Portable non-invasive dynamic monitoring and health management platform for diabetes mellitus based on multispectral AI fusion

Abstract

The invention relates to the technical field of spectroscopy, and discloses a portable diabetes non-invasive dynamic monitoring and health management platform based on multispectral AI fusion, and the platform comprises a multispectral detection unit which is used for collecting Raman spectrum, near infrared spectrum and fluorescence spectrum data of the skin surface of a user in real time; the quantum dot enhancement sensor adopts a CdSe / ZnS quantum dot material to enhance the acquired weak optical signal, so that the detection sensitivity and accuracy are improved; the AI analysis unit is used for comprehensively analyzing the collected multispectral data, living habit data (such as heart rate and exercise amount) and diet records through an AI model and predicting the blood glucose fluctuation trend. Through the fusion of a multispectral technology and the adoption of the quantum dot enhanced sensor, the noninvasive detection precision of diabetes mellitus is remarkably improved, and the error rate is reduced to 5% or below and is far lower than 8-12% of an international competitive product. The high-precision blood glucose monitoring can provide more accurate health data for the patient, so that the risk of misdiagnosis and missed diagnosis is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of spectroscopy, in particular to a portable diabetes non-invasive dynamic monitoring and health management platform based on multi-spectral AI fusion. BACKGROUND

[0002] Diabetes, especially type 2 diabetes, has become an important public health problem worldwide. With the change of lifestyle, the prevalence of diabetes is increasing year by year. Long-term poor control of diabetes not only affects the quality of life of patients, but also can lead to serious complications such as cardiovascular disease, retinopathy, renal failure, etc. The existing diabetes management method mainly relies on traditional blood glucose monitoring, which is blood glucose testing by fingertip blood sampling. This method not only has invasiveness, but also has the disadvantages of inconvenient operation, frequent detection easily causing discomfort of patients, etc.

[0003] The existing non-invasive monitoring technology, such as near-infrared spectroscopy, Raman spectroscopy and fluorescence spectroscopy, although can detect blood glucose under non-invasive conditions, but the existing technology has the following problems: large error: the error of the existing technology is usually between 8-12%, which is far lower than the standard of clinical blood glucose detection. Poor environmental adaptability: external environmental light, individual differences (such as skin color, stratum corneum thickness) and other factors will affect the detection accuracy. Lack of personalized health management: most of the existing technology only provides single blood glucose monitoring function, and fails to combine the individual characteristics of patients for health intervention, lacking personalized diabetes management strategy.

[0004] Therefore, the present application aims to provide an accurate, efficient, non-invasive and personalized diabetes management platform by the fusion of multi-spectral technology, quantum dot enhanced sensor, AI driven dynamic health management and portable hardware design, to overcome the shortcomings of the existing technology and improve the accuracy, real-time and personalization level of diabetes monitoring. SUMMARY

[0005] To achieve the above-mentioned purpose, the present application provides the following technical scheme: a portable diabetes non-invasive dynamic monitoring and health management platform based on multi-spectral AI fusion, which comprises:

[0006] A multi-spectral detection unit for real-time acquisition of Raman spectroscopy, near-infrared spectroscopy and fluorescence spectroscopy data of the user's skin surface;

[0007] A quantum dot enhanced sensor using CdSe / ZnS quantum dot material to enhance the weak optical signal collected, improve the detection sensitivity and accuracy;

[0008] An AI analysis unit for comprehensive analysis of the collected multi-spectral data, lifestyle data (such as heart rate, exercise amount) and dietary records by AI model, prediction of blood glucose fluctuation trend, and generation of personalized health management suggestions;

[0009] Portable hardware unit: includes a miniaturized spectral sensor and low-power hardware design, enabling long-term blood glucose monitoring with high efficiency and low cost.

[0010] Preferably, the AI ​​analysis unit includes:

[0011] LSTM network: used to predict blood sugar fluctuation trends over the next 6 hours with an accuracy of 92%;

[0012] Federated Learning System: Combines data from multiple hospitals for model training, optimizes prediction accuracy, and improves the sensitivity and accuracy of blood sugar fluctuation predictions.

[0013] Preferably, the AI ​​analysis unit can provide the user with the following personalized intervention plan:

[0014] AI nutritionist function: Generates personalized recipes based on the user's blood sugar fluctuation trends and diet records, and dynamically adjusts carbohydrate intake;

[0015] Exercise suggestion engine: Based on blood sugar fluctuation prediction data, it recommends the optimal exercise time and intensity to help users optimize their exercise plans and reduce the risk of blood sugar fluctuations.

[0016] Preferably, the portable hardware unit of the platform includes:

[0017] Miniaturized spectroscopy module: uses MEMS technology to integrate optical components, measures 10mm×10mm, and consumes less than 0.5W of power.

[0018] Pen-shaped detector design: weighs less than 50g, supports USB-C interface charging, can support 200 tests per charge, and is equipped with a disposable disinfection patch to avoid cross infection.

[0019] Preferably, the platform supports edge-cloud collaborative architecture:

[0020] Local data encryption storage: User data is stored locally in an encrypted manner to prevent sensitive data leakage;

[0021] Cloud-based desensitized data upload: Only de-identified feature values ​​are uploaded to the cloud to ensure user privacy.

[0022] Blockchain evidence storage: Use blockchain technology to encrypt and store all test results to ensure the authenticity and non-tamperability of the results.

[0023] Compared with the existing technology, the present invention provides a portable non-invasive dynamic monitoring and health management platform for diabetes based on multispectral AI fusion, which has the following beneficial effects:

[0024] 1. This portable, non-invasive, dynamic diabetes monitoring and health management platform, based on multispectral AI fusion, significantly improves the accuracy of non-invasive diabetes testing through the integration of multispectral technology and the use of quantum dot-enhanced sensors, reducing the error rate to below 5%, far lower than the 8-12% of international competitors. This high-precision blood glucose monitoring can provide patients with more accurate health data, thereby reducing the risk of misdiagnosis and missed diagnosis.

[0025] 2. This portable non-invasive dynamic diabetes monitoring and health management platform based on multi-spectral AI fusion and AI-based dynamic health management system can comprehensively analyze the user's blood sugar fluctuation data, exercise, diet and other aspects, and provide personalized health intervention plans, such as customized diet, dynamic exercise recommendations, etc., to help patients better control blood sugar and prevent diabetic complications.

[0026] 3. This portable, non-invasive, dynamic diabetes monitoring and health management platform, based on multispectral AI fusion, is designed as a portable device. It is lightweight and compact, making it easy to carry around daily and supporting long-term non-invasive monitoring. Users can monitor their blood sugar status at any time without frequent blood draws, greatly improving the convenience and comfort of monitoring.

[0027] 4. This portable non-invasive dynamic diabetes monitoring and health management platform based on multispectral AI fusion ensures the privacy and security of user data through the combination of edge computing and blockchain evidence storage technology. It complies with current data protection regulations and enhances user trust. It is particularly suitable for scenarios involving insurance claims.

[0028] 5. This portable non-invasive dynamic monitoring and health management platform for diabetes based on multi-spectral AI fusion has good scalability. In the future, its functions can be expanded according to user needs and technological development. For example, it can be used to monitor diseases such as hyperuricemia and dyslipidemia, or combined with industries such as agriculture and food to promote the innovative development of cross-industry applications. DETAILED DESCRIPTION

[0029] The following is a clear and complete description of the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of the present invention.

[0030] Example

[0031] Example of a portable non-invasive dynamic monitoring and health management platform for diabetes based on multispectral AI fusion

[0032] A portable non-invasive dynamic diabetes monitoring and health management platform based on multispectral AI fusion. The management platform includes:

[0033] Multispectral detection unit: used to collect Raman spectrum, near-infrared spectrum and fluorescence spectrum data of the user's skin surface in real time;

[0034] Quantum dot enhanced sensor: uses CdSe / ZnS quantum dot materials to enhance the collected weak optical signals, improving detection sensitivity and accuracy;

[0035] AI analysis unit: uses AI models to comprehensively analyze collected multispectral data, lifestyle data (such as heart rate and exercise volume), and dietary records to predict blood sugar fluctuation trends and generate personalized health management recommendations;

[0036] Portable hardware unit: includes a miniaturized spectral sensor and low-power hardware design, enabling long-term blood glucose monitoring with high efficiency and low cost.

[0037] Specifically, the AI ​​analysis unit includes:

[0038] LSTM network: used to predict blood sugar fluctuation trends over the next 6 hours with an accuracy of 92%;

[0039] Federated Learning System: Combines data from multiple hospitals for model training, optimizes prediction accuracy, and improves the sensitivity and accuracy of blood sugar fluctuation predictions.

[0040] Specifically, the AI ​​analysis unit can provide users with the following personalized intervention plans:

[0041] AI nutritionist function: Generates personalized recipes based on the user's blood sugar fluctuation trends and diet records, and dynamically adjusts carbohydrate intake;

[0042] Exercise suggestion engine: Based on blood sugar fluctuation prediction data, it recommends the optimal exercise time and intensity to help users optimize their exercise plans and reduce the risk of blood sugar fluctuations.

[0043] Specifically, the platform's portable hardware units include:

[0044] Miniaturized spectroscopy module: uses MEMS technology to integrate optical components, measures 10mm×10mm, and consumes less than 0.5W of power.

[0045] Pen-shaped detector design: weighs less than 50g, supports USB-C interface charging, can support 200 tests per charge, and is equipped with a disposable disinfection patch to avoid cross infection.

[0046] Specifically, the platform supports edge-cloud collaborative architecture:

[0047] Local data encryption storage: User data is stored locally in an encrypted manner to prevent sensitive data leakage;

[0048] Cloud-based desensitized data upload: Only de-identified feature values ​​are uploaded to the cloud to ensure user privacy.

[0049] Blockchain evidence storage: Use blockchain technology to encrypt and store all test results to ensure the authenticity and non-tamperability of the results.

[0050] Through the above-mentioned technical solution, the present invention significantly improves the accuracy of non-invasive diabetes testing by integrating multispectral technology and adopting quantum dot-enhanced sensors, reducing the error rate to below 5%, far lower than the 8-12% of international competitors. This high-precision blood glucose monitoring can provide patients with more accurate health data, thereby reducing the risk of misdiagnosis and missed diagnosis. The AI-based dynamic health management system can comprehensively analyze the user's blood glucose fluctuation data, exercise, diet, and other aspects to provide personalized health intervention plans, such as customized diet and dynamic exercise recommendations, to help patients better control blood glucose and prevent diabetic complications. The platform is designed as a portable device with a lightweight and compact size, making it easy to carry daily and supporting long-term non-invasive monitoring. Users can monitor their blood glucose status at any time without frequent blood draws, greatly improving the convenience and comfort of monitoring. By combining edge computing and blockchain evidence storage technology, the platform ensures the privacy and security of user data, complies with current data protection regulations, and enhances user trust, making it particularly suitable for scenarios involving insurance claims. The platform has good scalability and its functions can be expanded in the future according to user needs and technological development. For example, it can be used to monitor diseases such as hyperuricemia and dyslipidemia, or combined with industries such as agriculture and food to promote the innovative development of cross-industry applications.

[0051] Multispectral detection unit

[0052] Raman spectroscopy:

[0053] Raman spectroscopy uses laser light to illuminate the skin, identifying markers of sugar metabolism (such as glucose and AGEs) and acquiring corresponding spectral data. This data can reflect the sugar metabolism process in the skin layer and provide an accurate quantitative basis for blood sugar monitoring.

[0054] Near-infrared spectroscopy:

[0055] Near-infrared spectroscopy is primarily used to penetrate the dermis and detect subcutaneous blood glucose levels. Within the 950-1050nm wavelength range, near-infrared light can effectively penetrate the skin and capture changes in blood glucose levels.

[0056] Fluorescence spectrum:

[0057] Fluorescence spectroscopy is mainly used to capture the fluorescence signal generated by AGEs (advanced glycation end products) in the skin, indirectly reflecting long-term blood glucose control level and chronic complications of diabetes.

[0058] Quantum dot enhanced sensor:

[0059] The quantum dot sensor uses CdSe / ZnS quantum dot materials to amplify weak light signals, improving the detection sensitivity of the system and making trace amounts of optical signals more noticeable.

[0060] LSTM network predicts blood glucose fluctuations:

[0061] Using a long short-term memory (LSTM) network model, the platform can analyze historical blood glucose data and lifestyle data to predict the future 6-hour blood glucose fluctuation trend. Based on this prediction, the platform can issue an early warning when the blood glucose trend appears, and provide intervention suggestions.

[0062] Personalized intervention suggestions:

[0063] The system can combine individual lifestyle habits (such as diet, exercise) with blood glucose fluctuation trends to provide precise dietary control suggestions and exercise planning for users, reducing the occurrence of blood glucose fluctuations.

[0064] Miniaturized spectral module:

[0065] Through MEMS technology, the integrated optical element miniaturizes the spectral module, with a volume of only 10mm×10mm and a power consumption as low as 0.5W, making it suitable for long-term use.

[0066] Pen-type detector design:

[0067] The device adopts a pen-type design, weighing no more than 50g, supporting USB-C interface for charging, and a single charge can complete 200 detections, greatly improving the convenience and continuous use.

[0068] Through edge computing and cloud encryption storage technology, the platform can protect users' personal health data, while using blockchain technology to ensure the non-tamperability and integrity of the detection results, meeting the data security requirements of medical, insurance and other fields.

[0069] Although embodiments of the present application have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and alterations can be made without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A portable non-invasive dynamic diabetes monitoring and health management platform based on multispectral AI fusion, featuring: The management platform includes: Multispectral detection unit: used to collect Raman spectrum, near-infrared spectrum and fluorescence spectrum data of the user's skin surface in real time; Quantum dot enhanced sensor: uses CdSe / ZnS quantum dot materials to enhance the collected weak optical signals, improving detection sensitivity and accuracy; AI analysis unit: uses AI models to comprehensively analyze collected multispectral data, lifestyle data (such as heart rate and exercise volume), and dietary records to predict blood sugar fluctuation trends and generate personalized health management recommendations; Portable hardware unit: includes a miniaturized spectral sensor and low-power hardware design, enabling long-term blood glucose monitoring with high efficiency and low cost.

2. The portable non-invasive dynamic monitoring and health management platform for diabetes based on multispectral AI fusion according to claim 1 is characterized by: in, The AI ​​analysis unit includes: LSTM network: used to predict blood sugar fluctuation trends over the next 6 hours with an accuracy of 92%; Federated Learning System: Combines data from multiple hospitals for model training, optimizes prediction accuracy, and improves the sensitivity and accuracy of blood sugar fluctuation predictions.

3. The portable non-invasive dynamic monitoring and health management platform for diabetes based on multispectral AI fusion according to claim 1 is characterized by: The AI ​​analysis unit can provide users with the following personalized intervention plans: AI nutritionist function: Generates personalized recipes based on the user's blood sugar fluctuation trends and diet records, and dynamically adjusts carbohydrate intake; Exercise suggestion engine: Based on blood sugar fluctuation prediction data, it recommends the optimal exercise time and intensity to help users optimize their exercise plans and reduce the risk of blood sugar fluctuations.

4. The portable non-invasive dynamic monitoring and health management platform for diabetes based on multispectral AI fusion according to claim 1 is characterized by: The portable hardware unit of the platform includes: Miniaturized spectroscopy module: uses MEMS technology to integrate optical components, measures 10mm×10mm, and consumes less than 0.5W of power. Pen-shaped detector design: weighs less than 50g, supports USB-C interface charging, can support 200 tests per charge, and is equipped with a disposable disinfection patch to avoid cross infection.

5. The portable non-invasive dynamic monitoring and health management platform for diabetes based on multispectral AI fusion according to claim 1 is characterized by: The platform supports edge-cloud collaborative architecture: Local data encryption storage: User data is stored locally in an encrypted manner to prevent sensitive data leakage; Cloud-based desensitized data upload: Only de-identified feature values ​​are uploaded to the cloud to ensure user privacy. Blockchain evidence storage: Use blockchain technology to encrypt and store all test results to ensure the authenticity and non-tamperability of the results.