the corresponding treatment effects. For example, for diabetes, the system can collect the user's past blood glucose level fluctuations, glycosylated hemoglobin value changes, and blood glucose control effects under different treatment plans. Next, using these historical pathological data as the horizontal coordinate and the pathological state development trend as the vertical coordinate, the system can use mathematical algorithms to process these data, curve fit the pathological state development trend, and generate a prediction function curve graph that can reflect the development law of the disease. This curve graph acts as a prediction model, and by substituting the current pathological state data into it, the system can obtain the pathological state development prediction results for a preset time period, which refers to the possible trend of the current state under the original body mechanism response fluctuation curve. For example, for a patient with hypertension, the system can collect the patient's blood pressure data over the past few months or even years, including systolic and diastolic pressure values at different time points and blood pressure changes under different antihypertensive drugs or lifestyle interventions, which are all historical pathological data. Using the time sequence of blood pressure measurement values and the corresponding treatment conditions as the horizontal coordinate and the blood pressure control conditions, blood pressure fluctuation trends, and other pathological state development trends as the vertical coordinate, the system can draw a curve reflecting the patient's blood pressure change law, i.e., a prediction function curve graph, through a suitable curve fitting algorithm. If the patient's blood pressure is at a certain level, substituting this current state into the prediction function curve graph can predict how the user's blood pressure will change in the future, i.e., the trend. To timely detect and respond to possible changes in the disease, the system can set a warning threshold, which is equivalent to virtually drawing two warning curves above and below the user's corresponding pathological state load fitting curve based on the warning threshold. When the prediction result shows that the change curve of the pathological state may exceed this warning threshold, i.e., when the curves intersect, the system can immediately start a continuous alarm mechanism. This alarm mechanism can send alarm signals to the user through the user terminal device in various ways such as sound, vibration, pop-up window, etc., to ensure that the user can receive the notification in time and take appropriate action. For example, if the prediction result shows that the blood glucose level of a diabetic patient may rise sharply in the next few days, exceeding the safe range, the system can send a loud alarm through the user's mobile phone and pop up an emergency notification window, reminding the patient to seek medical treatment or adjust the treatment plan to cope with the possible situation.

[0033] In a preferred embodiment, the alarm signal linkage positioning-based nursing assistance includes: after the persistent alarm mechanism is started, the geographic position information of the user terminal is automatically acquired, and the geographic position information is matched with a preset medical institution database, wherein the medical institution database stores medical institution information around the geographic position information of the user terminal within a preset range; according to the matching result, a first medical institution is recommended in combination with a current pathological state development prediction result, and navigation path information is generated; the navigation path information and the alarm signal are pushed to the user through the user terminal device, and at the same time, a nursing assistance request is sent to the first medical institution; after the first medical institution receives the nursing assistance request, whether to accept is decided according to a real-time load condition of the first medical institution, and a feedback processing result is generated; if the first medical institution accepts the request, the navigation path information is generated and shared; if the first medical institution cannot accept the request, a medical institution is recommended for the user again until a response processing is performed.

[0034] Specifically, after the persistent alarm mechanism is started, the system rapidly and automatically acquires the geographic position information of the user terminal, which is crucial for subsequent nursing assistance. The system accurately matches the geographic position information with a preset medical institution database. The medical institution database stores detailed medical institution information around the geographic position information of the user terminal within a preset range, including hospitals, clinics, emergency centers, and key data such as their service range, professional characteristics, and real-time load condition. According to the matching result, the system intelligently recommends the most suitable first medical institution in combination with the current pathological state development prediction result. For example, if the user is predicted to have acute myocardial infarction and the condition develops rapidly, the system will preferentially recommend a nearby third-grade hospital with heart intervention surgery capability. At the same time, the system generates detailed navigation path information to ensure that the user can quickly and accurately reach the recommended medical institution. Next, the system pushes the navigation path information and the alarm signal to the user through the user terminal device to ensure that the user can first understand their condition and the medical institution to go to. At the same time, the system sends a nursing assistance request to the first medical institution, which contains key information such as the user's pathological state, prediction result, and geographic position, so that the medical institution can prepare for reception in advance. After receiving the nursing assistance request, the first medical institution quickly decides whether to accept according to its real-time load condition. If it accepts the request, the medical institution shares the navigation path information to ensure that the user can quickly reach it according to the optimal route. If the first medical institution cannot accept the request due to heavy load or other reasons, the system will recommend other suitable medical institutions for the user again until it finds a medical institution that can respond to the processing.

[0035] The health care service system based on the Internet provided by the embodiment of the application has at least the following technical effects:

[0036] 1. By combining private semantic libraries for speech recognition, the user's voice describing the symptoms can be accurately understood, and the guidance information stream containing the preliminary judgment of the disease, the related causes and the recommended examination items can be generated. Further, the pathological screening is carried out by using a hybrid cloud architecture network, and the processing architecture is dynamically selected according to the data volume, the urgency and the network load of the guidance information stream, which not only ensures the efficiency of data processing, but also ensures the accuracy of recognition, effectively improves the accuracy and response speed of pathological recognition, and provides more timely and reliable health consultation services for users.

[0037] 2. The current pathological state is combined to predict on the historical load response level, and a warning threshold is set. Once the prediction result shows that the pathological state may exceed the warning threshold, the system will immediately start a persistent alarm mechanism, and send alarm signals to the user in multiple ways through the user terminal device. This intelligent prediction and alarm mechanism helps users to understand the development trend of their own illness in a timely manner, take necessary preventive measures or medical action, and thus effectively avoid illness deterioration.

[0038] 3. When the persistent alarm mechanism is started, the geographic location information of the user terminal can be automatically obtained, and matched with the preset medical institution database to quickly recommend suitable medical institutions and generate navigation path information. At the same time, the system will also send a nursing rescue request to the medical institution, realizing the rapid linkage between the user and the medical institution. After receiving the request, the medical institution will decide whether to accept according to its real-time load situation, and generate a feedback processing result. This nursing rescue linkage mechanism greatly shortens the time from the discovery of the illness to the obtaining of professional medical rescue, improves the efficiency and success rate of nursing rescue, and provides more convenient and efficient health care services for users.

[0039] The above description of disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An internet-based health care service system, characterized in that, The execution steps include: Acquire user voice data, combine it with a private semantic library for speech recognition, and generate a triage information stream; The triage information flow is dynamically classified according to a preset triage strategy and then uploaded to a hybrid cloud architecture network for pathological screening to obtain the current pathological status. The hybrid cloud architecture network includes a public cloud architecture and a private cloud architecture. Based on the current pathological state, the pathological state development is predicted at the historical load response level. If the prediction result exceeds the warning threshold, a continuous alarm is triggered, and nursing care is provided based on the alarm signal linkage positioning. The preset traffic splitting strategy performs dynamic classification, and the execution steps include: Based on the data volume, urgency, and current network load of the triage information stream, dynamically select whether to upload the triage information stream to a public cloud architecture or a private cloud architecture; When the data volume, urgency level, and current network load are all below their respective thresholds, the triage information stream is uploaded to a private cloud architecture for pathological identification and screening. If any of the data volume, urgency level, or current network load exceeds its threshold, the triage information stream will be uploaded to the public cloud architecture for pathological identification and screening. When uploading to a private cloud architecture for pathological identification and screening, the following steps are performed: By combining a private cloud architecture, the triage information flow is localized to obtain recognition results; The private cloud architecture stores local user data, and the triage information flow is verified and analyzed based on the local user data to obtain pathological feature vectors. The pathological feature vector is input into a pre-trained local pathological recognition model, which is trained based on local user pathological case data, and outputs a first pathological recognition result. The quality of the first pathological identification result is evaluated. If the evaluation result meets the preset standard, the identification result is taken as the final identification result. If the evaluation results do not meet the preset standards, the triage information flow will be re-uploaded to the public cloud architecture for secondary identification and screening; the public cloud architecture is equipped with a general pathology identification model, which is trained based on large-scale cross-regional and cross-population pathology data; Nursing and rescue operations based on alarm signal-linked positioning include the following steps: When the continuous alarm mechanism is activated, the geographical location information of the user terminal is automatically obtained and matched with a preset medical institution database. The medical institution database stores information on medical institutions around the geographical location information of the user terminal within a preset range. Based on the matching results and the prediction of the current pathological condition, the primary medical institution is recommended, and navigation path information is generated. The navigation path information and alarm signals are pushed to the user through the user terminal device, and at the same time, a nursing assistance request is sent to the first medical institution. After receiving a request for nursing assistance, the first medical institution decides whether to accept it based on its real-time workload and generates a feedback processing result. If the first medical institution accepts the request, it generates and shares the navigation path information; if the first medical institution cannot accept the request, it recommends medical institutions to the user again until a response is initiated.

2. The Internet-based health care service system as described in claim 1, characterized in that, The process involves acquiring user voice data, performing speech recognition using a private semantic library, and generating a triage information stream. The steps include: Voice data is collected through user terminal devices, and the collected user voice data is preprocessed to obtain voice data to be converted. The speech data to be converted is input into the semantic recognition model, and combined with the pre-built private semantic library, the speech data to be converted is semantically recognized and converted into text data to obtain semantic request information; Based on the semantic request information, a symptom list is matched, and a triage information stream is generated according to the matching results. The triage information stream includes a preliminary diagnosis of the disease, relevant causes, and suggested examination items.

3. The Internet-based health care service system as described in claim 1, characterized in that, The triage information stream is uploaded to a hybrid cloud architecture network for pathological screening to obtain the current pathological status. The hybrid cloud architecture network includes both public and private cloud architectures. The execution steps include: The triage information stream is encrypted, dynamically classified according to a preset triage strategy, and then uploaded to a hybrid cloud architecture network to complete pathological identification and screening, and obtain identification results. The identification result is taken as the current pathological state.

4. The Internet-based health care service system as described in claim 3, characterized in that, When uploading to a public cloud architecture for pathological identification and screening, the following steps are performed: The triage information flow is uploaded to a public cloud architecture, and the triage information flow is standardized to meet the input requirements of a general pathological recognition model on the public cloud architecture. The standardized triage information flow is input into a general pathology recognition model on a public cloud architecture. The general pathology recognition model is trained based on large-scale cross-regional and cross-population pathology data and outputs a second pathology recognition result. The second pathological identification result returned by the public cloud is subject to security review. Once the review is passed, the identification result is used as the current pathological status.

5. The Internet-based health care service system as described in claim 1, characterized in that, Based on the current pathological state, a pathological state development prediction is made at the historical load response level. If the prediction result exceeds the warning threshold, a continuous alarm is triggered. The execution steps include: Collect historical pathological data related to the current pathological state, including changes in pathological indicators and treatment effects of the same disease at different time periods; Using the historical pathological data of the same disease at different time periods as the horizontal axis and the pathological state development trend as the vertical axis, a curve fitting is performed on the pathological state development trend to obtain a prediction function curve. Using the prediction function curve as a constraint, the current pathological state is predicted to obtain the pathological state development prediction result within a preset time period. Set an early warning threshold. When the prediction results show that the pathological state has developed beyond the early warning threshold, activate the continuous alarm mechanism and send an alarm signal to the user through the user terminal device via sound, vibration or pop-up window.

Images