Personalized health intervention and automated commercial service method and system based on multi-dimensional physiological monitoring

Abstract

The application discloses a kind of personalized health intervention and automation commodity service method and system based on multidimensional physiological monitoring, applied to intelligent medical technology field.The method comprises the following steps: collecting the multidimensional physiological health monitoring data of user;Input to health risk prediction model, output health risk prediction result;Based on the health risk prediction result of user, long-term health record and real-time physiological state, dynamically generate the personalized commodity demand constraint of user;Using multi-objective optimization algorithm to complete commodity matching screening, obtain at least one group of commodity combination satisfying the current health demand of user;Commodity combination is pushed to user terminal, if receiving the confirmation instruction of user, automatically generate order and complete ordering purchase.The present application can comprehensively evaluate the health status of user, realize more accurate health risk prediction;Consider individual differences of user, realize real personalized health intervention;Realize the full automation process from health suggestion to commodity purchase.

Description

Technical Field

[0001] This invention relates to the field of smart healthcare technology, and more specifically to a method and system for personalized health intervention and automated commodity services based on multi-dimensional physiological monitoring. Background Technology

[0002] With the rapid development of the Internet of Things (IoT) and sensor technology, personal health monitoring devices are becoming increasingly common, ranging from simple pedometers to professional medical-grade devices such as continuous glucose monitors and smart blood pressure monitors, providing users with increasingly rich physiological data collection capabilities. However, existing technologies still face the following significant bottlenecks in the entire health management process: 1. Limited Monitoring Dimensions: Most health monitoring devices focus on only a single physiological indicator, such as blood glucose, blood pressure, or heart rate. They lack the ability to integrate and analyze multi-dimensional physiological data. The human body is a complex organic whole, and various physiological indicators have close interrelationships. For example, fluctuations in blood glucose can affect blood pressure stability, and sleep apnea can lead to decreased blood oxygenation and subsequently cardiovascular problems. Monitoring a single indicator cannot comprehensively assess a user's overall health status and may easily overlook potential health risks. 2. Generalized Intervention Recommendations: Most existing health management devices provide general health recommendations, such as a low-salt diet and moderate exercise. They fail to provide personalized guidance based on individual differences such as the user's real-time physiological state, medical history, and allergies, making it difficult to achieve the desired health intervention effect. 3. Lack of Implementation: Health advice often remains at the information push level. After receiving the advice, users still need to search, filter, and purchase the necessary health-related products themselves. This cumbersome process leads to low user compliance, and many effective health suggestions cannot be translated into actual action. 4. Disconnect between Products and Services: Currently, there is a lack of a closed-loop system on the market that can automatically match and purchase suitable products (covering multiple categories such as food, medicine, health products, medical devices, and home monitoring consumables) from various e-commerce platforms, local pharmacies, and service platforms based on users' real-time health needs. Health management services and product purchase services are separated, failing to form a complete health management chain. Therefore, how to provide a method and system for personalized health intervention and automated product services based on multi-dimensional physiological monitoring is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0003] In view of this, the present invention provides a method and system for personalized health intervention and automated commodity services based on multi-dimensional physiological monitoring. Through multi-dimensional physiological monitoring, dynamic personalized demand modeling, and cross-platform automated purchasing, a closed loop of the entire health management process of "perception-analysis-decision-execution" is realized.

[0004] To achieve the above objectives, the present invention provides the following technical solution: A method for personalized health intervention and automated product services based on multi-dimensional physiological monitoring includes the following steps: S1. Collect multi-dimensional physiological health monitoring data from users through wearable devices, home medical devices, and user input interfaces; S2. Input multi-dimensional physiological health monitoring data into a pre-trained health risk prediction model and output the user's health risk prediction results within a preset future time window. S3. Based on the user's health risk prediction results, long-term health records and real-time physiological status, dynamically generate the user's personalized product demand constraints at the current moment. S4. Based on personalized product demand constraints and user personal preferences, a multi-objective optimization algorithm is used to complete product matching and screening, and obtain at least one set of products that meet the user's current health needs; S5. Push the product combination to the user's terminal. If the user confirms the order, automatically generate an order and complete the purchase.

[0005] Optional multidimensional physiological health monitoring data includes: continuous blood glucose monitoring data, blood pressure monitoring data, blood oxygen saturation data, body temperature data, respiratory rate data, electrocardiogram data, heart rate variability data, exercise and resting data, sleep structure data, stress index data, body fat percentage data, and data on women's menstrual cycle stages.

[0006] Optionally, the health risk prediction model includes a data input layer, a multi-branch feature extractor, a multi-dimensional attention fusion module, and a multi-task prediction head. The multi-branch feature extractor includes a time-series data feature extraction branch, a waveform data feature extraction branch, a statistical data feature extraction branch, and a periodic data feature extraction branch. The multi-dimensional attention fusion module performs weighted fusion of multi-dimensional physiological health monitoring data based on different prediction objectives, and the multi-task prediction head outputs the corresponding prediction results.

[0007] Optional health risk prediction results include: blood glucose prediction trajectory, probability of hypoglycemia risk, probability of hyperglycemia risk, warning of hypertension risk, warning of hypotension risk, warning of arrhythmia risk, risk of decreased blood oxygen, warning of abnormal body temperature, warning of nutritional deficiency risk, and warning of metabolic characteristics changes based on the female menstrual cycle.

[0008] Optional, personalized product demand constraints include: required nutrient types and intake amounts, recommended product categories, permissible glycemic index range, upper limit for sodium content, recommended potassium content, specific needs for blood pressure management, specific needs for blood oxygen management, specific needs for body temperature management, and specific needs for women's menstrual cycles.

[0009] Optionally, in S4, unstructured product information is parsed into structured attribute tags through natural language processing and product knowledge graph. The product knowledge graph includes a basic attribute layer, a semantic mapping layer, and a dynamic correction layer. The basic attribute layer contains standardized data of the product, the semantic mapping layer maps the non-standardized description of the product to the basic attribute layer, and the dynamic correction layer adjusts the health benefit score of the product in the knowledge graph based on the physiological response data fed back by the user after purchase.

[0010] Optionally, in S4, an objective function is constructed with the optimization objectives of minimizing health risks, maximizing health benefits, maximizing user preference matching, and minimizing the total price of the product combination, and a genetic algorithm is used to solve it.

[0011] Optionally, S5 also includes: if in fully automated managed mode and the safety score and health benefit score of the product combination exceed the preset threshold, an order will be automatically generated and the corresponding platform API will be called through the pre-authorized payment interface to complete the order purchase and delivery scheduling. The fully automated managed mode is equipped with a circuit breaker mechanism, including emergency circuit breaker for hypoglycemia, circuit breaker for hypertensive crisis, circuit breaker for hypotensive syncope risk, circuit breaker for product conflict detection, and circuit breaker for inventory timeliness.

[0012] Optionally, after S5, it also includes: S6. Record the actual physiological response data of users after each purchase of product combinations, and use the deviation between the actual effect and the predicted effect as a feedback signal to regularly optimize the health risk prediction model, product knowledge graph and user personal preferences.

[0013] A personalized health intervention and automated commodity service system based on multi-dimensional physiological monitoring, applying the aforementioned method for personalized health intervention and automated commodity service based on multi-dimensional physiological monitoring, includes: A multi-dimensional data acquisition terminal is used to collect multi-dimensional physiological health monitoring data from users; The cloud-based health center is equipped with a health risk prediction model, a dynamic constraint generator, a product knowledge graph, a multi-objective optimization matcher, and an automated decision engine. The health risk prediction model outputs the user's health risk prediction results within a preset future time window. The dynamic constraint generator generates the user's personalized product demand constraints at the current moment. The multi-objective optimization matcher completes product matching and filtering. The automated decision engine generates orders and completes the purchase. User interaction terminal, used for data display, solution confirmation and system configuration; A unified service gateway connects to multiple third-party platforms, enabling cross-platform product retrieval and order fulfillment.

[0014] As can be seen from the above technical solution, compared with the prior art, the present invention provides a personalized health intervention and automated commodity service method and system based on multi-dimensional physiological monitoring, which has the following beneficial effects: The present invention integrates multi-dimensional physiological data, which can comprehensively assess the user's health status and achieve more accurate health risk prediction; it fully considers individual differences of users and combines real-time physiological status to generate dynamic personalized demand constraints, thereby achieving truly personalized health intervention; it realizes a fully automated process from health advice to commodity purchase, which greatly reduces the user's operating costs and effectively improves the user's compliance with health intervention; it has a built-in multi-scenario safety circuit breaker mechanism, which can automatically take corresponding measures when an emergency health condition occurs, and ensure the safety of automated purchase through drug interaction detection; through a closed-loop feedback mechanism, it can continuously learn the user's individual physiological response characteristics, so that the service quality can continuously improve with the increase of usage time. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0016] Figure 1 This is a flowchart of the personalized health intervention and automated product service method of the present invention; Figure 2 This is a schematic diagram of the health risk prediction model of the present invention; Figure 3 This is a schematic diagram of the product knowledge graph structure and intelligent matching process of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] This invention discloses a method for personalized health intervention and automated commodity services based on multi-dimensional physiological monitoring, such as... Figure 1 As shown, it includes the following steps: S1. Collect multi-dimensional physiological health monitoring data from users through wearable devices, home medical devices, and user input interfaces; S2. Input multi-dimensional physiological health monitoring data into a pre-trained health risk prediction model and output the user's health risk prediction results within a preset future time window. S3. Based on the user's health risk prediction results, long-term health records and real-time physiological status, dynamically generate the user's personalized product demand constraints at the current moment. S4. Based on personalized product demand constraints and user personal preferences, a multi-objective optimization algorithm is used to complete product matching and screening, and obtain at least one set of products that meet the user's current health needs; S5. Push the product combination to the user's terminal. If the user confirms the order, automatically generate an order and complete the purchase.

[0019] Furthermore, multidimensional physiological health monitoring data includes: continuous blood glucose monitoring data, blood pressure monitoring data (including systolic blood pressure, diastolic blood pressure, and mean arterial pressure), blood oxygen saturation data, body temperature data, respiratory rate data, electrocardiogram data, heart rate variability data, exercise and resting data, sleep structure data, stress index data, body fat percentage data, and data on women's menstrual cycle stages.

[0020] In this embodiment of the invention, blood pressure monitoring data is acquired through a smart blood pressure monitor or a wearable device with blood pressure monitoring function, including ambulatory blood pressure monitoring data; blood oxygen saturation data is acquired through a pulse oximeter or a smartwatch; body temperature data is acquired through a smart body temperature patch or an infrared thermometer; and electrocardiogram (ECG) data is acquired through a single-lead or multi-lead ECG patch or a smartwatch. Data on the female menstrual cycle stage is acquired through one or a combination of the following methods: manually entered menstrual start date and cycle length; changes in basal body temperature and heart rate variability monitored by a smart wearable device, automatically calculating the current physiological stage (follicular phase, ovulation phase, luteal phase, menstruation) using a cycle prediction model; and the periodic nighttime blood glucose fluctuation pattern observed in continuous blood glucose monitoring data is used to assist in correcting the physiological stage judgment.

[0021] Furthermore, such as Figure 2 As shown, the health risk prediction model includes a data input layer, a multi-branch feature extractor, a multi-dimensional attention fusion module, and a multi-task prediction head. The multi-branch feature extractor includes a time-series data feature extraction branch, a waveform data feature extraction branch, a statistical data feature extraction branch, and a periodic data feature extraction branch. The multi-dimensional attention fusion module performs weighted fusion of multi-dimensional physiological health monitoring data based on different prediction objectives, and the multi-task prediction head outputs the corresponding prediction results.

[0022] In this embodiment of the invention, the data input layer preprocesses multi-dimensional physiological health monitoring data, including time-series alignment, outlier and missing value handling, feature encoding, and standardization. The time-series data feature extraction branch processes blood glucose, blood pressure, heart rate, blood oxygen, and body temperature data, using a 1D convolutional neural network to extract local time-series features and combining a long short-term memory network to capture long-term dependencies. The waveform data feature extraction branch processes electrocardiogram (ECG) data, using a ResNet-based ECG feature extraction network to extract key clinical waveform features. The statistical data feature extraction branch processes step count, sleep duration, deep sleep ratio, and stress index, using a fully connected layer to extract global statistical features. The periodic data feature extraction branch processes data from different stages of the female menstrual cycle, using an embedding layer to convert the cycle stages into low-dimensional dense vectors and capture the differences in metabolic characteristics between different cycle stages.

[0023] The multi-dimensional attention fusion module concatenates the feature vectors extracted from each branch into a unified multimodal feature matrix. For each prediction target (such as blood glucose prediction or blood pressure prediction), the model automatically calculates the correlation score between each physiological data dimension and the prediction target. The correlation score is then converted into attention weights between 0 and 1 using a softmax function, with the sum of all weights equal to 1. The feature vectors of each physiological data dimension are multiplied by their corresponding attention weights and summed to obtain the final feature vector that integrates multi-dimensional information, which is then input into the subsequent multi-task prediction head. For example, in the blood pressure prediction scenario, the attention weights for heart rate, exercise intensity, sleep stage, and stress index are strengthened because these indicators are highly correlated with short-term fluctuations in blood pressure. In the blood glucose prediction scenario, continuous historical blood glucose data and the female menstrual cycle stage are given higher weights. In the blood oxygen prediction scenario, the feature weights for sleep stage, respiratory rate, and heart rate variability are increased to capture the risk of decreased blood oxygenation caused by sleep apnea.

[0024] Furthermore, the health risk prediction results include: blood glucose prediction trajectory, probability of hypoglycemia risk, probability of hyperglycemia risk, warning of hypertension risk, warning of hypotension risk, warning of arrhythmia risk, risk of decreased blood oxygen, warning of abnormal body temperature, warning of nutritional deficiency risk, and indication of metabolic characteristics changes based on the female menstrual cycle.

[0025] In this embodiment of the invention, continuous physiological indicator prediction outputs predicted values ​​for blood glucose, systolic blood pressure, diastolic blood pressure, blood oxygen saturation, and body temperature at each time step within a preset future time window; based on the continuous prediction trajectory, the probability of occurrence of various acute health events is calculated, and risk levels are classified, specifically: Hypoglycemia and hyperglycemia risk: Calculate the percentage of time in the predicted trajectory when blood glucose is below 3.9 mmol / L (hypoglycemia) or above 11.1 mmol / L (hyperglycemia) and the lowest / highest blood glucose value, and output the risk probability; Risk of hypertension and hypotension: Calculate the probability of predicted systolic blood pressure ≥140 mmHg or diastolic blood pressure ≥90 mmHg (hypertension), systolic blood pressure <90 mmHg or diastolic blood pressure <60 mmHg (hypotension). When predicted systolic blood pressure ≥180 mmHg or diastolic blood pressure ≥120 mmHg, a hypertensive crisis warning is directly triggered. Risk of decreased blood oxygenation: Calculates and predicts the duration and minimum value of blood oxygen saturation <94%, and outputs the risk probability; Abnormal body temperature warning: Predicts the probability of body temperature ≥37.3℃ or <35℃ and the peak / trough values; Cardiac arrhythmia risk: Predicting the probability of future occurrence of arrhythmias such as premature beats and atrial fibrillation based on electrocardiogram characteristics.

[0026] Furthermore, personalized product demand constraints include: required nutrient types and intake amounts, recommended product categories, permissible glycemic index range, upper limit for sodium content, recommended potassium content, specific needs for blood pressure management, specific needs for blood oxygen management, specific needs for body temperature management, and specific needs for women's menstrual cycles.

[0027] In this embodiment of the invention, specific needs for blood pressure management include: When the risk of hypertension is predicted, the upper limit of sodium intake (e.g., <1500mg per day) should be added to the demand constraints, and low-sodium foods, high-potassium foods (e.g., bananas, potatoes) and health products with auxiliary antihypertensive effects (e.g., fish oil, coenzyme Q10) should be recommended. When a risk of low blood pressure is anticipated, it is recommended to increase salt intake appropriately, drink plenty of water, and possibly take vasopressors (with a doctor's prescription). Adjust the reminder time for purchasing antihypertensive medications according to the diurnal rhythm of blood pressure (dipper or non-dipper pattern).

[0028] Specific needs for blood oxygen management include: When blood oxygen saturation remains below 94%, it is recommended to purchase a home oxygen concentrator and blood oxygen monitor, and it is advised to seek medical attention. For patients with sleep apnea syndrome, it is recommended to purchase a ventilator and related consumables based on nighttime blood oxygen fluctuations.

[0029] Furthermore, such as Figure 3As shown, in S4, unstructured product information is parsed into structured attribute tags through natural language processing and a product knowledge graph. The product knowledge graph includes a basic attribute layer, a semantic mapping layer, and a dynamic correction layer. The basic attribute layer contains standardized data of the product, such as nutritional components (sodium, potassium, carbohydrates, etc.), drug ingredients, applicable population, contraindications, and medical device registration certificate numbers. The semantic mapping layer maps the product's non-standardized description (conversational title, introduction) to the basic attribute layer. The dynamic correction layer adjusts the product's health benefit score in the knowledge graph based on physiological response data (such as postprandial blood glucose and post-medication blood pressure) provided by the user after purchase.

[0030] Furthermore, in S4, an objective function is constructed with the optimization objectives of minimizing health risks, maximizing health benefits, maximizing user preference matching, and minimizing the total price of the product combination, and a genetic algorithm is used to solve it.

[0031] Furthermore, S5 also includes: if in fully automated managed mode and the safety score and health benefit score of the product combination exceed the preset threshold, an order will be automatically generated and the corresponding platform API will be called through the pre-authorized payment interface to complete the order purchase and delivery scheduling. The fully automated managed mode is equipped with a circuit breaker mechanism, including emergency circuit breaker for hypoglycemia, circuit breaker for hypertensive crisis, circuit breaker for hypotensive syncope risk, circuit breaker for product conflict detection, and circuit breaker for inventory timeliness.

[0032] In this embodiment of the invention, the circuit breaker mechanism is specifically as follows: Hypoglycemia emergency circuit breaker: When real-time blood glucose falls below a preset threshold and drops rapidly, the system automatically cancels the original planned order, prioritizes the purchase of fast-raising blood sugar foods from the nearest pharmacy or convenience store, and triggers an emergency notification; Hypertensive crisis circuit breaker: When real-time blood pressure is higher than 180 / 120 mmHg and does not drop, the system will automatically call the emergency center and send the location, and at the same time suggest purchasing emergency antihypertensive drugs (if available). Low blood pressure fainting risk circuit breaker: When blood pressure drops suddenly and is accompanied by abnormal heart rate, the system will automatically notify emergency contacts; Product conflict detection circuit breaker: Before purchasing medicines or health products, the system automatically detects interactions with the user's current medications (such as antihypertensive drugs or hypoglycemic drugs). If contraindications exist, the system will prevent the order from being placed and notify the user. Inventory and delivery time circuit breaker: If the selected product is out of stock or the delivery time is too long to meet the urgent needs, the system will automatically switch to the alternative solution or prompt the user.

[0033] Furthermore, after S5, it also includes: S6. Record the actual physiological response data of users after each purchase of product combinations, and use the deviation between the actual effect and the predicted effect as a feedback signal to regularly optimize the health risk prediction model, product knowledge graph and user personal preferences.

[0034] In one embodiment of the present invention, the user is 55 years old, has type 2 diabetes and hypertension, and wears the following device: CGM: Abbott Libre 3, uploads blood glucose levels every 15 minutes; Smart blood pressure monitor: Omron Bluetooth blood pressure monitor, measures twice a day, morning and evening, with data automatically synchronized; Smartwatch: Huawei Watch D (supports blood pressure measurement and blood oxygen monitoring), which can perform multiple blood pressure measurements and continuous blood oxygen monitoring throughout the day; Smart temperature patch: continuously monitors body temperature to help determine infection or metabolic status; ECG patch: Wear once a week to record a 24-hour ECG for screening arrhythmias; All data is uploaded to the cloud in real time via a mobile app.

[0035] At a certain point in time, the system received the following data: Blood glucose: 5.8 mmol / L, stable; Blood pressure: 145 / 92 mmHg, slightly high; Blood oxygen: 97%, normal; Heart rate: 78 bpm; Activity: Very few steps in the past 2 hours (sedentary); Stress index: moderately high; Medication record: Blood pressure medication (irbesartan) was taken in the morning. After comprehensive analysis, the predictive model outputs the risk prediction for the next 2 hours: Blood glucose: Remains stable with no risk of hypoglycemia; Blood pressure: May rise further to 155 / 95 mmHg, triggering a hypertension warning; Blood oxygen: Remains normal. Recommendation: Current blood pressure is poorly controlled and requires enhanced intervention.

[0036] Based on the prediction results, the system generates personalized product demand constraints for the current moment: Primary need: Lowering blood pressure, requiring low-sodium foods or health supplements that can help lower blood pressure; Secondary need: Long-term blood sugar management, maintaining a low-carbohydrate, high-fiber diet. Constraint structuring: Sodium intake: Sodium content per serving <200mg; Potassium intake: Recommended potassium content per serving >200mg (helps lower blood pressure); Carbohydrates: Carbohydrates per serving <20g, GI <55; Product type: Permitted food, health supplement; Specific categories: Low-sodium snacks, high-potassium fruits, foods rich in Omega-3 (such as deep-sea fish), Coenzyme Q10 supplements, etc.; Contraindications: Avoid interactions with antihypertensive drugs, such as grapefruit. System startup matching process: 1. Product Search: Simultaneously search for nearby supermarkets and online e-commerce platforms through a unified gateway. Search keywords include "low sodium", "salt-free", "high potassium", "deep-sea fish", "coenzyme Q10", etc.

[0037] 2. Knowledge Graph Analysis: The product "Low-Salt Soda Crackers (100g)" is described as having a sodium content of 80mg / 100g, carbohydrates of 65g, a medium glycemic index (GI), and is suitable for people with high blood pressure. The product "Banana (500g)" is described as having a potassium content of 350mg / 100g, carbohydrates of 22g / 100g, a medium glycemic index (GI), and low sodium content. The product "Deep-Sea Fish Oil Soft Capsules" is described as containing Omega-3, with the effects of helping to lower blood lipids and blood pressure, and is suitable for people with high blood pressure, high cholesterol, and high blood sugar. 3. Multi-objective optimization: The system comprehensively considers health needs (low sodium, high potassium, low carbohydrates, and blood pressure reduction), user preferences (liking cookies), and price; ultimately outputting two Pareto optimal solutions: Option A: 2 packs of low-sodium soda crackers + 1 banana + 1 bottle of deep-sea fish oil capsules. Total price: ¥89; Option B: 1 can of unsalted mixed nuts + 1 serving of spinach salad + 1 bottle of Coenzyme Q10 capsules. Total price: ¥120; The system is set to a "semi-automatic managed mode," meaning user confirmation is required. The system then presents two options, A and B, along with health benefit scores. The user selects option A and clicks confirm; the system then places orders through Meituan Maicai and JD Health respectively. Safety Circuit Breaker: If the user selects the fully automated managed mode, the system will automatically place the order. However, after the order is generated, if real-time monitoring shows a rapid increase in blood pressure to 180 / 110 mmHg, the system will immediately trigger the hypertensive crisis circuit breaker. 1. Automatically cancel the original order (if it has not yet been shipped); 2. Search for the nearest pharmacy and purchase emergency blood pressure medication; 3. Simultaneously send an alert via the app and SMS: "Critical blood pressure reading! Emergency blood pressure medication has been purchased for you. Please take it immediately and contact your doctor." After the user consumed Option A, the system continuously monitored their blood pressure changes 2 hours post-meal and found that the decrease in blood pressure was slightly less than expected. The system recorded the deviation and updated the blood pressure-lowering effect coefficient of "low-sodium soda crackers" on the user in the knowledge graph. At the same time, after collecting a large amount of feedback data, the predictive model was fine-tuned periodically to more accurately reflect the individual's response to food and medication.

[0038] In one embodiment of the present invention, if the user is a hypertensive patient, and the system detects that their blood pressure monitor battery is low and the cuff has been used for more than 2 years, and that they only have 5 blood glucose test strips left in stock, the system automatically generates a recommended combination: Blood pressure cuff (1 piece) + blood glucose test strips (50 strips) + blood pressure monitor batteries (4 pieces).

[0039] Once the user confirms, the system will automatically place the order.

[0040] and Figure 1 Corresponding to the method described above, this invention also discloses a personalized health intervention and automated commodity service system based on multi-dimensional physiological monitoring. The method for personalized health intervention and automated commodity service based on multi-dimensional physiological monitoring includes: A multi-dimensional data acquisition terminal is used to collect multi-dimensional physiological health monitoring data from users; The cloud-based health center is equipped with a health risk prediction model, a dynamic constraint generator, a product knowledge graph, a multi-objective optimization matcher, and an automated decision engine. The health risk prediction model outputs the user's health risk prediction results within a preset future time window. The dynamic constraint generator generates the user's personalized product demand constraints at the current moment. The multi-objective optimization matcher completes product matching and filtering. The automated decision engine generates orders and completes the purchase. User interaction terminal, used for data display, solution confirmation and system configuration; A unified service gateway connects to multiple third-party platforms, enabling cross-platform product retrieval and order fulfillment.

[0041] In this embodiment of the invention, the multi-dimensional data acquisition terminal specifically includes: a continuous glucose monitor (such as Dekang G6), a smart blood pressure monitor (upper arm type), a pulse oximeter (finger clip type), a smart body temperature patch (continuous monitoring), an ECG patch (single lead), a smart watch / bracelet (collecting heart rate, heart rate variability, blood oxygen, exercise steps, sleep, and stress), a body fat scale, and a user terminal APP (for manually inputting menstrual period, symptoms, etc.).

[0042] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0043] Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for personalized health intervention and automated commodity services based on multi-dimensional physiological monitoring, characterized in that, Includes the following steps: S1. Collect multi-dimensional physiological health monitoring data from users through wearable devices, home medical devices, and user input interfaces; S2. Input multi-dimensional physiological health monitoring data into a pre-trained health risk prediction model and output the user's health risk prediction results within a preset future time window. S3. Based on the user's health risk prediction results, long-term health records and real-time physiological status, dynamically generate the user's personalized product demand constraints at the current moment. S4. Based on personalized product demand constraints and user personal preferences, a multi-objective optimization algorithm is used to complete product matching and screening, and obtain at least one set of products that meet the user's current health needs; S5. Push the product combination to the user's terminal. If the user confirms the order, automatically generate an order and complete the purchase.

2. The method of claim 1, wherein the method further comprises: Multidimensional physiological health monitoring data includes: continuous blood glucose monitoring data, blood pressure monitoring data, blood oxygen saturation data, body temperature data, respiratory rate data, electrocardiogram data, heart rate variability data, exercise and resting data, sleep structure data, stress index data, body fat percentage data, and data on women's menstrual cycle stages.

3. The method of claim 1, wherein the method further comprises: The health risk prediction model includes a data input layer, a multi-branch feature extractor, a multi-dimensional attention fusion module, and a multi-task prediction head. The multi-branch feature extractor includes branches for extracting features from time-series data, waveform data, statistical data, and periodic data. The multi-dimensional attention fusion module performs weighted fusion of multi-dimensional physiological health monitoring data based on different prediction objectives. The multi-task prediction head outputs the corresponding prediction results.

4. The method of claim 3, wherein the method further comprises: Health risk prediction results include: blood glucose prediction trajectory, probability of hypoglycemia risk, probability of hyperglycemia risk, warning of hypertension risk, warning of hypotension risk, warning of arrhythmia risk, risk of decreased blood oxygen, warning of abnormal body temperature, warning of nutritional deficiency risk, and indication of metabolic characteristics changes based on women's menstrual cycle.

5. The method of claim 1, wherein the method further comprises: Personalized product demand constraints include: required nutrient types and intake amounts, recommended product categories, permissible glycemic index range, upper limit for sodium content, recommended potassium content, specific needs for blood pressure management, specific needs for blood oxygen management, specific needs for body temperature management, and specific needs for women's menstrual cycles.

6. The method of claim 1, wherein the method further comprises: In S4, unstructured product information is parsed into structured attribute tags through natural language processing and product knowledge graph. The product knowledge graph includes a basic attribute layer, a semantic mapping layer, and a dynamic correction layer. The basic attribute layer contains standardized product data, the semantic mapping layer maps the non-standardized description of the product to the basic attribute layer, and the dynamic correction layer adjusts the health benefit score of the product in the knowledge graph based on the physiological response data fed back by users after purchase.

7. The method of claim 1, wherein the method further comprises: In S4, the objective function is constructed with the optimization objectives of minimizing health risks, maximizing health benefits, maximizing user preference matching degree, and minimizing the total price of the product combination, and a genetic algorithm is used to solve it.

8. The method of claim 1, wherein the method further comprises: S5 also includes: if in fully automated managed mode and the safety score and health benefit score of the product combination exceed the preset threshold, an order will be automatically generated and the corresponding platform API will be called through the pre-authorized payment interface to complete the order purchase and delivery scheduling. The fully automated managed mode is equipped with a circuit breaker mechanism, including emergency circuit breaker for hypoglycemia, circuit breaker for hypertensive crisis, circuit breaker for hypotensive syncope risk, circuit breaker for product conflict detection, and circuit breaker for inventory timeliness.

9. The method of claim 1, wherein the method further comprises: Following S5 are: S6. Record the actual physiological response data of users after each purchase of product combinations, and use the deviation between the actual effect and the predicted effect as a feedback signal to regularly optimize the health risk prediction model, product knowledge graph and user personal preferences.

10. A personalized health intervention and automated commodity service system based on multi-dimensional physiological monitoring, employing the personalized health intervention and automated commodity service method based on multi-dimensional physiological monitoring as described in any one of claims 1-9, characterized in that, include: A multi-dimensional data acquisition terminal is used to collect multi-dimensional physiological health monitoring data from users; The cloud-based health center is equipped with a health risk prediction model, a dynamic constraint generator, a product knowledge graph, a multi-objective optimization matcher, and an automated decision engine. The health risk prediction model outputs the user's health risk prediction results within a preset future time window. The dynamic constraint generator generates the user's personalized product demand constraints at the current moment. The multi-objective optimization matcher completes product matching and filtering. The automated decision engine generates orders and completes the purchase. User interaction terminal, used for data display, solution confirmation and system configuration; A unified service gateway connects to multiple third-party platforms, enabling cross-platform product retrieval and order fulfillment.

Images