Intelligent vital sign health analysis system based on professional medical big language model

Abstract

The invention relates to the technical field of medical health, and discloses a vital sign health intelligent analysis system based on a professional medical big language model, which comprises a data acquisition module, a data processing module, a big language model analysis module and a user interaction interface, the data processing module is connected with the large language model analysis module, and the large language model analysis module is connected with the user interaction interface. By using an advanced data acquisition technology (such as a non-intrusive sensor) and a deep learning model (such as BERT, GPT and LSTM), the physiological data of the user can be monitored and analyzed in real time, the health trend and potential problems can be identified, the Bayesian fusion algorithm is combined to perform weighted fusion on the physiological data and the historical data of the user, and the user experience is improved. The system can provide more personalized and accurate health suggestions; therefore, not only is the accuracy of health monitoring improved, but also the requirements of different users are better met through personalized analysis.

Description

Technical Field

[0001] The present invention relates to the field of medical health technology, and in particular to a vital sign health intelligent analysis system based on a professional medical large language model. Background Art

[0002] With the rapid development of artificial intelligence technology, large language models have become one of the core technologies in the field of natural language processing. By learning a large amount of text data, large language models can understand and generate human language and provide support for various application scenarios, including machine translation, content creation, and dialogue systems. However, in the field of vital sign health analysis, the application of large language models is still in its infancy.

[0003] Traditional vital signs monitoring systems rely on sensor technology and data analysis algorithms to collect and process physiological signals such as heart rate, blood pressure, blood oxygen saturation, etc. These systems usually focus on accurate data collection and real-time monitoring, but have limitations in health analysis and prediction. They often lack a deep understanding of individual health data and cannot provide personalized health advice or predict potential health risks. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the shortcomings of the prior art, the present invention provides a vital sign health intelligent analysis system based on a professional medical big language model, which has the advantages of not only being able to monitor the user's vital sign data in real time, but also being able to analyze the data through a deep learning algorithm, thereby providing more accurate health status assessment and personalized recommendations, etc., solving the problems raised in the background technology.

[0006] (II) Technical solution

[0007] In order to achieve the above purpose of not only being able to monitor the user's vital signs data in real time, but also being able to analyze the data through a deep learning algorithm, thereby providing a more accurate health status assessment and personalized suggestions, the present invention provides the following technical solutions: a vital signs health intelligent analysis system based on a professional medical large language model, comprising a data acquisition module, a data processing module, a large language model analysis module and a user interaction interface, wherein the data acquisition module is connected to the data processing module, the data processing module is connected to the large language model analysis module, and the large language model analysis module is connected to the user interaction interface;

[0008] A data collection module for collecting physiological data of the user, including but not limited to heart rate, heart and lung sounds, blood pressure and body temperature;

[0009] The data processing module is used to pre-process the collected physiological data, including data cleaning and standardization, and use signal processing algorithms to remove noise and convert the data into a unified format;

[0010] The large language model analysis module uses a well-trained large language model to conduct in-depth analysis of pre-processed data to identify health trends and potential problems. The large language model is based on the Transformer architecture and combines physiological data and user historical data for multimodal analysis;

[0011] The user interaction interface is used to provide users with visual health reports and personalized health recommendations. The user interaction interface is implemented through a mobile application or a web application, and provides multiple forms of interaction such as charts, text descriptions, and voice feedback.

[0012] Preferably, the data acquisition module includes at least one non-invasive sensor for real-time monitoring of heart rate and cardiopulmonary sounds without interfering with the user's normal life.

[0013] Preferably, the data processing module receives the information collected by the data acquisition module, and uses a wavelet transform algorithm to remove noise from the data. The algorithm formula is as follows:

[0014]

[0015] Among them, W f (a,b) are wavelet transform coefficients, f(t) is the original signal, It is the value of the mother wavelet function after scaling and translation, a is the scale parameter that controls the scaling of the wavelet, b is the translation parameter that controls the position of the wavelet on the time axis, and dt represents the small time interval when integrating time t.

[0016] Preferably, the data processing module further includes detecting outliers, specifically using a Z-score method to detect outliers, and the calculation formula is as follows:

[0017]

[0018] Where Z is the calculated Z-score value, X is a data point in the data set, μ is the mean of the data set, and σ is the standard deviation of the data set.

[0019] Preferably, the large language model adopts a pre-trained BERT model or a GPT model, and the large language model analysis module also includes using a long short-term memory network to perform predictive analysis of time series data for identifying health trends.

[0020] Preferably, the Transformer architecture comprises at least 12 encoder layers.

[0021] Preferably, when performing multimodal analysis, the large language model analysis module uses a Bayesian fusion algorithm to perform weighted fusion on physiological data and user history data, and the algorithm formula is as follows:

[0022] P(yx)=k·P(yx p ) wp ·P(yx h ) wh

[0023] Among them, P(yx) is the predicted probability after fusion, P(yx p ) and P(yx h ) are the predicted probabilities based on physiological data and historical data, wp and wh are their weights, and k is the normalization factor.

[0024] Preferably, the user interaction interface provides a historical record query function for user health data.

[0025] Preferably, the system further comprises a security module, which is connected to the user interaction interface and is used for encrypting data transmission and storage.

[0026] Preferably, the user interaction interface also includes a feedback module, which allows the user to evaluate the provided health advice.

[0027] Compared with the prior art, the present invention provides a vital sign health intelligent analysis system based on a professional medical large language model, which has the following beneficial effects:

[0028] 1. The present invention provides a vital sign health intelligent analysis system based on a professional medical large language model. By using advanced data acquisition technology (such as non-invasive sensors) and deep learning models (such as BERT, GPT and LSTM), it can monitor and analyze the user's physiological data in real time, identify health trends and potential problems, and combine the Bayesian fusion algorithm to perform weighted fusion of physiological data and user historical data. The system can provide more personalized and accurate health recommendations; this not only improves the accuracy of health monitoring, but also better meets the needs of different users through personalized analysis.

[0029] 2. The present invention provides a vital sign health intelligent analysis system based on a professional medical large language model. The security module is set to ensure the security and privacy of user data during transmission and storage. Through encryption technology, the system effectively prevents data leakage and unauthorized access, and complies with relevant health information protection laws and regulations. This not only enhances the user's trust in the system, but also provides users with safe and reliable health management services. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a schematic diagram of the architecture of the system of the present invention;

[0031] Figure 2 Schematic diagram of the logic of the system of the present invention. DETAILED DESCRIPTION

[0032] 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.

[0033] See also Figure 1-Figure 2 , a vital sign health intelligent analysis system based on a professional medical big language model, comprising a data acquisition module, a data processing module, a big language model analysis module and a user interaction interface, the data acquisition module is connected to the data processing module, the data processing module is connected to the big language model analysis module, and the big language model analysis module is connected to the user interaction interface;

[0034] A data collection module for collecting physiological data of the user, including but not limited to heart rate, heart and lung sounds, blood pressure and body temperature;

[0035] The data processing module is used to pre-process the collected physiological data, including data cleaning and standardization, and use signal processing algorithms to remove noise and convert the data into a unified format;

[0036] The large language model analysis module uses a well-trained large language model to conduct in-depth analysis of pre-processed data to identify health trends and potential problems. The large language model is based on the Transformer architecture and combines physiological data and user historical data for multimodal analysis;

[0037] The user interaction interface is used to provide users with visual health reports and personalized health recommendations. The user interaction interface is implemented through mobile applications or web applications, and provides multiple forms of interaction such as charts, text descriptions and voice feedback.

[0038] Specifically, the data acquisition module includes at least one non-invasive sensor for real-time monitoring of heart rate and cardiopulmonary sounds without interfering with the user's normal life. The advantage is that non-invasive sensors do not need to be in direct contact with the user's skin or body, so the user can hardly feel their presence in daily life, which improves the user's comfort and convenience; by real-time monitoring of heart rate and cardiopulmonary sounds, the system can instantly capture the user's physiological changes, which is conducive to timely detection of health problems and health trend analysis; real-time monitoring can provide a continuous data stream, which helps to improve the accuracy and reliability of analysis, because data collection will not be restricted or interrupted by user activities.

[0039] Specifically, the data processing module receives the information collected by the data acquisition module, and uses the wavelet transform algorithm to remove noise from the data. The algorithm formula is as follows:

[0040]

[0041] Among them, W f (a,b) are wavelet transform coefficients, f(t) is the original signal, It is the value of the mother wavelet function after scaling and translation, a is the scale parameter, which controls the scaling of the wavelet, b is the translation parameter, which controls the position of the wavelet on the time axis, and dt represents the small time interval when integrating time t. The advantage is that the wavelet transform algorithm can decompose the signal collected by the data acquisition module into different frequency levels, so as to effectively identify and remove the noise components and retain the useful signal, which is conducive to improving the accuracy of physiological data analysis; the wavelet transform has a multi-scale characteristic and can analyze the signal in the time and frequency domain at the same time, which means that it can identify the transient characteristics and trends in the signal, which is very useful for understanding complex vital signs data; through the wavelet transform, the data can be effectively compressed, redundant information can be removed, and only the data part that has an important impact on health analysis can be retained, which helps to reduce storage requirements and improve system processing speed; after removing the noise, the signal-to-noise ratio of the data is improved, making the subsequent health trend analysis and potential problem identification more accurate.

[0042] Specifically, the data processing module also includes detecting outliers, specifically using the Z-score method to detect outliers, and the calculation formula is as follows:

[0043]

[0044] Among them, Z is the calculated Z-score value, X is a data point in the data set, μ is the mean of the data set, and σ is the standard deviation of the data set. The advantage is that the system calculates the Z-score value of the data point through a formula, and can identify those points that are far from the mean of the data set. These points may be caused by measurement errors, data entry errors, or other abnormal situations. Identifying and eliminating them can improve the overall quality of the data set and ensure that subsequent analysis is based on more accurate and reliable data; the Z-score method can simplify the process of data cleaning and preprocessing, making data processing more efficient.

[0045] Specifically, the large language model uses a pre-trained BERT model or GPT model, and the large language model analysis module also includes the use of long short-term memory networks for predictive analysis of time series data to identify health trends. The advantage is that BERT and GPT models, as pre-trained large language models, can deeply understand natural language and complex text data, which is very useful when processing and analyzing users' health data and historical records. Combined with long short-term memory networks, the system can process and predict time series data, which is conducive to analyzing the trend of vital sign data over time; because long short-term memory networks are good at capturing long-term dependencies, they can more accurately predict health trends and potential problems; large language models can combine physiological data and user historical data for multimodal analysis, which can more comprehensively evaluate the user's health status and provide more accurate health trend predictions.

[0046] Specifically, the Transformer architecture contains at least 12 encoder layers. The advantage is that more encoder layers mean that the large language model can learn deeper feature representations, which helps capture complex patterns and relationships in the data when processing complex physiological data and user history data; the deep Transformer architecture can usually provide higher accuracy because it can better understand the long-term dependencies and contextual information in the data. The Transformer architecture with at least 12 encoder layers can adapt to a variety of input data of different lengths and complexities, which enables the system to process multimodal data and time series data.

[0047] Specifically, when performing multimodal analysis, the large language model analysis module uses the Bayesian fusion algorithm to perform weighted fusion of physiological data and user history data. The algorithm formula is as follows:

[0048] P(yx)=k·P(yx p ) wp ·P(yx h ) wh

[0049] Among them, P(yx) is the predicted probability after fusion, P(yxp ) and P(yx h ) are the prediction probabilities based on physiological data and historical data, respectively, wp and wh are their weights, and k is the normalization factor. The advantage is that by assigning different weights to physiological data and user historical data, the Bayesian fusion algorithm can combine these data sources more intelligently to optimize the prediction and analysis results. The weights can be dynamically adjusted according to the reliability, relevance and predictive ability of the data. By comprehensively considering multiple data sources, the Bayesian fusion algorithm helps to improve the accuracy of identifying health trends and potential problems; the Bayesian fusion algorithm can consider the user's personalized historical data and provide more customized health analysis and suggestions for each user.

[0050] Specifically, the user interaction interface provides a historical record query function for user health data. The advantage is that users can track and review their health data history through the query function, which helps users understand their health trends and long-term changes. By accessing historical records, users can better manage their health status, identify potential health problems, and take timely actions; providing a historical record query function can increase user participation and interest in the health management system because users can see the progress and changes in their health status.

[0051] Specifically, the system also includes a security module, which is connected to the user interaction interface and is used for encrypted data transmission and storage. The advantage is that encryption technology can ensure the privacy of user health data during transmission and storage, thereby protecting user health data from leakage; encryption not only protects data from unauthorized access, but also ensures that data is not illegally tampered with during transmission and storage. With the popularization of telemedicine and health management services, the security module ensures the security of health consultation and data exchange through the network.

[0052] Specifically, the user interaction interface also includes a feedback module, which allows users to evaluate the health suggestions provided. The advantage is that by providing a feedback mechanism, the system can collect first-hand information on user feedback, understand user satisfaction with health suggestions and improvement suggestions, thereby continuously optimizing and improving services, and also increasing user participation and satisfaction with the system.

[0053] The system operates in the following steps:

[0054] First, the data acquisition module uses non-invasive sensors to monitor the user's heart rate, heart and lung sounds, blood pressure, body temperature and other physiological data in real time. The collected data is then transmitted to the data processing module to clean and standardize the collected physiological data, use the wavelet transform algorithm to remove noise, and use the Z-score method to detect outliers. The processed data is then sent to the large language model analysis module, which uses the pre-trained BERT or GPT model combined with the long short-term memory network for multimodal analysis to identify health trends and potential problems. The Bayesian fusion algorithm is used to perform weighted fusion of physiological data and user historical data to improve the accuracy of the analysis. The analysis results are encrypted, transmitted and stored through the security module to ensure the privacy and security of the user's health data. Users can log in to the user interaction interface to query visual health reports and personalized health suggestions. At the same time, users can also use the feedback module to provide feedback to the system.

[0055] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A vital sign health intelligent analysis system based on a professional medical large language model, characterized by: It includes a data acquisition module, a data processing module, a large language model analysis module and a user interaction interface, wherein the data acquisition module is connected to the data processing module, the data processing module is connected to the large language model analysis module, and the large language model analysis module is connected to the user interaction interface; A data collection module for collecting physiological data of the user, including but not limited to heart rate, heart and lung sounds, blood pressure and body temperature; The data processing module is used to pre-process the collected physiological data, including data cleaning and standardization, and use signal processing algorithms to remove noise and convert the data into a unified format; The large language model analysis module uses a well-trained large language model to conduct in-depth analysis of pre-processed data to identify health trends and potential problems. The large language model is based on the Transformer architecture and combines physiological data and user historical data for multimodal analysis; The user interaction interface is used to provide users with visual health reports and personalized health recommendations. The user interaction interface is implemented through a mobile application or a web application, and provides multiple forms of interaction such as charts, text descriptions, and voice feedback.

2. According to claim 1, a vital sign health intelligent analysis system based on a professional medical large language model is characterized by: The data acquisition module includes at least one non-invasive sensor for real-time monitoring of heart rate and cardiopulmonary sounds without interfering with the user's normal life.

3. According to claim 1, a vital sign health intelligent analysis system based on a professional medical large language model is characterized by: The data processing module receives the information collected by the data acquisition module and uses a wavelet transform algorithm to remove noise from the data. The algorithm formula is as follows: Among them, W f (a,b) are wavelet transform coefficients, f(t) is the original signal, It is the value of the mother wavelet function after scaling and translation, a is the scale parameter that controls the scaling of the wavelet, b is the translation parameter that controls the position of the wavelet on the time axis, and dt represents the small time interval when integrating time t.

4. According to claim 1, a vital sign health intelligent analysis system based on a professional medical large language model is characterized by: The data processing module also includes detecting outliers, specifically using the Z-score method to detect outliers, and the calculation formula is as follows: Where Z is the calculated Z-score value, X is a data point in the data set, μ is the mean of the data set, and σ is the standard deviation of the data set.

5. According to claim 1, a vital sign health intelligent analysis system based on a professional medical large language model is characterized by: The large language model adopts a pre-trained BERT model or a GPT model, and the large language model analysis module also includes using a long short-term memory network to perform predictive analysis of time series data for identifying health trends.

6. The vital sign health intelligent analysis system based on professional medical large language model according to claim 1 is characterized by: The Transformer architecture contains at least 12 encoder layers.

7. The vital sign health intelligent analysis system based on professional medical large language model according to claim 1 is characterized by: When performing multimodal analysis, the large language model analysis module uses a Bayesian fusion algorithm to perform weighted fusion on physiological data and user historical data. The algorithm formula is as follows: P(yx)=k·P(yx p ) wp ·P(yx h ) wh Among them, P(yx) is the predicted probability after fusion, P(yx p ) and P(yx h ) are the predicted probabilities based on physiological data and historical data, wp and wh are their weights, and k is the normalization factor.

8. The vital signs health intelligent analysis system based on professional medical large language model according to claim 1 is characterized by: The user interaction interface provides a historical record query function for user health data.

9. The vital signs health intelligent analysis system based on professional medical large language model according to claim 1 is characterized by: The system also includes a security module, which is connected to the user interaction interface and is used for encrypting data transmission and storage.

10. The vital sign health intelligent analysis system based on professional medical large language model according to claim 1 is characterized by: The user interaction interface also includes a feedback module, which allows the user to evaluate the provided health advice.

Images