Dynamic tracking, intervention method and system for bone health based on exercise and nutrition

Abstract

Embodiments of the present application disclose a method and system for dynamic tracking and intervention of bone health based on exercise and nutrition. The method comprises: collecting gait information, bone metabolite information and bone state measurement parameters of a target user using a wearable device; processing the gait information, bone metabolite information and bone state measurement parameters based on an artificial intelligence (AI) model to obtain bone state prediction parameters; performing bone health state assessment of the target user based on the bone state prediction parameters and health data of the target user to obtain health state prediction parameters; the health state prediction parameters include bone health state real-time evaluation indicators and bone health dynamic trend data; generating a targeted exercise plan and a diet plan based on the health state prediction parameters; the targeted exercise plan is used to guide the exercise of the target user; and the diet plan is used to guide the diet of the target user.

Description

Technical Field

[0001] This invention relates to the field of big health technology, and in particular to a method and system for dynamic tracking and intervention of bone health based on exercise and nutrition. Background Technology

[0002] The field of bone health management technology currently relies primarily on static testing, single-dimensional assessment, and standardized intervention programs. Traditional bone health management systems are centered on bone mineral density (BMD) testing, combined with empirical exercise and nutritional planning, forming a basic process of testing-assessment-intervention. For example, traditional BMD testing suffers from a lag: it is typically performed only 1-2 times per year, making it difficult to reflect the dynamic impact of daily exercise and nutrition on bone health in a timely manner. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide a method and system for dynamic tracking and intervention of bone health based on exercise and nutrition. The technical solution of the present invention is implemented as follows:

[0004] The first aspect of this disclosure provides a method for dynamic tracking and intervention of bone health based on exercise and nutrition, comprising: collecting gait information, bone metabolite information, and bone status measurement parameters of a target user using a wearable device; processing the gait information, bone metabolite information, and bone status measurement parameters based on an artificial intelligence (AI) model to obtain bone status prediction parameters; and outputting a prompt message when it is determined that the target user has a risk of falling based on the assessment indicators and the target user's baseline health data; the prompt message includes: a fall risk value, the movement that caused the fall, and measures to prevent the fall risk.

[0005] A targeted plan is generated based on the health status prediction parameters; the targeted plan includes: exercise planning and diet planning; wherein, the targeted exercise plan is used to guide the exercise of the target user; the diet plan is used to guide the diet of the target user; the targeted plan includes a long-term plan to counteract the long-term bone health risks indicated by the bone health dynamic trend data, and a short-term plan that is consistent with the goals of the long-term plan and avoids the fall risk corresponding to the immediate assessment indicators.

[0006] A second aspect of this disclosure provides a dynamic tracking and intervention system for bone health based on exercise and nutrition, comprising: a data acquisition module for acquiring gait information, bone metabolite information, and bone status measurement parameters of a target user using a wearable device; a prediction module for processing the gait information, bone metabolite information, and bone status measurement parameters based on an artificial intelligence (AI) model to obtain bone status prediction parameters; an evaluation module for evaluating the bone health status of the target user based on the bone status prediction parameters and the target user's health data to obtain health status prediction parameters; the health status prediction parameters include: real-time bone health status evaluation indicators and dynamic trend data of bone health; and a prompting module for prompting based on... The system outputs a warning message when the assessment indicators and the target user's baseline health data determine that the target user is at risk of falling. The warning message includes: a fall risk value, the movement that caused the fall, and measures to mitigate the fall risk. A planning module is used to generate a targeted plan based on the health status prediction parameters. The targeted plan includes: an exercise plan and a dietary plan. The targeted exercise plan guides the target user's exercise; the dietary plan guides the target user's diet. The targeted plan includes a long-term plan to counteract the long-term bone health risks indicated by the bone health dynamic trend data, and a short-term plan consistent with the long-term plan and avoiding the fall risk corresponding to the immediate assessment indicators.

[0007] A third aspect of this disclosure provides a computer-readable medium having instructions stored thereon that, when executed by one or more processors, cause the processors to perform the method for dynamic tracking and intervention of bone health based on exercise and nutrition as described in any of the foregoing technical solutions.

[0008] This method for dynamic tracking and intervention of bone health based on exercise and nutrition utilizes wearable devices to collect gait information, bone metabolite information, and bone status measurement parameters in real time, enabling comprehensive, dynamic, and accurate monitoring of users' bone health data. By leveraging artificial intelligence (AI) models to deeply analyze and process this data, precise bone status prediction parameters are obtained. Combined with user health data, this allows for a scientific assessment of bone health status, deriving immediate evaluation indicators and dynamic trend data. Based on this, targeted exercise and dietary plans are generated, tailored to individual differences, effectively guiding users' exercise and diet, achieving proactive prevention and precise intervention, improving the efficiency of bone health management, reducing the risk of bone diseases, and comprehensively protecting users' bone health. Attached Figure Description

[0009] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0010] Figure 1 A flowchart illustrating a method for dynamic tracking and intervention of bone health based on exercise and nutrition, provided in an embodiment of the present invention;

[0011] Figure 2 A flowchart illustrating another method for dynamic tracking and intervention of bone health based on exercise and nutrition, provided in an embodiment of the present invention;

[0012] Figure 3 A schematic diagram of a dynamic tracking and intervention system for bone health based on exercise and nutrition, provided as an embodiment of the present invention;

[0013] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0014] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. 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 should fall within the scope of protection of the present invention.

[0015] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0016] like Figure 1 As shown, this disclosure provides a method for dynamic tracking and intervention of bone health based on exercise and nutrition, including:

[0017] S1110: Uses wearable devices to collect gait information, bone metabolite information, and bone status measurement parameters of the target user;

[0018] S1120: Based on an artificial intelligence (AI) model, process the gait information, bone metabolite information, and bone status measurement parameters to obtain bone status prediction parameters;

[0019] S1130: Based on the bone status prediction parameters and the target user's health data, assess the target user's skeletal health status to obtain health status prediction parameters; the health status prediction parameters include: real-time skeletal health status assessment indicators and dynamic trend data of skeletal health.

[0020] S1140: When it is determined that the target user is at risk of falling based on the assessment indicators and the target user's baseline health data, a prompt message is output; the prompt message includes: fall risk value, the movement that caused the fall, and measures to prevent the fall risk;

[0021] S1150: Generate a targeted plan based on the health status prediction parameters; the targeted plan includes: exercise planning and diet planning; wherein, the targeted exercise plan is used to guide the exercise of the target user; the diet plan is used to guide the diet of the target user; the targeted plan includes a long-term plan to counteract the long-term bone health risks indicated by the bone health dynamic trend data, and a short-term plan that is consistent with the goals of the long-term plan and avoids the fall risk corresponding to the immediate assessment indicators.

[0022] This method for dynamically tracking and intervening in bone health based on exercise and nutrition can be used on various electronic devices. These devices can connect to big data service platforms such as health monitoring platforms and interact with the platform's servers.

[0023] For example, this method for dynamic tracking and intervention of bone health based on exercise and nutrition can be used in one or more electronic devices. For example, the method can be used in mobile devices, which may be handheld devices and / or wearable devices programmed for the target user, specifically including but not limited to mobile phones, wristbands, ankle bracelets, etc. In other embodiments, this method for dynamic tracking and intervention of bone health based on exercise and nutrition can also be used in medical assistive devices. For example, the medical assistive device may be a health status monitoring device for the target user in a hospital or clinic. The owner of the electronic device can be the target user themselves, or a relative, friend, or guardian of the target user.

[0024] The electronic device runs an application or app for health monitoring. In some embodiments, the application or app can monitor multiple health indicators for a health goal. For example, it can monitor the target user's bone health, muscle health, etc.

[0025] In some embodiments, the target user may be a target user or potential target user with various diseases caused by skeletal or muscular abnormalities, such as the elderly, infants learning to walk, or patients with skeletal or muscular diseases. Of course, in specific implementations, the target user can be any object, not limited to target users or potential users with skeletal or muscular abnormalities, and can also be any healthy person who is concerned about their blood sugar level.

[0026] In some embodiments, the target user includes, but is not limited to, people, and may even be pets or animals in a zoo.

[0027] In S1110, wearable devices and invasive or non-invasive detection technologies are used to collect multi-dimensional data. For example, wearable devices, such as portable smart devices like smart bracelets and smart insoles, can collect data from the user's body surface in real time. Other technologies that acquire physiological information without invasive procedures, such as near-infrared spectroscopy and sweat sensor technology, utilize principles like optics and bioelectricity to obtain physiological information.

[0028] In some embodiments, gait information may be any information of the target user when walking, including but not limited to at least one of the following: stride length, cadence, plantar pressure distribution during a single step, plantar pressure changes during continuous stepping, pressure comparison parameters of both feet, and other motion data.

[0029] Bone condition measurement parameters can be various parameters that actually reflect the condition of the bones, especially those related to the bones of the feet and / or legs. For example, bone mineral density (BMD) estimated using bioelectrical impedance analysis.

[0030] Users wear wearable devices, where accelerometers and pressure sensors collect gait information; at the same time, a near-infrared spectroscopy module illuminates the wrist to monitor the concentration of substances related to bone metabolism in the blood, and a sweat sensor detects trace amounts of calcium and phosphorus metabolites; the data is synchronously transmitted to a cloud server.

[0031] The integration of multiple technologies enables non-invasive, real-time, and multi-dimensional data collection, broadening the sources of bone health data, reducing the testing burden on users, and providing a rich and reliable data foundation for subsequent analysis.

[0032] In some embodiments, the use of wearable devices to collect gait information and bone metabolite information of the target user includes at least one of the following:

[0033] The gait information is collected using foot pressure-sensitive shoes or smart ankle bracelets;

[0034] Bone structure parameters were measured using a home ultrasound bone densitometer wristband.

[0035] Muscle parameters were obtained by measuring muscle condition using bioelectrical impedance analysis.

[0036] The bone metabolite information is collected using a bandage-type sensor; wherein, the bandage sensor can also be used to collect muscle movement information at its location; the muscle movement information is also used to predict the bone state prediction parameters.

[0037] For example, foot pressure sensing shoes or smart ankle bracelets can be used to collect gait information: Foot pressure sensing shoes have a built-in array of pressure sensors. When a user walks, the sensors can detect data such as pressure distribution and force application time in different areas of the sole in real time, and then calculate information such as stride length, stride frequency, and gait cycle; Smart ankle bracelets are equipped with accelerometers and gyroscopes. By capturing the acceleration and angular velocity of ankle movement, they can analyze the swing amplitude and rhythm changes during walking, accurately obtain gait characteristics, and provide motion data support for assessing the stress state of bones.

[0038] For example, a home-use ultrasonic bone densitometer bracelet is used to measure bone structure parameters: This bracelet integrates an ultrasonic bone density detection module, utilizing the characteristics of ultrasound waves propagating through bones, such as speed and attenuation, to non-invasively measure parameters such as bone density and bone sound velocity. By emitting ultrasound waves that penetrate the skin and muscles to reach the bones, and receiving and analyzing the reflected signals, it quickly assesses the degree of bone mineralization and structural strength, helping to determine the basic state of bone health.

[0039] For example, bioelectrical impedance analysis (BIA) is used to measure muscle condition and obtain skeletal-muscle parameters: the device injects a weak, safe current into the body through electrodes on the skin surface, and measures the impedance value of the current within the body based on the differences in conductivity of different tissues (muscle, fat, bone, etc.). Based on a specific algorithm, information such as muscle content, muscle mass, and skeletal-muscle ratio (muscle mass / bone density) is extracted from the impedance data. Combined with the mechanical relationship between bones and muscles, skeletal-muscle synergy parameters are calculated, providing a basis for analyzing the stress environment of bones.

[0040] For example, a bandage-type sensor is used to collect information on bone metabolites and muscle movement: the bandage-type sensor is equipped with a microfluidic chip and a biosensor, which can adsorb sweat from the skin surface and perform biochemical analysis on trace amounts of calcium and phosphorus metabolites in the sweat to obtain information on bone metabolites; at the same time, its built-in strain sensor can sense the stretching and deformation of muscles in the area covered by the bandage, and collect information such as the frequency and amplitude of muscle movement. This data is fused with other parameters to help AI models predict bone status more comprehensively.

[0041] In S1120, data is processed based on an artificial intelligence (AI) model to obtain bone condition prediction parameters. The AI ​​model is a deep learning model, such as a convolutional neural network, trained on a large amount of bone health data. In some embodiments, gait information, non-invasively detected bone metabolites, and bone condition measurement parameters are input into the AI ​​model. The model automatically extracts data features, analyzes the correlations between data using a trained algorithm, predicts the changing trends of bone strength, density, etc., over a future period, and outputs bone condition prediction parameters.

[0042] For example, the AI ​​model structure can be a hybrid structure combining a multi-layer convolutional neural network (CNN) and a long short-term memory network (LSTM). The CNN part consists of convolutional layers and pooling layers, used to extract local features from the data; the LSTM layer is responsible for processing data with time-series characteristics and mining long-term dependencies between data; the final connection is a fully connected layer and an output layer, outputting the final prediction result. The processing of the AI ​​model structure can include: data preprocessing and input: gait information, bone metabolite information, and bone state measurement parameters are organized into a multi-dimensional data matrix in a unified format, and then input into the model after standardization. For example, gait data such as stride length and gait frequency are arranged into a matrix according to time series. CNN feature extraction: the data enters the CNN layer, the convolutional layer scans the data through convolutional kernels to extract local features such as gait patterns and metabolite change trends; the pooling layer reduces the dimensionality of the features, retains key information, and reduces the amount of computation. LSTM Temporal Analysis: Feature data processed by CNN is input into LSTM layers. LSTM selectively retains or discards historical information through forget gates, input gates, and output gates, mining the correlation of data in the time dimension and analyzing the changing patterns of bone health status over time. Fully Connected Layers and Output: The feature vector output by LSTM is integrated by fully connected layers and transformed into values ​​that conform to the range of bone status prediction parameters through activation functions. Finally, the output layer outputs bone status prediction parameters such as bone strength and density for a future period of time.

[0043] AI models can efficiently process complex, multi-source data, accurately uncover hidden information in the data, and achieve forward-looking predictions of bone status, providing strong support for early warning of bone health risks.

[0044] In S1130, bone health status is assessed by combining multi-source data, and health status prediction parameters are generated.

[0045] For example, health status prediction parameters can cover basic information such as age, gender, and medical history; however, they are not limited to these, and more importantly, include the current health level and future trend curve. By integrating bone status prediction parameters with user health data, and based on preset medical assessment algorithms and standards, the user's bone health is quantitatively scored and trend analyzed to generate health status prediction parameters, intuitively presenting the current health level and future trends.

[0046] It integrates multi-dimensional data for in-depth assessment, providing comprehensive and accurate bone health prediction results to help users and medical personnel clearly understand their bone health status and facilitate the development of personalized management plans.

[0047] In some embodiments, the bone health status of the target user is assessed based on the bone status prediction parameters and the target user's health data to obtain health status prediction parameters, including: using a three-dimensional AI fracture risk model to obtain the fracture risk index of the target user and the fracture change trend with a preset period as the time unit based on the bone structure index, mechanical load index, and fall risk index.

[0048] In some embodiments, the system employs a 3D AI fracture risk model, deeply integrating bone condition prediction parameters with target user health data to accurately assess bone health status. This AI model, built on deep learning algorithms and trained and optimized through massive amounts of clinical cases and bone health data, can accurately identify the correlations between complex data.

[0049] During the assessment, the AI ​​model first extracts bone structure indicators, such as bone density and trabecular bone structure. These data are obtained through home wearable bone densitometers, medical imaging, and other methods, reflecting the basic strength and structural stability of the bones. The mechanical load indicators are obtained from gait information and muscle movement data collected by wearable devices, which are used to analyze the pressure and stress distribution on the bones during daily activities. The fall risk indicator is determined by combining health data such as the user's age, balance ability test data, and historical fall records.

[0050] The AI ​​model takes the aforementioned indicators as input, extracts and analyzes features through a multi-layered neural network, and uses complex algorithms to calculate the fracture risk index for the target user, presenting the current fracture risk level in a quantifiable numerical form. Simultaneously, using a preset period (e.g., monthly or quarterly) as the time unit, it analyzes the dynamic changes of various indicators and, combined with time series prediction algorithms, generates a fracture trend curve, clearly showing the rising and falling trends of fracture risk over a future period. This provides users and medical personnel with comprehensive and forward-looking bone health assessment results, enabling timely development of prevention and intervention strategies.

[0051] In some embodiments, a three-dimensional AI fracture risk model is used to obtain the fracture risk index of the target user and the fracture change trend over a preset period using the bone structure index, mechanical load index, and fall risk index. This includes: determining skeletal functional redundancy using the three-dimensional AI fracture risk model based on the muscle-to-bone ratio index, joint stability coefficient, and dynamic impact absorption rate; correcting the skeletal functional redundancy using the mechanical load index, fall risk index, and metabolic environment; and determining the fracture risk index of the target user and the fracture change trend over a preset period based on the skeletal functional redundancy.

[0052] The 3D AI-based risk model first uses multimodal data fusion technology to perform in-depth analysis of the muscle-to-bone ratio, joint stability coefficient, and dynamic shock absorption rate to obtain skeletal functional redundancy. For example, skeletal functional redundancy is an indicator that measures the reserve capacity of the skeleton to withstand daily activities and potential risks beyond basic physiological needs.

[0053] In some embodiments, the muscle-to-bone ratio, calculated from muscle mass measured by bioelectrical impedance analysis and ultrasound bone mineral density data, reflects the synergistic mechanical relationship between muscle and bone. AI models use convolutional neural networks (CNNs) to analyze the impact of this ratio on bone stress distribution; for example, a low muscle-to-bone ratio may lead to higher loads on the bones.

[0054] In some embodiments, the joint stability coefficient is determined by using a long short-term memory network (LSTM) to analyze the regularity and symmetry of joint movement trajectories based on gait information and joint angle data collected by wearable devices, thereby quantifying the joint's shock resistance during movement.

[0055] In some embodiments, dynamic shock absorption rate: combining plantar pressure sensor data and accelerometer information, it assesses the bone's cushioning efficiency against impact forces during movement. The model uses time-series analysis techniques to identify impact peaks and decay curves during the gait cycle and calculates the bone's dynamic protective capability.

[0056] In some embodiments, the AI ​​model further incorporates mechanical load, fall risk, and metabolic environmental variables to correct the initial redundancy. For example, mechanical load correction: by continuously monitoring the distribution of skeletal stress during daily activities (such as the peak pressure on the knee joint when climbing stairs), the finite element analysis algorithm is used to simulate the microstructural changes of bones under different loads, thus correcting the redundancy threshold. For example, fall risk correction: by fusing balance test data, historical fall records, and environmental perception data (such as the ratio of indoor to outdoor activities), a probabilistic graphical model is constructed to calculate the potential threat of accidental impacts to skeletal function. As another example, metabolic environment correction: by analyzing the correlation between blood biochemical indicators (such as calcium and phosphorus metabolite concentrations) and bone metabolism markers, the gradient boosting tree algorithm is used to quantify the impact of metabolic abnormalities on bone repair capacity.

[0057] Furthermore, based on the corrected skeletal functional redundancy, the model employing a dual time-series prediction architecture may include:

[0058] Short-term risk prediction: Gated recurrent unit (GRU) networks are used to analyze monthly data fluctuations, capturing the immediate impact of acute factors (such as sports injuries and changes in nutritional intake) on fracture risk. Long-term trend analysis: Quarterly data is processed using an attention-weighted Transformer model to identify the cumulative effect of chronic factors (such as age-related bone loss) on fracture risk. The final output fracture risk index uses a percentile grading system (such as T-score equivalent transformation) and generates a risk change curve containing a first threshold (95%) confidence interval, providing accurate time window predictions for clinical intervention. For example, when the AI ​​model predicts that the fracture risk will increase beyond the second threshold (e.g., 20%) in the next 3 or 6 months, the system will automatically trigger an alert and recommend enhanced intervention measures.

[0059] In some embodiments, a causal graph model (Do-Calculus) is embedded in the 3D AI fracture risk model to distinguish between correlated and causal risk factors. For example, it is found that insufficient muscle protection has a greater causal contribution to fracture risk than low bone mineral density (even with normal bone mineral density, low muscle mass can still lead to a high fracture risk). Based on this, the intervention priority is adjusted (prioritizing increasing muscle mass rather than simply supplementing calcium).

[0060] In some embodiments, the AI ​​model includes:

[0061] A short-term volatility early warning sub-model is included to address acute risks. For example, an LSTM-Attention hybrid model is used for acute risk events such as falls and bone stress injuries from strenuous exercise.

[0062] LSTM layer: Leveraging the temporal memory capabilities of Long Short-Term Memory (LSTM) networks, it captures the temporal dependencies of gait parameters (such as plantar pressure mutation sequences) and metabolite indicators (such as the time series of sudden drops in blood calcium concentration), and mines the dynamic correlations from the past to the present in continuous data.

[0063] Attention layer: Through an attention weight allocation mechanism, it focuses on abnormal changes at key time points (such as the surge in bone metabolism markers after a jump), enhancing the sensitivity to sudden abnormalities and triggering real-time warnings to achieve rapid response to sudden risks. For example, when an increased risk of falling is detected, protective recommendations are immediately pushed out.

[0064] A long-term trend prediction sub-model is developed to address long-term evolution. Targeting the long-term evolution of bone mass with slow variables such as age and hormones, and combining the physical constraints of the digital twin, a Transformer + physical constraint loss function architecture is adopted: Transformer layer: Leveraging the modeling advantages of self-attention mechanisms for long sequences, it learns the cumulative impact of slow variables such as age and hormone levels on bone mass (these variables change slowly over time and require global temporal correlation); Physical constraint loss function: Introduces prior physical laws of bone growth (e.g., bone loss rate does not exceed physiological thresholds) to constrain the model training process, preventing overfitting due to noisy data and ensuring that long-term predictions conform to the objective laws of bone physiological evolution (e.g., bone mass changes are consistent with the physiological growth / aging rhythm).

[0065] In S1140, baseline health data serves as a personalized baseline for the target user's skeletal health, encompassing baseline bone mineral density, normal gait characteristics, and baseline levels of bone metabolites, representing a reference state without significant pathology or risk. Real-time assessment indicators are quantitative / qualitative descriptions of the user's current skeletal status based on current monitoring data (such as current gait stability score and bone metabolite fluctuation values).

[0066] In S1140, determining whether a risk warning has been triggered may include:

[0067] When the deviation between the real-time assessment metric and the baseline health data exceeds a preset threshold, a prompt message is output. For example, if a user's baseline gait stability score is 85 (out of 100, representing their normal level of stability), and the real-time assessment score is 60 (a deviation of 25 points from the baseline, exceeding the preset threshold by 20 points), it indicates that the current gait stability is far below their normal level, posing a risk of falling, and a prompt should be output.

[0068] When the deviation between the real-time assessment indicator and the baseline health data is within a preset threshold, no prompts will be output to reduce unnecessary prompts. For example, a real-time gait score of 80 (a deviation of 5 points from the baseline score of 85, which is less than the threshold of 20 points) indicates that the current state is close to normal and there is no significant real-time risk, so no prompt is needed.

[0069] In some embodiments, the fall risk value can be based on gait information, bone metabolite information, and bone status measurement parameters collected by wearable devices, combined with the target user's baseline health data. It is an indicator quantified or graded by an evaluation model, reflecting the likelihood of the target user falling due to bone health-related factors. For example, if, after evaluation by an AI model, the deviation between the target user's gait stability parameters (such as stride variability and center of gravity shift) and the baseline health data exceeds a preset risk threshold, a fall risk value of 8 is calculated (risk levels are divided into 1-10, with higher values ​​indicating higher risk), indicating a high probability of the user falling.

[0070] In some embodiments, the AI ​​model also predicts the motion that would cause the target user to fall. For example,

[0071] By analyzing data such as gait information and bone status measurement parameters of target users, specific movement behaviors or scenarios that may cause or increase the probability of a user's fall due to a mismatch with the user's current skeletal health status (such as insufficient bone density, limited joint mobility, etc.). For example, if a user has decreased lumbar spine stability due to osteoporosis, the wearable device detects that when the user suddenly bends over to lift heavy objects, the abnormal amplitude of the spinal posture increases significantly, and the level of bone injury-related markers in bone metabolites rises instantaneously. Therefore, the sudden bending over to lift heavy objects is identified as the movement that caused the user's fall.

[0072] In some embodiments, as a means of mitigating risk, the AI ​​model may also develop specific and actionable action plans or interventions to reduce the probability of falling, based on the movement that caused the fall or the fall risk factors identified in the target user's current skeletal health. For example, it may suggest that some elderly people avoid walking on mountain roads.

[0073] In some embodiments, to enhance the necessity of risk warnings, it is possible to determine whether to output a warning by using two thresholds: the degree of deviation between the immediate assessment value and the baseline, and the fall risk level. Specifically, this includes, but is not limited to, at least one of the following:

[0074] Deviation threshold: The degree of difference between the real-time assessed skeletal status indicators (such as gait stability score) and the user's baseline health data exceeds a preset threshold.

[0075] Fall risk threshold: The fall risk value output by the model reaches or exceeds the preset risk level threshold (the risk value is divided into low risk 1-3, medium risk 4-7, and high risk 8-10, with medium risk and above (≥4) preset as the warning threshold).

[0076] In this embodiment, a prompt is only output when both the error rate and the risk value meet the standard, thereby filtering out scenarios triggered by a single condition but with insufficient actual risk and increasing the necessity of the prompt.

[0077] This dual-threshold approach to determining whether to output alerts takes into account both the dynamic differences in an individual's health status (reflected by baseline deviation to indicate individual-specific changes) and ensures that the severity of the risk meets the standard (reflected by risk value levels to indicate the degree of objective harm). Compared to single-condition triggering alerts, this significantly reduces false alarms or unnecessary alerts, making alerts more accurate and necessary. It avoids desensitizing users to frequent alerts and provides timely warnings when intervention is truly needed, thus improving the intelligence and humanization of bone health risk management.

[0078] In S1150, a targeted plan is generated based on health prediction parameters. This targeted plan includes a targeted exercise plan and a targeted diet plan. For example, the targeted exercise plan is a personalized exercise program, and the targeted diet plan is a customized nutrition intake guide.

[0079] For example, based on health status prediction parameters and combined with kinesiology and nutrition knowledge bases, the system generates suitable exercise type, intensity, and frequency plans, as well as dietary supplementation plans for nutrients such as calcium and vitamin D, for users' bone health problems and risks.

[0080] The generated exercise and diet plans are highly tailored to users' actual needs, effectively guiding them to improve bone health through daily behaviors, enhancing the pertinence and effectiveness of prevention and intervention, and reducing the likelihood of bone diseases.

[0081] In some embodiments, short-term planning sets up a first period and a second period. The first period aims to mitigate immediate risks, while the second period aims to consolidate the improved risk outcomes. This generates exercise and dietary plans, primarily high-frequency, easily implemented intervention programs tailored to the target user's physical condition based on short-term assessment indicators. For example, immediate risks to skeletal health (such as falls due to gait instability) are often related to short-term behaviors (such as sudden fatigue or temporary postural abnormalities). The first period (e.g., 3 days) can quickly correct immediate problems, while the second period (e.g., 1 week) can establish initial behavioral habits, preventing repeated risks and aligning with the management logic of requiring rapid response to immediate risks. For example, the first period is shorter than the second period.

[0082] In some embodiments, long-term planning aims to halt the deterioration of bone condition (such as a continuous decline in bone density) and establish sustainable healthy habits. It utilizes trend monitoring data from the third phase to specify a phased, systematic intervention program with a fourth phase. The third phase is longer than the second phase. The fourth phase is longer than the third phase. For example, bone metabolism has a repair cycle of 3-6 months. Because changes in bone density require more than 3 months to become detectable, 6 months allows for observation of the actual impact of the intervention on bone condition and timely adjustments to the program, avoiding the problems of too short a cycle not showing results or too long a cycle missing the optimal intervention opportunity.

[0083] Short-term planning provides support for achieving long-term goals. For example, the long-term goal of exercising three times a week relies on a week of consolidation training in the short-term plan to build a foundation—users first adapt to the exercise rhythm by doing 15 minutes of static squats every day for a week, and then gradually increase to 3 times a week for 30 minutes of exercise within 6 months, so as to avoid difficulty in starting due to the long-term goal being too far away; the habit of supplementing 200mg of calcium per day in the short-term plan also provides behavioral groundwork for the long-term goal of 800mg of calcium intake per day.

[0084] Long-term planning provides directional constraints for short-term planning. For example, all interventions in short-term planning must align with long-term goals. For instance, choosing leg training over upper body training for a one-week wall squat exercise is precisely because the long-term goal is to stimulate bone formation and improve bone density. As weight-bearing bones, the legs provide a more direct training effect. If upper body training is chosen in the short term, although it can improve muscle mass, it is unrelated to long-term bone health goals and would cause the intervention to deviate from its intended direction.

[0085] The collaborative customization of long-term and short-term planning solves the problems of traditional planning: short-term plans lack clear objectives, while long-term plans are difficult to implement. In the short term, real-time data is used to quickly mitigate risks, preventing users from being injured due to delayed intervention. In the long term, trend data is used to prevent deterioration, preventing users from developing bone diseases due to ignoring long-term trends. This compatibility forms a closed-loop management system encompassing real-time, medium-term, and long-term goals, ensuring both the timeliness of intervention and the sustainability of results, significantly improving the accuracy of bone health management and user compliance.

[0086] In summary, the alert information, based on the deviation between real-time assessment indicators and baseline health data, accurately outputs fall risk values ​​and mitigation measures, rapidly reducing short-term risks such as falls. Targeted planning, using real-time data to determine short-term interventions, consolidates the risk improvement effects of the alerts, while long-term trend data helps prevent bone deterioration. The two work synergistically: the alerts provide precise risk targets for targeted planning, preventing plans from deviating from actual needs; targeted planning, through short- and long-term interventions, complements the alerts, addressing both immediate risks and preventing recurrence and long-term bone diseases, forming a closed loop of alert-intervention-protection, significantly improving the timeliness and sustainability of bone health management.

[0087] In some embodiments, generating targeted exercise and dietary plans based on the health status prediction parameters includes:

[0088] The planned targeted exercise arrangement is determined based on the health status prediction parameters; the targeted exercise arrangement includes at least a targeted exercise plan for the type and / or intensity of the targeted exercise.

[0089] A targeted diet plan is generated based on the targeted exercise schedule and health status prediction parameters.

[0090] The system determines targeted exercise programs to be beneficial to the health of the target user. For example, based on real-time assessment indicators and dynamic trend data of bone health in the health status prediction parameters, combined with knowledge of exercise physiology, the system develops personalized targeted exercise programs. If the assessment indicators show that the user's bone density is low and declining, weight-bearing exercises, such as brisk walking and climbing stairs, are given priority to promote bone cell growth through the stimulation of bones by gravity. For users with poor joint flexibility, low-impact exercises such as yoga and swimming are chosen, which can both strengthen muscles and enhance bone stability while avoiding excessive wear and tear on joints. At the same time, the exercise intensity is precisely set according to the user's physical fitness and bone endurance, such as controlling the brisk walking speed to 80-100 steps per minute and the swimming duration to 30-45 minutes, to ensure the effectiveness and safety of the exercise.

[0091] For example, targeted motion arrangement generation may include:

[0092] Recommended targeted exercises based on the user's bone condition (e.g., lumbar spine osteopenia):

[0093] Low bone mass: Prioritize resistance training, such as machine training.

[0094] High risk of falling: Strengthen balance training, such as balance beam exercises or yoga balance training;

[0095] For arthritis patients: Water-based exercises, such as swimming, are recommended.

[0096] In some embodiments, multi-agent reinforcement learning (MARL) is introduced into the existing targeted exercise planning module: Agent 1 (motor module): generates an initial exercise plan (such as resistance training) based on the user's bone status (such as lumbar vertebral osteopenia); Agent 2 (nutrition module): recommends a nutritional supplement plan (such as deep-sea fish + sunshine) based on the user's metabolic status (such as vitamin D deficiency); the two agents iteratively optimize the strategy combination by sharing a reward function (such as bone mass growth rate + fall risk reduction), breaking the static binding relationship between exercise and nutrition.

[0097] In some embodiments, targeted dietary plans are generated to help target users achieve nutritional balance and bone improvement. Based on targeted exercise schedules and health status prediction parameters, dietary plans are generated by integrating nutritional principles. For users who need to increase bone density through weight-bearing exercise, the plan increases the intake of foods rich in calcium and vitamin D, such as milk, fish, and nuts, as vitamin D promotes calcium absorption. For users who engage in low-impact exercise, the plan considers supplementing with protein and Omega-3 fatty acids, such as chicken breast and deep-sea fish, to help repair minor muscle damage during exercise and improve exercise performance. Simultaneously, based on the user's metabolic status and bone metabolite information, the proportions of various nutrients are adjusted to achieve synergy between diet and exercise, comprehensively improving bone health.

[0098] For example, dynamic nutritional adaptation may include, but is not limited to, at least one of the following:

[0099] Based on dietary records and metabolic markers (such as serum 25(OH)D levels), adjust nutrient plans in real time: For those deficient in vitamin D: a combination of deep-sea fish and sunlight exposure from 10:00 to 14:00 is recommended; For those with insufficient stomach acid: calcium citrate is recommended as a substitute for calcium carbonate.

[0100] like Figure 2 As shown, in some embodiments, the method includes:

[0101] S1160: Based on the actual exercise status and actual diet of the target user, modify the targeted exercise arrangement and the targeted diet plan.

[0102] Through wearable devices, diet tracking apps, and other tools, we continuously collect data on the actual exercise and dietary status of target users. Regarding exercise, we record in detail the type, duration, intensity, and frequency of each user's workout, such as using smart bracelets to monitor steps and heart rate changes, and motion sensors to capture the degree of proper form in movements. For diet, users manually enter the types and quantities of food they eat daily, or use image recognition technology to scan food and automatically record its nutritional components. This data, combined with dietary check-in data from the community health platform, constructs a complete dietary profile.

[0103] The system compares and analyzes the collected actual data with the initial targeted exercise plan and dietary plan. Using statistical and machine learning algorithms, it determines whether the user has met the planned standards. For example, it analyzes whether the actual exercise intensity meets the requirements for improving bone density and whether the intake of nutrients such as calcium and vitamin D in the diet is sufficient. If the actual exercise intensity is insufficient and the bone density improvement effect does not meet expectations, the system will appropriately increase the exercise duration or adjust the exercise type, upgrading low-intensity walking to moderate-intensity brisk walking. If protein intake in the diet is consistently lower than the planned value, the system will optimize the dietary plan, recommend more protein-rich foods, and provide simple recipes and cooking tutorials.

[0104] Based on the analysis results, targeted exercise arrangements and dietary plans are dynamically adjusted to generate new personalized plans, which are then promptly pushed to users through community health platforms, social media messages, and other channels to ensure that the plans always fit the user's actual situation and effectively support bone health management.

[0105] In some embodiments, the adaptive intervention system may include, but is not limited to, at least one of the following:

[0106] Dynamic difficulty adjustment: For example, automatically upgrading the program based on the user's progress in fitness (such as from wall push-ups to resistance band training to free weights).

[0107] Nutrient absorption optimization: For example, when poor calcium absorption is detected, the program automatically increases the intake of co-nutrients such as vitamin K2 and magnesium.

[0108] In some cases, certain types or intensities of exercise may be too difficult for the target user, so adjustments to the difficulty level can be considered. Similarly, if a user dislikes certain dishes, the dishes or drinks can be adjusted based on their nutritional needs.

[0109] In some embodiments, the method further includes:

[0110] The targeted exercise and diet plans were published in the community.

[0111] Build or participate in communities based on the status of the target users;

[0112] Based on the targeted exercise plan and dietary plan, challenge activities related to the community;

[0113] Guide and monitor the target users' participation in the challenge activities through virtual reality (VR) and / or augmented reality (AR);

[0114] Based on the participation, the challenge rankings of the target users will be tracked and reported.

[0115] When publishing targeted exercise and diet plans in the community, a dedicated community health platform can be built to push personalized plans in the form of pictures, text, and short videos. At the same time, the plans can be categorized and displayed according to user tags such as age and health status, making it easy for people with different needs to quickly access them. Intelligent search functions can also be embedded to make it easy for users to find specific plans at any time, thereby improving the reach of health knowledge.

[0116] Communities are created or members participate in based on the target user's profile. First, cluster analysis is performed on user health data and behavioral preferences to group users with similar needs, such as a low bone density improvement group or a post-operative rehabilitation exchange group. Each community is staffed with professional health consultants who regularly organize online lectures and Q&A sessions to promote experience sharing and mutual support among users.

[0117] Community challenge activities are conducted based on targeted planning, with a tiered task system designed, such as a 21-day calcium intake challenge requiring three resistance training sessions per week. The activities include a badge and points reward system, with points redeemable for health products or services to motivate user participation.

[0118] The system utilizes Virtual Reality (VR) and Augmented Reality (AR) technologies to guide and supervise challenge activities. After users wear the devices, VR simulates the movement scenario and corrects movement techniques in real time; AR overlays virtual prompts onto the real environment, such as indicating correct dietary combinations. The system automatically records user progress and generates visual feedback data.

[0119] The challenge rankings are tracked and reported based on participation, and the leaderboard is updated in real time within the community to showcase users' progress. Individual progress reports and success stories are also published to foster a healthy competitive environment and encourage users to continuously pay attention to and improve their bone health.

[0120] The embodiments disclosed herein include multimodal evaluation, intelligent matching, 3D prediction, adaptive intervention, and gamified community, and further include the following improvements:

[0121] First, the multimodal skeletal health assessment system introduces a multi-source heterogeneous data fusion algorithm to construct a digital twin model of skeletal health. It not only collects basic data such as gait, muscle mass, and bone density, but also integrates environmental data (such as the intensity of sunlight and air quality in the place of residence), genetic data (such as VDR gene polymorphism), and behavioral data (such as sleep quality and stress level).

[0122] We construct a digital twin model of bone health, using physical modeling and data-driven methods to simulate the dynamic changes of an individual's bones over time and predict bone mass trends over a period of 3 or 6 months.

[0123] Thus, short-term predictions are based on real-time monitoring data, while digital twin models introduce long-term dynamic simulations, representing monitoring across time scales. Multi-source heterogeneous data fusion algorithms (such as federated learning + time series modeling) can improve prediction robustness.

[0124] For example, based on the existing three-dimensional AI fracture risk model, physical information neural networks (PINNs) are introduced, and the bone biomechanical equations (such as Wolff's law: bones reshape themselves to adapt to mechanical loads) are embedded as constraints in the AI ​​training process, so that the digital twin not only learns the correlation of data, but also follows the physical laws of bone growth.

[0125] Regarding the intelligent sports-nutrition matching engine, a reinforcement learning dynamic optimization mechanism is introduced to construct a personalized sports-nutrition strategy space. It no longer relies on preset rules (such as resistance training for low bone mass), but instead uses reinforcement learning (such as Deep Q-Learning) to dynamically optimize exercise prescriptions and nutrition planning based on users' historical interaction data (such as exercise completion rate and nutrition intake feedback). The sports-nutrition strategy space maps exercise type (resistance / aerobic / balance), intensity, frequency, and nutrient (calcium / vitamin D / protein) combinations into a searchable strategy space, guiding strategy optimization through reward functions (such as bone mass growth rate and fall risk reduction). Compared to static matching rules, the reinforcement learning mechanism enables dynamic adaptive optimization. The construction of the strategy space decouples sports and nutrition as independent variables, and then achieves synergistic effects through joint optimization, unlike existing linear correlation models.

[0126] Furthermore, the three-dimensional fracture risk prediction model incorporates a causal inference model to distinguish between correlational and causal risk factors, thereby improving the accuracy of targeted exercise and dietary planning. Based on the integration of bone structure, mechanical load, and fall risk, causal inference algorithms (such as Do-Calculus and causal graph models) are introduced to identify the true causal risk factors leading to fractures (such as insufficient muscle protection versus simply low bone mineral density). The model outputs a Causal Fracture Risk Index (CFRI) to differentiate between high-risk groups (such as those with low muscle mass but normal bone mineral density) and low-risk groups (such as those with low bone mineral density but strong muscle protection). Combining data-driven correlation and causal inference models eliminates spurious correlations and improves prediction accuracy. The CFRI provides a causal basis for fracture prevention, unlike existing models that rely solely on statistical risk.

[0127] An adaptive intervention system incorporates biofeedback closed-loop control to achieve dynamic calibration of nutrient absorption. It not only detects calcium malabsorption (e.g., through serum calcium / PTH levels) but also introduces a biofeedback closed-loop control mechanism: real-time monitoring of blood calcium concentration and gut microbiota metabolites (such as short-chain fatty acids) via wearable devices dynamically adjusts the dosage of co-nutrients such as vitamin K2 and magnesium. A dynamic calibration model for nutrient absorption is constructed, providing closed-loop feedback of data from the entire nutrient intake, absorption, metabolism, and bone metabolism chain to achieve precise nutrient intervention.

[0128] For example, taking the increase of vitamin K2 as an example, a closed-loop control mechanism can dynamically adjust the amount of vitamin K2 in real time, and determine whether it is necessary to stop taking the corresponding supplements. For example, the introduction of gut microbiota metabolites links nutrient absorption with the microbiome, expanding the boundaries of traditional nutrition.

[0129] This gamified health community incorporates a blockchain incentive mechanism to build a trusted health behavior ledger. User exercise / nutrition behavior data is stored on the blockchain, creating an immutable ledger. Bone strength rewards are automatically distributed via smart contracts (e.g., completing jump training earns an NFT badge). A decentralized autonomous organization (DAO) model is introduced, where users vote to determine community challenge rules and reward distribution, enhancing participation and sustainability.

[0130] During data collection, an active stimulation and feedback loop is introduced to achieve active data acquisition and a closed-loop acquisition process. If only wearable devices are used to passively collect data (such as gait and metabolites), the data coverage may be incomplete due to the user's limited daily activities (e.g., sedentary individuals lack impact data). Introducing an active stimulation module, which adds a micro-vibration stimulator integrated into the insole or wristband, allows for controlled vibrations (frequency 5-50Hz, amplitude 0.1-0.5mm) to actively induce specific gait patterns (e.g., simulating foot pressure changes during running), forcing users to generate more diverse motion data.

[0131] In dynamic data quality assessment, since active stimulation may introduce noise (such as sensor drift caused by vibration), a data reliability assessment model is developed. For example, feature engineering: extracting statistical features (mean, variance, frequency domain entropy) of gait parameters before and after vibration; model design: a lightweight XGBoost classifier to determine data validity (labels are calibrated offline using high-precision laboratory equipment). Assuming the collected data is absolutely reliable, this invention resolves the contradiction between data coverage and quality through active stimulation and dynamic quality assessment. In adaptive optimization of stimulation parameters, for example, dynamically adjusting vibration parameters (frequency, amplitude, duration) based on reinforcement learning (PPO algorithm): state space: current gait data quality score, user fatigue (heart rate variability HRV); action space: combination of vibration parameters; reward function: increased proportion of valid data + reduced user discomfort (quantified through questionnaire feedback).

[0132] In this way, a closed loop of active stimulation → data quality assessment → parameter adaptive optimization is formed, which upgrades the acquisition module from passive recording to active exploration, significantly improving the integrity of multimodal data.

[0133] Secondly, from the generality of exercise planning and the construction of personalized exercise to metabolic coupling models, relying solely on bone status (e.g., resistance training recommended for low bone mass) while ignoring the user's metabolic capacity (e.g., differences in mitochondrial function leading to varying energy expenditure under the same exercise) may result in ineffective interventions. Therefore, a metabolic capacity assessment module should be introduced. A new respiratory gas analysis sensor (integrated into a mask or chest strap) should be added to monitor oxygen uptake (VO2) and carbon dioxide excretion (VCO2) in real time during exercise. Key indicators calculated may include, but are not limited to, at least one of the following: maximum oxygen uptake (VO2max): reflecting cardiorespiratory endurance; fat oxidation rate: measuring fat metabolism efficiency; pyruvate threshold: determining the anaerobic metabolism initiation point. Thus, metabolic capacity directly affects exercise performance (e.g., high-intensity training in individuals with low VO2max may lead to fatigue accumulation rather than bone mass growth).

[0134] The exercise-to-metabolism coupling model, constructing a dual-path neural network, can include: Path 1 (traditional): bone status → exercise type recommendation (e.g., low bone mass → resistance training); Path 2 (new): metabolic capacity → exercise intensity adjustment (e.g., reducing the percentage of resistance training load for individuals with low VO2max); Fusion layer: weighted summation of the outputs of the two paths to generate the final exercise prescription (considering the balance between bone health needs and metabolic tolerance). This breaks through the limitation of focusing solely on bone status by introducing metabolic capacity as a key variable for regulating exercise intensity.

[0135] The integrated intervention from metabolism to nutrition can include: dynamically adjusting nutritional plans based on metabolic data during exercise; if a low fat oxidation rate is detected (indicating excessive carbohydrate dependence), increasing dietary fiber intake (delaying glucose absorption and promoting fat mobilization); if the pyruvate threshold is advanced (indicating premature anaerobic metabolism), supplementing with branched-chain amino acids (BCAAs) to reduce muscle protein breakdown. This forms an exercise-metabolism coupling model → personalized intensity adjustment → integrated metabolism-nutrition intervention chain, achieving a leap from a single bone health goal to multi-system synergistic optimization.

[0136] Finally, the solution to the lag in risk prediction lies in developing a skeletal digital twin + quantum computing accelerated prediction system. If a 3D AI fracture risk model relies on historical data to predict future trends, it cannot respond in real time to sudden physiological changes (such as bone metabolic disorders caused by acute inflammation). The solution is to introduce real-time simulation using a skeletal digital twin. Based on the digital twin model, an organ-level physics engine (based on finite element analysis, FEM) can be introduced. This can include: dividing the bone into micrometer-level mesh units to simulate the stress distribution of the trabecular bone microstructure under mechanical load; and real-time access to gait data (plantar pressure) and muscle contraction force (sEMG signal) to dynamically update the mesh stress state. In this way, FEM can accurately predict the risk of localized bone micro-damage (such as stress concentration areas), overcoming the shortcomings of traditional models that only focus on macroscopic bone density.

[0137] Online quantum computing is employed to accelerate predictions. For example, for computationally intensive FEM simulations (with mesh elements reaching millions), quantum annealing algorithms are used for optimization: the stress distribution problem is transformed into a quadratic unconstrained binary optimization (QUBO) form; a quantum annealer (such as the D-Wave system) is used to quickly find the global optimum, reducing the simulation time from hours to minutes. This eliminates reliance on serial computation using classical computers, enabling real-time dynamic predictions through quantum acceleration, supporting responses to sudden risks.

[0138] The emergency risk intervention protocol can include: when the digital twin detects that local stress exceeds a threshold (e.g., the instantaneous impact force on the ankle joint during running exceeds the safety limit), a three-level intervention protocol is triggered: Level 1: Instant reminder to adjust movement (displaying the correct posture via AR glasses); Level 2: Temporarily reduce exercise intensity (prompting deceleration via vibration from a smart bracelet); Level 3: Push nutritional supplementation plans (e.g., quickly replenishing electrolytes to relieve muscle cramps). This forms a closed loop of real-time simulation → quantum acceleration → emergency intervention, upgrading risk prediction from post-event analysis to pre-event prevention, completely solving the problem of lag.

[0139] like Figure 3As shown, this disclosure provides a dynamic tracking and intervention system for bone health based on exercise and nutrition, including: a data acquisition module 3110, used to collect gait information, bone metabolite information, and bone status measurement parameters of a target user using a wearable device; a prediction module 3120, used to process the gait information, bone metabolite information, and bone status measurement parameters based on an artificial intelligence (AI) model to obtain bone status prediction parameters; an evaluation module 3130, used to evaluate the bone health status of the target user based on the bone status prediction parameters and the target user's health data to obtain health status prediction parameters; the health status prediction parameters include: real-time bone health status evaluation indicators and dynamic trend data of bone health; and a prompting module 3140. When the target user is determined to have a fall risk based on the immediate assessment indicators and the target user's baseline health data, a prompt message is output. The prompt message includes: a fall risk value, the movement that caused the fall, and measures to mitigate the fall risk. A planning module 3150 is used to generate a targeted plan based on the health status prediction parameters. The targeted plan includes: an exercise plan and a dietary plan. The targeted exercise plan guides the target user's exercise; the dietary plan guides the target user's diet. The targeted plan includes a long-term plan to counteract the long-term bone health risks indicated by the dynamic trend data of bone health, and a short-term plan consistent with the goals of the long-term plan and avoiding the fall risk corresponding to the immediate assessment indicators.

[0140] The system may include one or more electronic devices, which can, for example... Figure 4 As shown.

[0141] In some embodiments, the acquisition module 3110 is specifically configured to perform at least one of the following:

[0142] The gait information is collected using foot pressure-sensitive shoes or smart ankle bracelets;

[0143] Bone structure parameters were measured using a home ultrasound bone densitometer wristband.

[0144] Muscle parameters were obtained by measuring muscle condition using bioelectrical impedance analysis.

[0145] The bone metabolite information is collected using a bandage-type sensor; wherein, the bandage sensor can also be used to collect muscle movement information at its location; the muscle movement information is also used to predict the bone state prediction parameters.

[0146] In some embodiments, the planning module 3140 is specifically used to determine a planned targeted exercise arrangement based on the health status prediction parameters; the targeted exercise arrangement includes at least a targeted exercise plan of targeted exercise type and / or exercise intensity; and to generate a targeted diet plan based on the targeted exercise arrangement and the health status prediction parameters.

[0147] In some embodiments, the system further includes: a correction module, configured to correct the targeted exercise arrangement and the targeted diet plan based on the target user's actual exercise status and actual diet.

[0148] In some embodiments, the assessment module is specifically used to obtain the fracture risk index of the target user and the fracture change trend with a preset period as the time unit by using a three-dimensional AI fracture risk model with the bone structure index, mechanical load index and fall risk index.

[0149] In some embodiments, the prediction module is further configured to use a three-dimensional AI fracture risk model to determine skeletal functional redundancy based on the muscle-to-bone ratio index, joint stability coefficient, and dynamic shock absorption rate; to correct the skeletal functional redundancy using the mechanical load index, fall risk index, and metabolic environment; and to determine the fracture risk index of the target user and the fracture change trend with a preset period as the time unit based on the skeletal functional redundancy.

[0150] In some embodiments, the system further includes: a publishing module for publishing the targeted exercise plan and diet plan in a community; a component module for forming or participating in a community based on the target user's status; a challenge module for conducting challenge activities related to the community based on the targeted exercise plan and diet plan; a monitoring module for guiding and monitoring the target user's participation in the challenge activities through virtual reality (VR) and / or augmented reality (AR); and a tracking module for tracking and reporting the target user's challenge ranking based on the participation.

[0151] Figure 4 As shown, this application embodiment provides an electronic device including a processor 10 and a memory 11. Optionally, the device may further include a communication interface 12 and a bus 9. The processor 10, communication interface 12, and memory 11 can communicate with each other via the bus 9. The communication interface 12 can be used for information transmission. The processor 10 can call logical instructions in the memory 11 to execute the queue-based voiceprint data processing method of the above embodiment.

[0152] Furthermore, the logical instructions in the aforementioned memory 11 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium.

[0153] The memory 11, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as the program instructions / modules corresponding to the methods in the embodiments of this application. The processor 10 executes functional applications and data processing by running the program instructions / modules stored in the memory 11, thereby realizing the dynamic tracking and intervention method for bone health based on exercise and nutrition in the above embodiments.

[0154] The memory 11 may include a program storage area and a data storage area. The program storage area may store the operating system and application programs required for at least one function; the data storage area may store data created based on the use of the electronic device. Furthermore, the memory 11 may include high-speed random access memory and may also include non-volatile memory.

[0155] This application provides a computer program product, which includes a computer program stored on a storage medium. The computer program includes program instructions, which, when executed by a computer, cause the computer to perform the above-mentioned method for dynamic tracking and intervention of bone health based on exercise and nutrition.

[0156] The aforementioned computer-readable storage medium may be a transient computer-readable storage medium or a non-transitory computer-readable storage medium.

[0157] The technical solutions of this application embodiment can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes one or more instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in this application embodiment. The aforementioned storage medium can be a non-transitory storage medium, including various media capable of storing program code such as USB flash drives, portable hard drives, read-only memory, random access memory, magnetic disks, or optical disks, or it can be a transient storage medium.

[0158] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0159] The embodiments or examples disclosed in this application are not exhaustive, but merely illustrative of some embodiments or examples, and are not intended to limit the scope of protection of this disclosure. Unless contradictory, each step in a particular embodiment or example can be implemented as an independent embodiment, and the steps can be arbitrarily combined. For example, a solution after removing some steps in a particular embodiment or example can also be implemented as an independent embodiment, and the order of the steps in a particular embodiment or example can be arbitrarily interchanged. Furthermore, optional methods or examples in a particular embodiment or example can be arbitrarily combined; moreover, embodiments or examples can be arbitrarily combined. For example, some or all steps of different embodiments or examples can be arbitrarily combined, and a particular embodiment or example can be arbitrarily combined with optional methods or examples of other embodiments or examples.

[0160] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0161] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0162] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0163] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0164] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for dynamic tracking and intervention of bone health based on exercise and nutrition, characterized in that, include; Wearable devices are used to collect gait information, bone metabolite information, and bone status measurement parameters of target users; the gait information, bone metabolite information, and bone status measurement parameters are processed based on an artificial intelligence (AI) model to obtain bone status prediction parameters; Based on the bone status prediction parameters and the target user's health data, the target user's skeletal health status is assessed to obtain health status prediction parameters, including: determining skeletal functional redundancy using a three-dimensional AI fracture risk model based on the muscle-to-bone ratio index, joint stability coefficient, and dynamic impact absorption rate; correcting the skeletal functional redundancy using the mechanical load index, fall risk index, and metabolic environment; and determining the target user's fracture risk index and fracture change trend over a preset period based on the corrected skeletal functional redundancy. The health status prediction parameters include: real-time skeletal health status assessment indicators and dynamic skeletal health trend data. The muscle-to-bone ratio index reflects the synergistic biomechanical relationship between muscles and bones; the joint stability coefficient, obtained from gait information and joint angle data, reflects the regularity and symmetry of joint movement trajectories to quantify the joint's impact resistance during movement; the dynamic impact absorption rate, assessed by combining plantar pressure sensor data and acceleration set information, evaluates the bone's buffering efficiency against impact forces during movement, which is used by the AI ​​model to identify impact peaks and attenuation curves in the gait cycle through time-series analysis to calculate the bone's dynamic protection capability; and the bone functional redundancy measures the bone's reserve capacity beyond basic physiological needs when subjected to daily activities and potential risks. When the target user is determined to be at risk of falling based on the real-time assessment indicators of bone health status and the target user's baseline health data, a prompt message is output; the prompt message includes: fall risk value, the movement that caused the fall, and measures to prevent the fall risk; A targeted plan is generated based on the health status prediction parameters; the targeted plan includes: exercise planning and diet planning; wherein, the targeted exercise plan is used to guide the exercise of the target user; the diet plan is used to guide the diet of the target user; the targeted plan includes a long-term plan to counteract the long-term bone health risks indicated by the bone health dynamic trend data, and a short-term plan that is consistent with the goals of the long-term plan and avoids the fall risk corresponding to the immediate assessment indicators.

2. The method according to claim 1, characterized in that, The method of using wearable devices to collect gait information and bone metabolite information of target users includes at least one of the following: The gait information is collected using foot pressure-sensitive shoes or smart ankle bracelets; Bone structure parameters were measured using a home ultrasound bone densitometer wristband. Muscle parameters were obtained by measuring muscle condition using bioelectrical impedance analysis. The bone metabolite information is collected using a bandage-type sensor; wherein, the bandage sensor can also be used to collect muscle movement information at its location; the muscle movement information is also used to predict the bone state prediction parameters.

3. The method according to claim 1 or 2, characterized in that, The step of generating targeted planning based on the health status prediction parameters includes: The planned targeted exercise arrangement is determined based on the health status prediction parameters; the targeted exercise arrangement includes at least a targeted exercise plan for the type and / or intensity of the targeted exercise. A targeted diet plan is generated based on the targeted exercise schedule and health status prediction parameters.

4. The method according to claim 3, characterized in that, The method includes: Based on the target user's actual exercise status and dietary situation, the targeted exercise arrangement and the targeted diet plan are modified.

5. The method according to claim 2, characterized in that, The method of collecting gait information using foot pressure sensing shoes or smart ankle bracelets includes: using the pressure sensor array built into the foot pressure sensing shoes to sense the pressure distribution and force application time in various areas of the sole to calculate stride length, stride frequency, and gait cycle; and / or, using interchangeable accelerometers and gyroscopes to capture the acceleration and angular velocity of ankle movement to analyze the swing amplitude and rhythm changes of the target user when walking. And / or, The use of bioelectrical impedance analysis to measure muscle condition and obtain musculoskeletal parameters includes: injecting a weak and safe current into the human body through surface electrodes, measuring the impedance data of the current in the body, and separating muscle content, muscle mass, and musculoskeletal ratio from the impedance data. And / or, The information on bone metabolites is collected using a bandage-type sensor, including: the bandage-type sensor is equipped with a microfluidic chip and a biosensor to adsorb sweat from the skin surface and perform biochemical analysis on trace amounts of calcium and phosphorus metabolites in the sweat to obtain bone metabolite information; and the strain sensor built into the bandage-type sensor can sense the stretching and deformation of muscles in the bandage-covered area and collect the frequency and amplitude of muscle movement. The bone structure index, the mechanical load index, and the fall risk index are used together to obtain the fracture risk index of the target user and the fracture change trend with a preset period as the time unit.

6. The method according to claim 1, characterized in that, Based on the corrected skeletal functional redundancy, the fracture risk index of the target user and the fracture change trend over a preset period are determined, including: The gated recurrent unit network of the AI ​​model is used to analyze monthly data fluctuations and capture the immediate impact of acute factors on fracture risk. The AI ​​model's attention-weighted transformation model is used to process quarterly data to identify the cumulative effect of chronic factors on fracture risk. The fracture risk index is output based on the immediate impact and the cumulative effect.

7. The method according to claim 1 or 2, characterized in that, The method further includes: The targeted exercise and diet plans were published in the community. Build or participate in communities based on the status of the target users; Based on the targeted exercise plan and dietary plan, challenge activities related to the community; Guide and monitor the target users' participation in the challenge activities through virtual reality (VR) and / or augmented reality (AR); Based on the participation, the challenge rankings of the target users will be tracked and reported.

8. A dynamic tracking and intervention system for bone health based on exercise and nutrition, characterized in that, include: The data acquisition module is used to collect gait information, bone metabolite information, and bone status measurement parameters of the target user using wearable devices. The prediction module is used to process the gait information, bone metabolite information and bone status measurement parameters based on the artificial intelligence (AI) model to obtain bone status prediction parameters. The assessment module is used to assess the bone health status of the target user based on the bone status prediction parameters and the target user's health data, and obtain the health status prediction parameters. The health status prediction parameters include: real-time assessment indicators of skeletal health status and dynamic trend data of skeletal health; the assessment module is specifically used to determine skeletal functional redundancy based on the muscle-to-bone ratio index, joint stability coefficient, and dynamic impact absorption rate using a three-dimensional AI fracture risk model; to correct the skeletal functional redundancy using the mechanical load index, fall risk index, and metabolic environment; and to determine the fracture risk index of the target user and the fracture change trend with a preset period as the time unit based on the corrected skeletal functional redundancy; the muscle-to-bone ratio index is used to reflect the synergistic mechanical relationship between muscles and bones; the joint stability coefficient is obtained by analyzing the gait information and joint angle data to reflect the regularity and symmetry of joint movement trajectory, thereby quantifying the impact resistance of joints during movement; the dynamic impact absorption rate is evaluated by combining plantar pressure sensor data and acceleration set information to assess the buffering efficiency of bones against impact forces during movement, which is used by the AI ​​model to identify the impact peak and attenuation curve in the gait cycle through time-series analysis technology to calculate the dynamic protection capacity of bones; the skeletal functional redundancy is used to measure the reserve capacity of bones beyond basic physiological needs when subjected to daily activities and potential risks. The prompting module is used to output prompting information when it is determined that the target user is at risk of falling based on the real-time assessment indicators of bone health status and the target user's baseline health data; the prompting information includes: fall risk value, the movement that caused the fall, and measures to prevent the fall risk; The planning module is used to generate targeted plans based on the health status prediction parameters; the targeted plans include: exercise plans and diet plans; wherein, the targeted exercise plans are used to guide the exercise of the target user; the diet plans are used to guide the diet of the target user; the targeted plans include long-term plans to combat the long-term bone health risks indicated by the bone health dynamic trend data, and short-term plans that are consistent with the goals of the long-term plans and avoid the fall risks corresponding to the immediate assessment indicators.

9. The system according to claim 8, characterized in that, The acquisition module is specifically used to perform at least one of the following: The gait information is collected using foot pressure-sensitive shoes or smart ankle bracelets; Bone structure parameters were measured using a home ultrasound bone densitometer wristband. Muscle parameters were obtained by measuring muscle condition using bioelectrical impedance analysis. The bone metabolite information is collected using a bandage-type sensor; wherein, the bandage sensor can also be used to collect muscle movement information at its location; the muscle movement information is also used to predict the bone state prediction parameters.

10. A computer-readable medium having instructions stored thereon that, when executed by one or more processors, cause the processors to perform the method as described in any one of claims 1 to 7.

Images