Diabetes exercise rehabilitation and diet collaborative management method, device and equipment and medium

Abstract

The invention relates to a diabetes exercise rehabilitation and diet collaborative management method, device and equipment and a medium. The method comprises the following steps: acquiring individual feature data, physiological data and daily behavior data of a diabetic patient, and constructing a personalized portrait of a user; generating an exercise plan and a diet plan according to a set period based on the personalized portrait of the user; in response to the obtained real-time user data, dynamically correcting the exercise plan and the diet plan to obtain an exercise and diet collaborative plan; the real-time user data comprises blood glucose data, diet intake data and exercise data; the exercise and diet collaborative plan comprises exercise items arranged based on time points and diet arrangement; generating a corresponding push instruction according to the time point; the pushing instruction is used for indicating to push corresponding exercise items or diet arrangements to the terminal of the diabetic patient. By adopting the method, the problem of inadaptation caused by a rigid fixed plan can be avoided by an exercise-diet collaborative intervention mechanism, and the management response capability of external variable change is improved.

Description

Technical Field

[0001] The present invention belongs to the field of personalized medical health management, and in particular relates to a method, device, equipment and medium for coordinated management of diabetic exercise rehabilitation and diet. Background Art

[0002] With the development of medical and health management technology, data-driven personalized health intervention technology has emerged. By integrating intelligent sensing devices, physiological signal monitoring, big data analysis and artificial intelligence algorithms, it can achieve dynamic tracking of individual health status and intervention recommendations.

[0003] Traditional health intervention technologies for diagnosing chronic diseases such as diabetes usually rely on experience or fixed templates, arrange daily diet and exercise according to the requirements of medical institutions, and push daily diet and exercise plans to patients.

[0004] However, there is a lack of linkage mechanism between daily diet and exercise arrangements, making it difficult to effectively respond to deviations and disturbances in the patient's actual execution process. It is often impossible to consider the actual completion of the exercise plan in real time, nor to make feedback adjustments based on actual food intake. It is also difficult to dynamically correct and personalize the patient's daily behavior. Summary of the Invention

[0005] Based on this, it is necessary to provide a method, device, equipment and medium for the coordinated management of diabetes exercise rehabilitation and diet that can be dynamically adjusted according to actual deviations to address the above technical problems.

[0006] In a first aspect, the present application provides a method for coordinated management of diabetes exercise rehabilitation and diet, comprising:

[0007] Obtain individual characteristic data, physiological data, and daily behavior data of diabetic patients and build personalized user portraits;

[0008] Generate exercise and diet plans based on user-specific profiles and set cycles;

[0009] In response to the acquired real-time user data, the exercise plan and diet plan are dynamically modified to obtain an exercise-diet collaborative plan; the real-time user data includes blood sugar data, dietary intake data, and exercise data; the exercise-diet collaborative plan includes exercise activities and diet arrangements scheduled based on time points;

[0010] Generate corresponding push instructions according to time points; the push instructions are used to instruct the corresponding exercise items or diet arrangements to be pushed to the diabetic patient terminal.

[0011] In one embodiment, generating an exercise plan and a diet plan based on a user's personalized profile and at a set period includes:

[0012] Based on the exercise type-energy expenditure-blood glucose response library model, the user's personalized profile is used to predict the impact of different exercise combinations and daily behavior data on energy expenditure and blood glucose, and an exercise plan for a certain period is generated; the exercise plan includes the type of exercise, intensity, and duration.

[0013] Based on the food-nutrient-glycemic response library model, a diet plan is obtained by using the user's personalized image and the expected exercise consumption corresponding to the exercise plan through an optimization algorithm; the diet plan includes the type of ingredients, the amount used, and the time period for meals; the food-nutrient-glycemic response library model includes food types and their corresponding GI values.

[0014] In one embodiment, in response to the acquired real-time user data, the exercise plan and diet plan are dynamically modified to obtain an exercise-diet collaborative plan, including:

[0015] Calculate the movement behavior deviation according to the movement data and the movement plan to obtain the movement deviation value;

[0016] Calculate dietary behavior deviation based on dietary intake data and dietary plan to obtain dietary deviation value;

[0017] Based on the multi-objective planning reinforcement model, the exercise plan and diet plan are modified according to the exercise deviation value and diet deviation value to obtain the exercise and diet synergy plan.

[0018] In one embodiment, the motion deviation value is obtained by the following formula:

[0019] ΔE=E_expected-E_actual

[0020] Among them, ΔE is the exercise deviation value; E_expected is the planned exercise calorie consumption value; E_actual is the actual recorded calorie consumption value;

[0021] The dietary deviation value is obtained by the following formula:

[0022] ΔC=C_actual-C_expected

[0023] Among them, ΔC is the dietary deviation value; C_actual is the actual calorie intake; C_expected is the planned calorie intake.

[0024] In one embodiment, based on the multi-objective planning reinforcement model, the exercise plan and the diet plan are modified according to the exercise deviation value and the diet deviation value to obtain an exercise-diet synergy plan, including:

[0025] If the movement deviation value exceeds the first positive threshold, the target is adjusted upward based on energy consumption, and an adjusted movement plan is obtained according to the multi-objective planning reinforcement model;

[0026] If the movement deviation value is less than the first negative threshold, based on increasing the dietary arrangement goal, an adjusted dietary plan is obtained according to the multi-objective planning reinforcement model;

[0027] If the dietary deviation value exceeds the second positive threshold, an adjusted exercise plan is obtained based on the addition of short-term light exercise goals according to the multi-objective planning reinforcement model;

[0028] If the dietary deviation value is less than the second negative threshold, an adjusted dietary plan is obtained according to the multi-objective planning reinforcement model based on the low GI food intake goal.

[0029] In one embodiment, the method includes:

[0030] Record the user's real-time data acquired historically within a preset time period, and obtain diet records, exercise records, and blood sugar records;

[0031] A three-dimensional dynamic graph of diet, exercise, and blood sugar is constructed by combining diet records, exercise records, and blood sugar records in a time series in the form of triples. The graph includes food nodes, exercise nodes, and blood sugar status nodes, as well as edge relationships between the nodes.

[0032] A graph convolutional network based on the graph path attention mechanism is used to learn the path of the diet-exercise-blood sugar three-dimensional dynamic graph and obtain path constraints.

[0033] The generation strategies corresponding to the exercise plan and diet plan are constrained according to the path constraints.

[0034] In one embodiment, the method further comprises:

[0035] According to the blood sugar status node, the map path that causes high blood sugar or low blood sugar trend in the diet-exercise-blood sugar three-dimensional dynamic map is marked as a risk path;

[0036] Identify whether there are anomalies in real-time user data that match risk paths through graph anomaly detection;

[0037] If there is an abnormal situation, the warning instruction will be triggered; the warning instruction is used to instruct the corresponding warning information to be pushed to the diabetic patient terminal; the warning information includes hypoglycemia warning and hyperglycemia warning.

[0038] In a second aspect, the present application also provides a diabetes exercise rehabilitation and diet collaborative management device, comprising:

[0039] Personalization module, used to obtain individual characteristic data, physiological data, and daily behavior data of diabetic patients and build personalized user portraits;

[0040] The planning module is used to generate exercise and diet plans based on the user's personalized profile according to the set cycle;

[0041] A dynamic adjustment module, configured to dynamically modify the exercise plan and diet plan in response to the acquired real-time user data to obtain an exercise-diet collaborative plan;

[0042] The interactive module is used to generate corresponding push instructions according to time points; the push instructions are used to instruct the corresponding exercise items or diet arrangements to be pushed to the diabetic patient terminal.

[0043] In a third aspect, the present application also provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the steps of any of the above-mentioned methods for coordinated management of diabetes exercise rehabilitation and diet.

[0044] In a fourth aspect, the present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above-mentioned methods for coordinated management of diabetes exercise rehabilitation and diet.

[0045] The above-mentioned diabetes exercise rehabilitation and diet collaborative management methods, devices, equipment and media integrate individual characteristics, physiological status and behavioral habits to form a dynamic, multi-dimensional personalized health model, providing a scientific basis for plan formulation and intervention; with exercise and diet as the core control variables, systematically plan and adjust daily energy input and output to effectively ensure the stability of blood sugar levels; through real-time monitoring of user status and dynamic adjustment of intervention plans, avoid the maladaptation problems caused by rigid fixed plans and improve the system's responsiveness to changes in external variables; push the adjusted intervention plan to users via smart terminals, enhancing the closed-loop and implementation rate of rehabilitation execution. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0047] Figure 1 This is a flow chart of the diabetes exercise rehabilitation and diet collaborative management method of the present invention;

[0048] Figure 2 Schematic diagram of the step-by-step process of step S103;

[0049] Figure 3Schematic diagram of the process of optimizing exercise plan and diet plan through dynamic graphs of the present invention;

[0050] Figure 4 This is a structural diagram of the diabetes exercise rehabilitation and diet collaborative management device of the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0052] In one embodiment, Figure 1 As shown, a method for coordinating diabetes exercise rehabilitation and diet management is provided. This embodiment uses the method applied to a terminal as an example. It is understandable that the method can also be applied to a server, or to a system including a terminal and a server, and implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0053] S101. Obtain individual characteristic data, physiological data, and daily behavior data of diabetic patients, and build a personalized user portrait.

[0054] For example, in the process of diabetes exercise rehabilitation and dietary management, achieving personalized and dynamic recommendations requires accurate understanding of an individual's basic health status and behavioral characteristics. Individual characteristic data typically includes the patient's age, gender, height, weight, occupation, daily routine, family dining habits, dietary preferences, dietary restrictions, and previous dietary control experience. Furthermore, individual physiological parameters such as metabolic type and insulin sensitivity need to be collected. These can be obtained from medical records, physical examination reports, or questionnaires. Physiological data focuses on indicators directly related to diabetes control, such as fasting blood glucose, postprandial blood glucose, glycated hemoglobin (HbA1c), basal metabolic rate, and body fat percentage. These can be imported through smart blood glucose meters, wearable physiological monitoring devices, or hospital diagnostic interfaces. Daily behavioral data primarily refers to the patient's daily exercise trajectory, steps, heart rate, energy expenditure, and meal times necessary for daily living, such as commuting.

[0055] Schematically, a multimodal information fusion algorithm is used to model user profiles, creating a structured representation of a user's behavioral habits, health status, and historical response results. This creates a personalized user profile that combines static characteristics such as body shape and metabolic capacity with dynamic behavioral patterns such as dietary deviations. By constructing this profile, each user is provided with a foundation for intervention that is differentiated from other patients and aligns with their actual physical condition and long-term control goals.

[0056] S102: Generate an exercise plan and a diet plan based on the user's personalized portrait according to a set period.

[0057] In principle, based on the user's personalized portrait, an exercise and diet coordination plan that meets individual needs is generated on a daily, weekly or other set cycle basis. Specifically, while maintaining a balance between energy intake and expenditure, blood sugar fluctuations are stabilized to the maximum extent possible and patient compliance is improved.

[0058] Among them, the formulation of the exercise plan takes into account the individual's metabolic capacity, previous exercise response patterns, insulin sensitivity, and related complications such as knee joints, and automatically matches the recommended time period, exercise type, and duration in combination with their lifestyle rhythm. For example, for a young or middle-aged patient with a relatively stable blood sugar response and no obvious complications, it can be recommended to do 30 minutes of moderate-intensity brisk walking in the morning and evening each day; for elderly patients with limited mobility, it is more likely to generate low-intensity stretching and indoor gymnastics recommendations spread throughout the day. The diet plan is based on the dual goals of nutritional balance and blood sugar control. It is recommended according to the model of three meals a day and optional snacks. The calories, carbohydrate content, glycemic index (GI value), fiber and protein ratio of the food are comprehensively calculated. Combined with the user's preferences, taboos and regional availability, a highly executable daily diet menu is constructed.

[0059] S103. In response to the acquired real-time user data, the exercise plan and diet plan are dynamically modified to obtain an exercise-diet collaborative plan; the real-time user data includes blood sugar data, dietary intake data, and exercise data; the exercise-diet collaborative plan includes exercise events and diet arrangements arranged based on time points.

[0060] Indicatively, since users inevitably have plan deviations in real life, such as impromptu dinners, exercise interruptions, or overeating, the actual execution status is dynamically evaluated and subsequent plans are adjusted in a timely manner to achieve dynamic collaborative management of diet and exercise. Optionally, real-time user data can be collected through continuous access to wearable devices, blood glucose meters, and diet record interfaces. Among them, blood glucose data provides a direct basis for judging the current blood glucose control status; exercise data includes the type, intensity, time, and energy consumed of the exercise completed; and diet data usually includes the type, quantity, and time of food intake.

[0061] Specifically, the deviation between the current data and the recommended plan is calculated to identify whether there is obvious excessive intake, lack of exercise or abnormal blood sugar. If a deviation is found, the dynamic correction mechanism is immediately activated. For example, if the carbohydrate intake at dinner exceeds the standard, the user may be advised to do 15 minutes of light physical activity before going to bed; if the user has not exercised all day, the staple food portion will be automatically reduced for breakfast the next day, and the exercise period will be postponed to a time window when blood sugar is more stable.

[0062] Optionally, in response to a user searching for alternative foods in a diet plan, or image recognition of a dining scene or menu text, alternative foods and corresponding dosage options are provided based on food similarity features and GI adjustment relationships.

[0063] S104. Generate corresponding push instructions according to the time points; the push instructions are used to instruct to push corresponding exercise items or diet arrangements to the diabetic patient terminal.

[0064] In principle, based on the dependency of user self-execution and the uncertainty of environmental changes, a push mechanism based on time points is constructed to ensure that the intervention plan can be delivered to the user terminal in a timely and clear manner, and a visual and interactive support interface is provided. For example, for an exercise plan of taking a 15-minute brisk walk at 3 pm, a push notification “taking a 15-minute brisk walk at 3 pm” is sent to the user at 2:30 pm.

[0065] Push instructions can be transmitted to user terminals through multiple channels, such as smartphone APP, bracelet vibration reminder, voice assistant broadcast, etc., to adapt to the acceptance habits of different users.

[0066] Optionally, a user feedback mechanism is supported, allowing users to confirm whether they have completed the recommended tasks and explain the reasons if they have not done so, which can be further used to update the user behavior graph and personalized preference model.

[0067] The aforementioned method for collaborative management of diabetes exercise rehabilitation and diet, by acquiring individual characteristic data, physiological data, and daily behavior data from diabetic patients and constructing personalized user profiles, can comprehensively characterize the patient's physical type, lifestyle habits, and metabolic characteristics, facilitating personalized matching of subsequent exercise and diet plans, and improving the accuracy and adaptability of health interventions. Cyclic generation of exercise and diet plans based on personalized profiles ensures the overall logical consistency and sustainability of the health intervention plan within the cycle. It can also match plans to the patient's daily rhythm, reduce sudden changes, and improve compliance. By dynamically revising the original exercise and diet plans in response to real-time user data, it can effectively address behavioral deviations or sudden blood sugar fluctuations that occur during patient implementation, ensuring real-time intervention adjustments, thereby improving the safety and stability of overall blood sugar control. The exercise and diet collaborative plan is delivered to the patient's terminal device via push instructions, ensuring that the plan is accurately executed on time, helping to improve patient compliance with the rehabilitation plan and the actual implementation rate of the plan, further enhancing the overall intervention effect.

[0068] In one embodiment, generating an exercise plan and a diet plan based on a user's personalized profile and at a set period includes:

[0069] S21. Based on the exercise type-energy expenditure-blood glucose response library model, the impact of energy expenditure of different exercise combinations and daily behavior data on blood glucose is predicted according to the user's personalized portrait, and an exercise plan for a certain period is obtained; the exercise plan includes running type, intensity and duration.

[0070] In principle, based on the constructed user personalized profile, the user's health status characteristics, daily life rhythm, occupational habits, and metabolic characteristics in the current cycle are obtained. The blood sugar fluctuation response caused by different exercise methods is predicted by combining the exercise type-energy expenditure-blood sugar response library model. This library model is a ternary association graph model trained by combining clinical data and public metabolic models. The exercise types include aerobic exercise, resistance exercise, mixed interval exercise, etc. The energy expenditure dimension is based on the conversion of metabolic equivalents (METs) per unit time and individual basal metabolic rate (BMR), while the blood sugar response dimension records the speed and magnitude of blood sugar drop caused by the same exercise stimulus under different individual metabolic types. By inputting the user profile information as the initial condition into the graph model, combined with the user's previous behavior data and optional exercise combinations, multiple scheme simulations can be performed using algorithms such as simulated annealing, genetic programming, or reinforcement learning to predict the blood sugar regulation trend brought about by different exercise combinations, and finally screen out the optimal solution to generate a periodic exercise plan. The generated exercise plan includes the type of exercise for each day or stage, the intensity including the target heart rate zone or MET value, and the duration, and is marked by time period to ensure that the execution rhythm is consistent with lifestyle habits.

[0071] S22. Based on the food-nutrient-glucose response library model, a diet plan is obtained by using the user's personalized image and the expected exercise consumption corresponding to the exercise plan through an optimization algorithm; the diet plan includes the type of ingredients, the amount used, and the time period for meals; the food-nutrient-glucose response library model includes the food types and their corresponding GI values.

[0072] Schematically, the energy expenditure expectation and blood glucose fluctuation curve corresponding to the exercise plan are used to deduce the most appropriate diet arrangement strategy by calling the food-nutrient-blood glucose response library model. This model is built on a large number of nutrition databases and individual diet experiment data. The core association graph structure uses food type-nutrient-blood glucose response as a ternary path mapping. Food types cover a broad range of daily staple foods, side dishes, fruits, beverages, and alternative ingredients. The nutrient dimension records the values ​​of calories, protein, carbohydrates, fat, fiber, minerals, etc. The blood glucose response dimension includes the standard GI value (Glycemic Index) and the predicted response corrected by individual insulin sensitivity. Exemplarily, linear programming, Lagrange multiplier method, or multi-objective constrained optimization algorithm is used to ensure that the total calorie intake meets the energy expenditure compensation of exercise, while ensuring that the nutritional structure of each meal is reasonable and the blood glucose fluctuation between meals is stable. Some taboo foods such as high-sugar beverages and individual allergy categories are automatically blocked. Among them, the introduction of adaptive variables in the optimization objective function ensures that the output plan not only meets the macro indicators but also conforms to the user's dietary preferences and feasibility. The final diet plan clearly lists the recommended ingredients, recommended amounts and corresponding meal time periods for each meal. For example, the breakfast recommendation is: 30g oatmeal, 1 egg, 200ml low-fat milk, consumed before 8:00.

[0073] Furthermore, there is a two-way coupling between the exercise plan and the diet plan. The diet arrangement not only serves the purpose of energy replenishment and blood sugar regulation during exercise, but also affects the timing and type of subsequent exercise. For example, a diabetic patient has mild type II diabetes. The personalized portrait shows that he works in a sedentary office, has low insulin sensitivity, and has a tendency to have high blood sugar after breakfast. Based on this portrait, it is recommended to arrange a 45-minute resistance training at noon, and to consume a high-protein breakfast and lunch before and after. The diet plan recommends replacing traditional white bread with low-GI cereals for breakfast, and moderately increasing the fiber intake at lunch to slow down the sugar release rate, thereby matching the energy demand and blood sugar change rhythm during exercise.

[0074] In one embodiment, Figure 2 As shown, in response to the acquired real-time user data, the exercise plan and diet plan are dynamically modified to obtain an exercise and diet collaborative plan, including:

[0075] S201. Calculate movement behavior deviation based on movement data and movement plan to obtain a movement deviation value.

[0076] The patient's actual exercise activity during the current cycle is compared with the initially generated exercise plan. For example, actual exercise activity is recorded using motion sensors such as smart bracelets, mobile phone accelerometers, or exercise apps. Key data captured includes exercise start and end times, type, actual calories burned, duration, average heart rate, and GPS tracking. This data is then matched to the type, intensity, and duration of exercise in the intended exercise plan.

[0077] Optionally, during the deviation calculation process, a weighted difference scoring method is used to uniformly map multiple dimensions to a comprehensive movement deviation value (MPD, Movement Plan Deviation), where the deviation types include but are not limited to incomplete, insufficient intensity, shortened duration, misaligned execution time, etc. The higher the absolute value of the deviation value, the more the patient's exercise plan execution in the current cycle deviates from the expected goal.

[0078] S202. Calculate dietary behavior deviation based on dietary intake data and the dietary plan to obtain a dietary deviation value.

[0079] Indicatively, actual eating behaviors are obtained by users manually entering food intake information in a mobile terminal, parsing captured meal pictures through image recognition algorithms, or linking with smart plates and nutrition monitoring equipment. Based on the food-nutrient composition-blood sugar response library model, the food intake is converted into quantifiable nutritional intake values ​​such as total calories, carbohydrate intake, protein ratio, etc., and compared with the expected intake values ​​in the diet plan. Optionally, by constructing a dietary deviation assessment model, corresponding indicator systems are established for behaviors such as excessive intake, insufficient intake, missed meals, missed intake times, and food type mismatches, and finally a multi-dimensional weighted dietary deviation value (FPD, Food Plan Deviation) is calculated. This deviation value not only quantifies the overall deviation of intake behavior, but can also be traced back to specific meals and food dimensions.

[0080] S203. Based on the multi-objective planning reinforcement model, the exercise plan and the diet plan are modified according to the exercise deviation value and the diet deviation value to obtain an exercise-diet synergy plan.

[0081] Schematically, a multi-objective planning reinforcement model is used to jointly analyze the deviation values ​​of exercise and diet, and to optimize and modify the subsequent plan to generate a new exercise and diet synergy plan. Specifically, the model takes minimizing the risk of blood sugar fluctuations, maximizing the plan's executability, and ensuring nutritional balance as the joint objective function, and introduces various realistic conditions such as behavioral inertia constraints, time scheduling flexibility, and metabolic load limits as planning constraints. For example, when the exercise deviation value is large but the diet is relatively stable, it can be recommended to increase low-intensity exercise the next day or to schedule exercise the day before; when the dietary intake is significantly exceeded, the subsequent carbohydrate intake ratio may be automatically reduced or the protein ratio increased, while the duration of the next aerobic exercise session may be appropriately extended. Optionally, when a continuous deviation trend is detected, the plan intensity can be gradually reduced through a guidance mechanism to adapt to the user's behavioral habits, and then slowly adjusted back to the ideal state to improve compliance.

[0082] For example, a patient missed their midday resistance training session on a particular day, recording an exercise deviation of 0.72. They also recorded a total calorie intake at dinner that was 30% higher than planned, resulting in a diet deviation of 0.64. The reinforcement optimization model determined that the current blood sugar risk level might be elevated, and then generated the following exercise-diet synergy plan: It recommended delaying the next meal by half an hour and reducing staple food intake by 20%, or arranging a 45-minute moderate-intensity aerobic exercise session, with supplementary blood sugar-slowing foods provided before and after meals. This regulatory process is dynamically learned by the reinforcement model, which records behavioral responses to continuously train the strategy network and improve future response accuracy.

[0083] The above method realizes sensitive perception and responsive adjustment of actual behavioral deviations, transforming the entire diabetes management system from static plan generation to a dynamic closed-loop system with self-correction capabilities, thereby more effectively dealing with patients' plan deviation problems in complex life situations, and improving the stability, individual adaptability and health benefits of diabetes rehabilitation management.

[0084] In one embodiment, the motion deviation value is obtained by the following formula:

[0085] ΔE=E_expected-E_actual

[0086] Among them, ΔE is the exercise deviation value; E_expected is the planned exercise calorie consumption value; E_actual is the actual recorded calorie consumption value;

[0087] The dietary deviation value is obtained by the following formula:

[0088] ΔC=C_actual-C_expected

[0089] Among them, ΔC is the dietary deviation value; C_actual is the actual calorie intake; C_expected is the planned calorie intake.

[0090] In one embodiment, based on the multi-objective planning reinforcement model, the exercise plan and the diet plan are modified according to the exercise deviation value and the diet deviation value to obtain an exercise-diet synergy plan, including:

[0091] S31. If the movement deviation value exceeds the first positive threshold, based on the energy consumption upward adjustment target, an adjusted movement plan is obtained according to the multi-objective planning reinforcement model.

[0092] Indicatively, it is determined whether the movement deviation value in the current cycle exceeds the set first positive threshold. If the movement deviation value exceeds the first positive threshold, it indicates that the user has obvious lack of exercise behavior or quality decline. Activate the intervention strategy based on the increase in energy consumption, and strengthen the model to increase the metabolic load as the optimization goal. Under the premise of maintaining a stable rhythm of life for the user, generate an adjusted exercise plan by adjusting the type of exercise, extending the duration or increasing the intensity. Optionally, in the multi-objective reinforcement model, the strategy constraints include not increasing nighttime exercise, avoiding continuous high-intensity training, and maintaining a reasonable interval with the time of meal intake, etc., to ensure that the executability of the plan is coordinated with the goal of stable blood sugar control.

[0093] S32: If the movement deviation value is less than the first negative threshold, based on increasing the diet arrangement goal, an adjusted diet plan is obtained according to the multi-objective planning reinforcement model.

[0094] Indicatively, if the movement deviation value is lower than the first negative threshold, it means that the user may have exceeded the target or overtrained, and it is necessary to avoid problems such as hypoglycemia caused by excessive energy consumption or excessive exercise. The goal based on the increase in dietary arrangements is activated, and the reinforcement model is guided to compensate for the nutritional supply and balance the energy level as the optimization direction, and a corresponding diet correction plan is generated. Specifically, the calorie supply of the main meal is moderately increased, the carbohydrate-water ratio is increased, or high-GI food is arranged as a snack, such as adding bananas, bread and other snacks to quickly replenish muscle glycogen. The optimization process of this strategy is also controlled by the food-blood sugar response model and the patient's insulin sensitivity parameters to ensure that the adjusted intake arrangement is still within the safe control range.

[0095] S33. If the dietary deviation value exceeds the second positive threshold, an adjusted exercise plan is obtained based on adding a short-term light exercise goal and according to the multi-objective planning reinforcement model.

[0096] Determine whether the dietary deviation value exceeds the set second positive threshold, indicating that the current cycle's calorie or carbohydrate intake is significantly exceeded. To reduce the risk of post-meal blood sugar surges, the short-term light exercise adjustment goal is activated, guiding users to perform light, controllable, low-load exercise interventions within a short period of time to improve their immediate glucose metabolism capacity. Under this goal, the reinforcement model typically recommends activities such as 15-20 minutes of low-intensity walking, stretching, or dynamic yoga, starting 10 minutes after a meal, and automatically adjusts the timing of these activities in relation to meals to optimize blood sugar control and avoid large insulin fluctuations and delayed hyperglycemia.

[0097] S34. If the diet deviation value is less than the second negative threshold, an adjusted diet plan is obtained according to the multi-objective planning reinforcement model based on the low GI food intake goal.

[0098] If the dietary deviation value falls below the second negative threshold, it indicates insufficient intake, potentially accompanied by a lack of important nutrients. This activates the low-GI food intake strategy, guiding the patient to supplement subsequent meals or snacks with slow-release low-GI foods, such as oats, sweet potatoes, chickpeas, or low-fat dairy products. This compensates for the carbon source needed for blood sugar stability and reduces the risk of hypoglycemia or nighttime blood sugar fluctuations due to insufficient intake. The objective function optimized by the reinforcement model in this strategy is to improve the ability to release glucose steadily after a meal, control total calorie intake, and take into account subjective preference variables such as dietary acceptability.

[0099] The above method realizes the linkage mapping of behavioral deviation and target adjustment, so that the exercise and diet plan has the perception-response-intervention closed-loop capability, improves the accuracy and adaptability of the intervention process, and avoids the physiological burden or execution barriers caused by blind adjustment, providing a more realistic compliance solution for diabetes rehabilitation.

[0100] In one embodiment, Figure 3 As shown, the method includes:

[0101] S301. Record the user's real-time data historically acquired within a preset time period to obtain diet records, exercise records, and blood sugar records.

[0102] For example, a pre-set time window, such as the past 30 days, is selected to automatically retrieve real-time data collected from the user's terminal or access device during that period. This includes dietary intake records (e.g., food type, nutrient composition, and time of intake); exercise records (e.g., type, duration, intensity, start and end times); and blood sugar profiles. By standardizing and timestamping these records, an hourly data timeline is constructed.

[0103] S302. Construct a diet-exercise-blood sugar three-dimensional dynamic graph by combining diet records, exercise records, and blood sugar records in a time series in the form of triples; the diet-exercise-blood sugar three-dimensional dynamic graph includes food nodes, exercise nodes, blood sugar status nodes, and edge relationships between the nodes.

[0104] Schematically, the data is organized into a three-dimensional dynamic graph of diet-exercise-blood sugar in the form of triples (head, relation, tail). In this graph, food intake behavior, exercise behavior and blood sugar status are abstracted into different types of nodes, respectively called food nodes, exercise nodes and blood sugar status nodes. The edge relationship between the nodes represents the semantic association between different behaviors or states. For example, eating rice for breakfast, taking a 30-minute walk after the meal, and blood sugar dropping by 1.2 mmol / L can be represented as a path from the rice node, through the walking node to the blood sugar drop node. The edge relationship can be annotated with attributes such as time delay and metabolic impact direction. Furthermore, the dynamic graph also has the ability to dynamically evolve the time layer, and can update the behavior chains of different time periods on a daily or weekly basis, so that the graph has scalability and plasticity over time.

[0105] S303. Use a graph convolutional network based on a graph path attention mechanism to perform path learning on the diet-exercise-blood sugar three-dimensional dynamic graph to obtain path constraints.

[0106] Schematically, by constructing a graph convolutional neural network (GCN) model combined with a graph path attention mechanism (Path Attention Mechanism), path-level feature extraction is performed on the three-dimensional dynamic graph. This model not only uses the local feature propagation method of adjacent nodes in the traditional GCN structure to obtain contextual information, but also introduces weight modeling of historical paths, that is, automatically identifying which paths in the diet-exercise-blood sugar path appear more frequently, are more closely related to blood sugar control, or have a more stable impact. For example, in a user's historical data, it is identified that the path of 20 minutes of walking for 15 minutes after a meal appears many times before the blood sugar stable node, so the weight of this path is significantly improved, becoming an important constraint for subsequent recommended behaviors. The path attention mechanism automatically adjusts the contribution of each path in the graph through the backpropagation algorithm, so that the model has the ability to perceive the key nodes and causal logic in complex behavior chains.

[0107] S304: Constraining the generation strategies corresponding to the exercise plan and the diet plan according to the path constraints.

[0108] The path constraint rules generated by the above-mentioned path learning results are used to perform soft constraints or structural modifications on the generation strategies of exercise plans and diet plans. Specifically, before generating the diet or exercise schedule for the next cycle, the high-weight causal path in the graph path will be called as the priority structure for the recommended behavior combination. For example, if the model identifies the path of breakfast, jogging, and stable blood sugar as a high-weight path, then this combination will be given priority in the breakfast arrangement or morning exercise generation strategy; if a path shows a negative trend of high-fat dinner, no exercise, and high blood sugar at night, then this combination will be avoided in subsequent plans, thereby achieving experience guidance and risk avoidance at the recommendation level.

[0109] The above method, through the introduction of the graph constraint mechanism, establishes a behavioral regulation pathway of structural backtracking-causal extraction-path guidance, providing a strategic basis for exercise and diet recommendations based on the real behavioral patterns of users, effectively improving the feasibility, effectiveness and individual adaptability of the output plan, and avoiding the problems of isolated planning, short-term optimization and long-term imbalance.

[0110] In one embodiment, the method further comprises:

[0111] S41. Marking a graph path that triggers a high blood sugar or low blood sugar trend in the diet-exercise-blood sugar three-dimensional dynamic graph as a risk path based on the blood sugar status node.

[0112] Schematically, the paths that appear in the constructed three-dimensional dynamic map of diet-exercise-blood sugar are marked as risky. Specifically, the changing trends of blood sugar status nodes in historical data are used to identify which paths often cause blood sugar levels to rise or fall rapidly after a behavior. These paths are assigned attribute labels of high blood sugar risk or low blood sugar risk in the map and included in the risk path database. Path identification not only considers the correspondence between a single behavior and blood sugar, but also covers trend evolution patterns under multi-step paths and multi-factor combinations, thereby improving the coverage of complex risk chains.

[0113] S42. Identify whether there are abnormal situations in real-time user data that meet the risk path through graph anomaly detection.

[0114] Schematically, real-time user behavior data is mapped to a graph structure, and a graph anomaly detection algorithm is used to identify whether the current state is close to or has already matched a risk path. For example, graph anomaly detection can use a comparison mechanism based on graph embedding vector space distances, or a path matching model to determine whether the current node sequence is approaching a high-risk path. If the degree of overlap between the behavior chain formed by the real-time data and a historically identified high-risk path within a certain time window exceeds a threshold, it is determined to be a potential anomaly risk.

[0115] S43. If an abnormal situation occurs, a warning instruction is triggered; the warning instruction is used to instruct the corresponding warning information to be pushed to the diabetic patient terminal; the warning information includes hypoglycemia warning and hyperglycemia warning.

[0116] If an abnormal situation occurs, an early warning instruction is immediately generated and triggered. This instruction instructs the corresponding risk reminder information to be pushed to the user terminal, such as a smartwatch, mobile phone app, or portable medical reminder device. The push content is divided into two categories based on the risk type: hypoglycemia warning and hyperglycemia warning. The information not only includes the current risk category but also behavioral suggestions. For example, the current behavior may cause hyperglycemia, please do light exercise, or if a hypoglycemia trend is detected, please supplement with glycemic-raising foods.

[0117] The above method realizes the static prediction and dynamic correction of individual behavior, and also introduces risk modeling and real-time abnormal response functions into the graph structure, greatly improving the practicality and safety of the system in diabetes management scenarios.

[0118] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0119] Based on the same inventive concept, the present application also provides a diabetes exercise rehabilitation and diet collaborative management device for implementing the aforementioned diabetes exercise rehabilitation and diet collaborative management method. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more diabetes exercise rehabilitation and diet collaborative management device embodiments provided below can be found in the above-mentioned limitations of the diabetes exercise rehabilitation and diet collaborative management method, and will not be repeated here.

[0120] In an exemplary embodiment, Figure 4 As shown, a diabetes exercise rehabilitation and diet collaborative management device is provided, comprising:

[0121] Personalization module 401 is used to obtain individual characteristic data, physiological data and daily behavior data of diabetic patients and build a personalized user portrait;

[0122] Planning module 402, for generating exercise plans and diet plans according to set periods based on the user's personalized profile;

[0123] A dynamic adjustment module 403 is used to dynamically modify the exercise plan and diet plan in response to the acquired real-time user data to obtain an exercise-diet collaborative plan;

[0124] The interaction module 404 is used to generate corresponding push instructions according to time points; the push instructions are used to instruct to push corresponding exercise items or diet arrangements to the diabetic patient terminal.

[0125] In one embodiment, the planning module 402 is further configured to predict the effects of energy expenditure on blood glucose of different exercise combinations and daily behavior data based on the exercise type-energy expenditure-blood glucose response library model according to the user's personalized profile, and obtain an exercise plan for a certain period;

[0126] The planning module 402 is further configured to obtain a diet plan based on the food-nutrient component-blood glucose response library model by using an optimization algorithm and the expected exercise consumption corresponding to the user's personalized image and exercise plan.

[0127] In one embodiment, it further includes:

[0128] A motion deviation module is used to calculate the motion behavior deviation based on the motion data and the motion plan to obtain the motion deviation value;

[0129] A dietary deviation module is used to calculate dietary behavior deviation based on dietary intake data and dietary plan to obtain a dietary deviation value;

[0130] The dynamic adjustment module 403 is further configured to modify the exercise plan and the diet plan according to the exercise deviation value and the diet deviation value based on the multi-objective planning reinforcement model to obtain an exercise-diet synergy plan.

[0131] In one embodiment, it further includes:

[0132] The history record module records the user's real-time data acquired in the preset time period, and obtains diet records, exercise records and blood sugar records;

[0133] A dynamic graph module is used to construct a three-dimensional dynamic graph of diet, exercise and blood sugar by combining diet records, exercise records and blood sugar records in a time series in the form of triplets;

[0134] The path constraint module is used to use a graph convolutional network based on the graph path attention mechanism to perform path learning on the diet-exercise-blood sugar three-dimensional dynamic graph to obtain path constraints;

[0135] The strategy adjustment module is used to constrain the generation strategies corresponding to the exercise plan and diet plan according to the path constraints.

[0136] In one embodiment, it further includes:

[0137] A risk marking module is used to mark the graph path that causes high blood sugar or low blood sugar trend in the diet-exercise-blood sugar three-dimensional dynamic graph as a risk path based on the blood sugar status node;

[0138] Similarity recognition module, used to identify whether there are anomalies in real-time user data that meet risk paths through graph anomaly detection;

[0139] The early warning module is used to trigger an early warning instruction if an abnormal situation occurs; the early warning instruction is used to instruct the corresponding early warning information to be pushed to the diabetic patient terminal.

[0140] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0141] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0142] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial description of the method embodiments. The device embodiments described above are merely illustrative, wherein the components described as separate parts may or may not be physically separated, and the parts displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the disclosed solution. A person of ordinary skill in the art can understand and implement it without expending creative work.

[0143] The above-described embodiments merely represent several implementation methods of the embodiments of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that a person skilled in the art may make various modifications and improvements without departing from the concept of the embodiments of the present application, and these modifications and improvements fall within the scope of protection of the embodiments of the present application.

Claims

1. A method for coordinated management of diabetes exercise rehabilitation and diet, characterized in that: The method comprises: Obtain individual characteristic data, physiological data, and daily behavior data of diabetic patients and build personalized user portraits; Generate an exercise plan and a diet plan based on the user's personalized profile according to a set period; In response to the acquired real-time user data, the exercise plan and diet plan are dynamically modified to obtain an exercise-diet collaborative plan; the real-time user data includes blood sugar data, dietary intake data, and exercise data; the exercise-diet collaborative plan includes exercise activities and diet arrangements scheduled based on time points; A corresponding push instruction is generated according to the time point; the push instruction is used to instruct to push the corresponding exercise event or diet arrangement to the diabetic patient terminal.

2. The method according to claim 1, characterized in that Generating an exercise plan and a diet plan based on the user's personalized portrait according to a set period includes: Based on the exercise type-energy expenditure-blood glucose response library model, the effects of different exercise combinations and the energy expenditure of the daily behavior data on blood glucose are predicted according to the user's personalized profile, and the exercise plan for a certain period is obtained; the exercise plan includes the type of exercise, intensity, and duration; Based on the food-nutrient component-blood glucose response library model, the diet plan is obtained by an optimization algorithm using the user's personalized image and the expected exercise consumption corresponding to the exercise plan; the diet plan includes the types of ingredients, amounts and meal time periods; the food-nutrient component-blood glucose response library model includes food types and their corresponding GI values.

3. The method according to claim 1, characterized in that The step of dynamically revising the exercise plan and diet plan in response to the acquired real-time user data to obtain an exercise-diet collaborative plan includes: Calculating the movement behavior deviation according to the movement data and the movement plan to obtain a movement deviation value; Calculating dietary behavior deviation based on the dietary intake data and the dietary plan to obtain a dietary deviation value; Based on the multi-objective planning reinforcement model, the exercise plan and the diet plan are modified according to the exercise deviation value and the diet deviation value to obtain an exercise and diet coordination plan.

4. The method according to claim 3, wherein: The motion deviation value is obtained by the following formula: ΔE=E_expected-E_actual Wherein, ΔE is the exercise deviation value; E_expected is the planned exercise calorie consumption value; E_actual is the actual recorded calorie consumption value; The dietary deviation value is obtained by the following formula: ΔC=C_actual-C_expected Wherein, ΔC is the dietary deviation value; C_actual is the actual calorie intake; and C_expected is the planned calorie intake.

5. The method according to claim 4, characterized in that The multi-objective planning reinforcement model modifies the exercise plan and the diet plan according to the exercise deviation value and the diet deviation value to obtain an exercise-diet synergy plan, including: If the movement deviation value exceeds a first positive threshold, based on the energy consumption upward adjustment target, an adjusted movement plan is obtained according to the multi-objective planning reinforcement model; If the movement deviation value is less than a first negative threshold, based on increasing the dietary arrangement goal, obtaining an adjusted dietary plan according to the multi-objective planning reinforcement model; If the dietary deviation value exceeds a second positive threshold, an adjusted exercise plan is obtained according to a multi-objective planning reinforcement model based on adding a short-term light exercise goal; If the diet deviation value is less than the second negative threshold, an adjusted diet plan is obtained according to the multi-objective planning reinforcement model based on the low GI food intake goal.

6. The method according to any one of claims 1 to 5, characterized in that The method comprises: Record the user's real-time data acquired historically within a preset time period to obtain diet records, exercise records, and blood sugar records; The diet records, exercise records, and blood sugar records are combined in a triplet form in a time series to construct a diet-exercise-blood sugar three-dimensional dynamic graph; the diet-exercise-blood sugar three-dimensional dynamic graph includes food nodes, exercise nodes, blood sugar status nodes, and edge relationships between the nodes; Utilizing a graph convolutional network based on a graph path attention mechanism to perform path learning on the diet-exercise-blood sugar three-dimensional dynamic graph to obtain path constraints; The generation strategies corresponding to the exercise plan and the diet plan are constrained according to the path constraint.

7. The method according to claim 6, characterized in that The method further comprises: Marking a graph path that triggers a high blood sugar or low blood sugar trend in the diet-exercise-blood sugar three-dimensional dynamic graph as a risk path according to the blood sugar status node; Identify whether the real-time user data has any anomalies that conform to the risk path through graph anomaly detection; If the abnormal situation exists, a warning instruction is triggered; the warning instruction is used to instruct to push corresponding warning information to the diabetic patient terminal; the warning information includes hypoglycemia warning and hyperglycemia warning.

8. A diabetes exercise rehabilitation and diet collaborative management device, characterized in that: The device comprises: Personalization module, used to obtain individual characteristic data, physiological data, and daily behavior data of diabetic patients and build personalized user portraits; A planning module, configured to generate an exercise plan and a diet plan based on a set period based on the user's personalized profile; A dynamic adjustment module, configured to dynamically modify the exercise plan and diet plan in response to the acquired real-time user data to obtain an exercise-diet collaborative plan; The interactive module is used to generate corresponding push instructions according to time points; the push instructions are used to instruct to push the corresponding exercise items or diet arrangements to the diabetic patient terminal.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images