AI-based question-answering system for adjusting rehabilitation plans for hip fractures in the elderly.

Abstract

This invention relates to the field of medical rehabilitation technology, and more particularly to an AI-based question-and-answer-based rehabilitation plan adjustment system for elderly patients with hip fractures. This system establishes an individualized evolutionary trajectory by analyzing medical images and motion data, and constructs a motion-pain correlation model to dynamically adjust the exercise safety boundaries and load control scheme. Furthermore, it determines differentiated follow-up strategies based on rehabilitation progress and resource status, and continuously optimizes the model and scheme using follow-up effect data. This invention achieves personalized dynamic adjustment of rehabilitation plans and efficient resource allocation, improving rehabilitation safety and effectiveness.

Description

Technical Field

[0001] This invention relates to the field of medical rehabilitation technology, and in particular to an AI-based question-and-answer system for adjusting rehabilitation plans for hip fractures in the elderly. Background Technology

[0002] Hip fractures are a common and serious injury among the elderly, and their rehabilitation process is lengthy and complex, requiring the development and dynamic adjustment of rehabilitation plans based on individual patient conditions. Traditional rehabilitation plan adjustments mainly rely on regular clinical assessments by rehabilitation physicians or therapists. Physicians assess rehabilitation progress based on their professional experience by reviewing the patient's imaging reports, inquiring about subjective pain levels, and conducting physical examinations such as limited joint range of motion, and then manually adjust subsequent rehabilitation training programs. The core of this model is intermittent, manual decision-making based on fixed time points.

[0003] Current practices have significant shortcomings. The intervals between rehabilitation assessments are too long, making it impossible to capture subtle functional changes and potential pain triggers in patients' daily activities in real time. Rehabilitation plans are mostly phased, general programs, lacking continuous quantitative tracking and prediction of individual patient bone healing dynamics and the non-linear trajectory of muscle function recovery. Adjustments rely on physicians' experience and judgment, making it difficult to accurately quantify the dynamic relationship between exercise load and pain response. This results in rehabilitation plans that are either too conservative, prolonging the recovery period, or too aggressive, increasing the risk of re-injury.

[0004] Another significant drawback lies in the inadequate allocation of medical resources and follow-up interventions. In community or home-based rehabilitation settings, monitoring of patient rehabilitation progress is weak, and follow-up appointments are typically based on fixed schedules rather than the patient's real-time rehabilitation status and risk. This easily leads to low-risk patients consuming unnecessary medical resources, while abnormal conditions in high-risk patients fail to be identified in a timely manner and receive intensive intervention. The feedback loop for rehabilitation effectiveness is slow, adjustment strategies lag behind, and a personalized adaptive optimization mechanism for rehabilitation plans based on continuous data feedback cannot be formed. Summary of the Invention

[0005] This invention provides an AI-based question-and-answer system for adjusting rehabilitation plans for hip fractures in the elderly, which can solve the problems in the prior art.

[0006] A first aspect of the present invention provides an AI-based question-and-answer-based system for adjusting rehabilitation plans for hip fractures in the elderly, comprising:

[0007] The data analysis unit is used to acquire hip bone medical imaging data, motion monitoring data and rehabilitation history data of the target elderly user, analyze the hip bone medical imaging data to extract bone structure features, establish an individualized evolutionary trajectory in combination with the rehabilitation history data, and determine the current rehabilitation stage and functional recovery expectation based on the individualized evolutionary trajectory.

[0008] The exercise safety unit is used to construct an exercise-pain correlation model. By analyzing the correspondence between load changes and user pain feedback in the exercise monitoring data, it identifies the characteristics of exercise patterns that aggravate pain, dynamically adjusts the exercise safety boundary according to the current rehabilitation stage in the individualized evolution trajectory, and generates a load control scheme based on the exercise safety boundary.

[0009] The follow-up strategy unit is used to obtain the actual deviation of the exercise safety boundary based on the execution of the load control scheme, calculate the follow-up urgency score in combination with the expected progress of functional recovery, and determine the follow-up timing and intervention intensity based on the follow-up urgency score and the real-time availability of community medical resources, thereby forming a differentiated follow-up execution strategy.

[0010] The model and scheme unit is used to execute the follow-up execution strategy and collect rehabilitation effect data after follow-up. The rehabilitation effect data is compared with the expected trajectory in the individualized evolution trajectory. The comparison results are used to update the boundary parameters of the exercise-pain association model and the adjustment strategy of the load control scheme.

[0011] The data parsing unit is also used for:

[0012] The hip bone medical imaging data is analyzed to extract bone structure features. An individualized evolutionary trajectory is established by combining this trajectory with the rehabilitation history data. Based on this individualized evolutionary trajectory, the current rehabilitation stage and expected functional recovery are determined, including:

[0013] Bone morphology analysis was performed on hip bone medical imaging data to extract structural degeneration features reflecting changes in bone density distribution and joint space. The evolution rate of these structural degeneration features was then analyzed by time series comparison to generate a degeneration rate curve characterizing the dynamic process of bone degeneration.

[0014] Extract the functional recovery range and recovery time corresponding to different rehabilitation stages from rehabilitation history data, correlate and map the functional recovery range with the degeneration rate curve, identify the nonlinear response relationship between bone degeneration rate and functional recovery ability, and construct an individualized evolutionary trajectory that reflects the sensitivity of individual bone status to rehabilitation response.

[0015] The structural degradation features are located in the individualized evolutionary trajectory. Based on the location results, the rehabilitation stage identifier of the current bone status is determined. Based on the response relationship in the individualized evolutionary trajectory, the functional recovery potential and expected recovery period corresponding to the rehabilitation stage identifier are deduced.

[0016] By mapping the functional recovery magnitude to the degradation rate curve, the nonlinear response relationship between bone degradation rate and functional recovery ability is identified, and an individualized evolutionary trajectory reflecting the sensitivity of an individual's bone status to rehabilitation response is constructed, including:

[0017] The functional recovery range is segmented according to the time dimension to obtain the recovery rate change characteristics in different time periods. The recovery rate change characteristics are paired with the degradation rate values ​​in the corresponding time periods in the degradation rate curve to establish a paired dataset between degradation rate and recovery rate.

[0018] Nonlinear fitting analysis is performed on the paired dataset to identify the differences in response patterns of functional recovery ability when the degradation rate is in different numerical ranges. A sensitivity function reflecting the change of response sensitivity with degradation rate is extracted, and an individualized evolutionary trajectory is constructed based on the sensitivity function.

[0019] The motion safety unit is also used for:

[0020] A movement-pain correlation model is constructed. By analyzing the correspondence between load changes and user pain feedback in the movement monitoring data, the characteristics of movement patterns that exacerbate pain are identified. The movement safety boundary is dynamically adjusted based on the current rehabilitation stage in the individualized evolutionary trajectory. A load control scheme is generated based on the movement safety boundary, including:

[0021] The load change sequence is extracted from the exercise monitoring data. The load change sequence is time-aligned with the user pain feedback value at the corresponding time. The time-aligned load change sequence and pain feedback value are correlated and analyzed to identify the load change pattern that causes the pain feedback value to exceed the baseline level. The exercise intensity threshold and duration threshold in the load change pattern are extracted as pain triggering features. Based on the pain triggering features, an exercise-pain correlation model is constructed.

[0022] Obtain the current rehabilitation stage identifier in the individualized evolutionary trajectory, adjust the pain triggering features in the movement-pain association model according to the functional recovery potential corresponding to the current rehabilitation stage identifier, generate a movement safety boundary that matches the current rehabilitation stage, and generate a load control scheme based on the movement safety boundary.

[0023] The follow-up strategy unit is also used for:

[0024] Based on the execution status of the load control scheme, the actual deviation from the exercise safety boundary is obtained. Combined with the expected progress of functional recovery, a follow-up urgency score is calculated. Based on the follow-up urgency score and the real-time availability of community medical resources, the timing and intensity of follow-up are determined, forming a differentiated follow-up execution strategy, including:

[0025] Monitor the actual execution data of the load control scheme, quantify the deviation between the motion intensity sequence in the actual execution data and the allowable range of the motion safety boundary, identify the time distribution pattern of the deviation and the cumulative trend of the deviation magnitude, and generate the actual deviation degree curve;

[0026] The expected recovery trajectory and the current actual recovery trajectory are obtained. The deviation area between the trajectories is calculated as the progress deviation. The cumulative trend characteristics of the actual deviation curve are nonlinearly mapped to the progress deviation. When the cumulative trend accelerates and the deviation exceeds the preset baseline, the urgency amplification mechanism is triggered to generate a dynamically adjusted follow-up urgency score.

[0027] Based on the follow-up urgency score, a resource matching priority sequence is defined. Follow-up resources are selected from the medical personnel closest to the available time and the medical equipment with the lowest load status according to the resource matching priority sequence. At the same time, the intervention intensity level is determined according to the numerical gradient of the follow-up urgency score. A differentiated follow-up execution strategy is formed based on the resource matching priority sequence and the intervention intensity level.

[0028] The cumulative trend characteristics of the actual deviation curve are nonlinearly mapped to the progress deviation. When the cumulative trend accelerates and the deviation exceeds a preset baseline, an urgency amplification mechanism is triggered to generate a dynamically adjusted follow-up urgency score, including:

[0029] Time series analysis is performed on the actual deviation curve to extract the rate of change sequence reflecting the change of deviation magnitude over time. The second derivative of the rate of change sequence is calculated to obtain the acceleration characteristic value. When the acceleration characteristic value is positive, it is determined that the cumulative trend is in an accelerating upward state.

[0030] The deviation from the achieved schedule is compared with the preset baseline to obtain the result of the deviation exceeding the limit. The cumulative trend feature and the deviation from the achieved schedule are input into a nonlinear mapping relationship. The nonlinear mapping relationship generates a mapping output value by applying an exponential transformation to the cumulative trend feature and using the deviation from the achieved schedule as a weight adjustment parameter.

[0031] When the acceleration characteristic value is positive and the deviation exceeds the preset baseline, the mapping output value is multiplied by the urgency amplification factor to trigger the urgency amplification mechanism, and a dynamically adjusted follow-up urgency score is generated based on the triggering result of the urgency amplification mechanism.

[0032] The model and scheme unit is also used for:

[0033] The process of comparing the rehabilitation effect data with the expected trajectory in the individualized evolutionary trajectory, and using the comparison results to update the boundary parameters of the exercise-pain correlation model and the adjustment strategy of the load control scheme, includes:

[0034] The functional recovery indicators in the rehabilitation effect data are compared point by point with the expected recovery indicators at the corresponding time points of the expected trajectory in the individualized evolution trajectory. The deviation sequence between the actual recovery indicators and the expected recovery indicators is calculated, and the deviation sequence is analyzed to identify the advanced or lagging state of the recovery process.

[0035] The direction of boundary parameter adjustment is determined based on the trend analysis results of the deviation sequence. When the deviation sequence shows a state of recovery process ahead, the intensity of the pain trigger feature restriction in the movement-pain association model is reduced to relax the movement safety boundary. When the deviation sequence shows a state of recovery process lag, the intensity of the pain trigger feature restriction is increased to tighten the movement safety boundary. The boundary parameter update value is generated based on the direction of boundary parameter adjustment.

[0036] The updated boundary parameter values ​​are applied to the exercise-pain correlation model to update the allowable range of exercise intensity and duration defined by the exercise safety boundary. Based on the updated allowable range of exercise intensity and duration, the exercise intensity progression rhythm and duration extension strategy in the load control scheme are regenerated, thus completing the update of the boundary parameters of the exercise-pain correlation model and the adjustment strategy of the load control scheme.

[0037] The beneficial effects of this method are as follows:

[0038] This method integrates medical imaging and motion monitoring data to construct individualized evolutionary trajectories, enabling precise identification of bone structure characteristics and rehabilitation stages in elderly patients with hip fractures. This allows for the setting of functional recovery expectations that align with each patient's physiological recovery patterns. This assessment approach, based on individual dynamic data, avoids the drawbacks of a "one-size-fits-all" approach in traditional rehabilitation plans, significantly improving the accuracy of rehabilitation stage determination and the rationality of expected goals, thus laying a scientific foundation for subsequent personalized interventions.

[0039] By establishing a movement-pain correlation model and dynamically adjusting movement safety boundaries, the system can intelligently identify specific movement patterns that exacerbate pain and generate load control plans accordingly. This mechanism enables refined and adaptive management of rehabilitation training load, effectively controlling pain risks while ensuring rehabilitation effectiveness, avoiding secondary injuries or interruptions to the rehabilitation process caused by improper movement, and improving the safety of rehabilitation training and patient compliance.

[0040] By introducing a follow-up urgency scoring mechanism, the actual deviations from the rehabilitation plan are combined with the progress of functional recovery, and the status of community medical resources are linked to intelligently determine the timing and intensity of follow-up visits. This forms a resource-efficient differentiated follow-up strategy, ensuring that medical resources are prioritized for patients most in need of intervention, thereby improving the timeliness and effectiveness of follow-ups and optimizing the allocation of limited community medical resources. Attached Figure Description

[0041] Figure 1 This is a system architecture diagram of an AI-based question-and-answer-based rehabilitation plan adjustment system for elderly patients with hip fractures, as described in an embodiment of the present invention.

[0042] Figure 2 This is a diagram illustrating the dynamic adjustment architecture of the follow-up urgency score in an embodiment of the present invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0045] Figure 1 This is a system architecture diagram of an AI-based question-and-answer-based rehabilitation plan adjustment system for elderly hip fractures, as described in an embodiment of the present invention. Figure 1 As shown, the AI-based question-and-answer system for adjusting rehabilitation plans for hip fractures in the elderly includes:

[0046] The data analysis unit is used to acquire hip bone medical imaging data, motion monitoring data and rehabilitation history data of the target elderly user, analyze the hip bone medical imaging data to extract bone structure features, establish an individualized evolutionary trajectory in combination with the rehabilitation history data, and determine the current rehabilitation stage and functional recovery expectation based on the individualized evolutionary trajectory.

[0047] The exercise safety unit is used to construct an exercise-pain correlation model. By analyzing the correspondence between load changes and user pain feedback in the exercise monitoring data, it identifies the characteristics of exercise patterns that aggravate pain, dynamically adjusts the exercise safety boundary according to the current rehabilitation stage in the individualized evolution trajectory, and generates a load control scheme based on the exercise safety boundary.

[0048] The follow-up strategy unit is used to obtain the actual deviation of the exercise safety boundary based on the execution of the load control scheme, calculate the follow-up urgency score in combination with the expected progress of functional recovery, and determine the follow-up timing and intervention intensity based on the follow-up urgency score and the real-time availability of community medical resources, thereby forming a differentiated follow-up execution strategy.

[0049] The model and scheme unit is used to execute the follow-up execution strategy and collect rehabilitation effect data after follow-up. The rehabilitation effect data is compared with the expected trajectory in the individualized evolution trajectory. The comparison results are used to update the boundary parameters of the exercise-pain association model and the adjustment strategy of the load control scheme.

[0050] In one optional implementation, the data parsing unit is further configured to:

[0051] The hip bone medical imaging data is analyzed to extract bone structure features. An individualized evolutionary trajectory is established by combining this trajectory with the rehabilitation history data. Based on this individualized evolutionary trajectory, the current rehabilitation stage and expected functional recovery are determined, including:

[0052] Bone morphology analysis was performed on hip bone medical imaging data to extract structural degeneration features reflecting changes in bone density distribution and joint space. The evolution rate of these structural degeneration features was then analyzed by time series comparison to generate a degeneration rate curve characterizing the dynamic process of bone degeneration.

[0053] Extract the functional recovery range and recovery time corresponding to different rehabilitation stages from rehabilitation history data, correlate and map the functional recovery range with the degeneration rate curve, identify the nonlinear response relationship between bone degeneration rate and functional recovery ability, and construct an individualized evolutionary trajectory that reflects the sensitivity of individual bone status to rehabilitation response.

[0054] The structural degradation features are located in the individualized evolutionary trajectory. Based on the location results, the rehabilitation stage identifier of the current bone status is determined. Based on the response relationship in the individualized evolutionary trajectory, the functional recovery potential and expected recovery period corresponding to the rehabilitation stage identifier are deduced.

[0055] When performing bone morphology analysis on hip bone medical imaging data, three-dimensional image data of the hip joint region is obtained through CT scans or MRI imaging. This three-dimensional image data includes grayscale information of key anatomical structures such as the femoral neck, femoral head, and acetabulum. Statistical analysis is performed on the grayscale value distribution in the image data. By setting a grayscale threshold range, the bone region and soft tissue region are segmented. Regions with grayscale values ​​higher than the threshold are identified as cortical bone with higher bone density, while regions with grayscale values ​​in the middle range correspond to cancellous bone structure. When extracting structural degeneration features reflecting bone density distribution, multiple measurement sections are set along the longitudinal axis of the femoral neck, with each section spaced 3 to 5 millimeters apart. The average grayscale value of the bone region on each section is calculated as the bone mineral density index at that location. Simultaneously, the joint space width between the femoral head and acetabulum is measured. Measurement points are selected at the anterior, superior, and posterior parts of the joint space, and the distance of the joint space at each measurement point is recorded. The reduction in the joint space width reflects the degree of wear and tear of the articular cartilage and the state of joint degeneration.

[0056] When analyzing the rate of evolution of structural degeneration characteristics through time-series comparative analysis, hip bone medical imaging data of the target elderly users taken at different time points are retrieved. Bone mineral density (BMD) and joint space width data extracted at each time point are arranged chronologically. The change in BMD between adjacent time points is calculated, and the rate of bone mineral density decay is obtained by dividing the change by the time interval. This rate of decay reflects the speed of bone loss. The formula for calculating the rate of bone mineral density decay is: ,in The rate of bone mineral density loss, The bone mineral density index at time point i is... This refers to the bone mineral density index at the previous time point. The time interval is used. Similar processing is applied to changes in joint space width, calculating the rate of space narrowing, which indicates the progression of articular cartilage degeneration. When generating degeneration rate curves characterizing the dynamic process of bone degeneration, two independent rate change curves are plotted with time on the horizontal axis and the rate of bone mineral density decay and the rate of joint space narrowing on the vertical axes. Fluctuations in the rate values ​​on the degeneration rate curves reflect the impact of rehabilitation interventions or pathological progression on the bone degeneration process; positive or negative changes in the slope of the curves indicate an accelerating or decelerating trend in the degeneration process.

[0057] When extracting functional recovery range and recovery time corresponding to different rehabilitation stages from rehabilitation history data, the rehabilitation history data includes functional indicators such as hip joint range of motion, walking distance, and pain scores recorded by the user during the rehabilitation process. The rehabilitation process is divided into multiple stages, such as the early postoperative period, the weight-bearing transition period, and the functional reconstruction period. The start and end times of each stage are determined based on the surgery date and rehabilitation milestones. Within each rehabilitation stage, the starting and ending values ​​of functional indicators are statistically analyzed. The functional recovery range is calculated by the difference between the ending and starting values. For example, if the hip flexion angle recovers from 45 degrees in the early postoperative period to 90 degrees at the end of the weight-bearing transition period, then the functional recovery range for that stage is 45 degrees. The recovery time is calculated by the number of days between the start and end dates of the stage; for example, if a stage lasts 42 days, then the recovery time is 42 days.

[0058] When mapping the functional recovery amplitude to the degeneration rate curve, the time intervals of the rehabilitation stage are aligned with the corresponding time periods of the degeneration rate curve on the time axis. The average value of the degeneration rate curve within that time period is extracted as the characteristic value of bone degeneration rate for that stage. A scatter distribution is established between the functional recovery amplitude and the degeneration rate characteristic value to observe the correlation pattern between the two. When identifying the nonlinear response relationship between bone degeneration rate and functional recovery ability, it was found that stages with lower degeneration rates correspond to larger functional recovery amplitudes, but this correspondence is not a simple linear proportion but exhibits nonlinear characteristics. When the degeneration rate exceeds a certain critical value, the functional recovery amplitude decreases significantly, indicating that bone degeneration to a certain extent limits the potential for functional recovery.

[0059] When constructing an individualized evolutionary trajectory reflecting the sensitivity of an individual's bone status to rehabilitation response, a piecewise function or polynomial function is used to fit the nonlinear relationship between the functional recovery amplitude and the characteristic value of the degeneration rate. The mathematical expression of the individualized evolutionary trajectory is: ,in To the extent of functional recovery, The degradation rate is a characteristic value. , , , These are the fitting coefficients. This fitting function constitutes the mathematical expression of the individualized evolutionary trajectory. Different degradation rate intervals in the individualized evolutionary trajectory correspond to different response sensitivity coefficients, which are obtained by differentiating the evolutionary trajectory function: ,in The response sensitivity coefficient quantifies the intensity of bone status's response to rehabilitation intervention.

[0060] When locating structural degradation features within an individualized evolutionary trajectory, the latest hip bone medical imaging data is acquired at the current moment, and the current bone mineral density and joint space width are extracted using the aforementioned method. The degradation rate between the two most recent imaging datasets is calculated, and this degradation rate is used as the degradation rate feature value of the current bone state. Find the point on the degradation rate coordinate axis of the individualized evolutionary trajectory that corresponds to the current degradation rate eigenvalue. The coordinates of this point are... This reflects the relative position of the current bone status within the entire evolutionary trajectory. When determining the rehabilitation stage marker of the current bone status based on the positioning results, multiple stage intervals are pre-divided on the individualized evolutionary trajectory. Each interval corresponds to a different rehabilitation stage marker, such as a high-sensitivity period, a stable period, or a low-response period. By determining which stage interval the current location falls into, the corresponding rehabilitation stage marker is determined.

[0061] When extrapolating the functional recovery potential and expected recovery period corresponding to the rehabilitation stage identifier based on the response relationship in the individualized evolutionary trajectory, the functional recovery potential is obtained by querying the functional recovery amplitude value corresponding to the current deterioration rate feature value in the individualized evolutionary trajectory fitting function: This value represents the degree of functional improvement achievable with standard rehabilitation intervention under the current bone condition. The projected recovery cycle is based on the average recovery time within similar degeneration rate intervals in historical rehabilitation data. Rehabilitation stage records with degeneration rate characteristic values ​​close to the current value are extracted from historical data, and the actual recovery time of these stages is statistically analyzed, calculating the mean and standard deviation. ,in For the expected recovery period, The recovery time for the i-th similar stage, The number of similar stages is represented. The mean serves as the median estimate of the expected recovery period, while the standard deviation reflects the range of individual differences in the recovery period. When an individualized evolutionary trajectory indicates a low response period, the estimated functional recovery potential is lower, and the expected recovery period is correspondingly longer. This provides a quantitative basis for subsequent adjustments to the rehabilitation plan.

[0062] Throughout the analysis, multidimensional features of bone morphology were systematically extracted and quantified, including spatial distribution patterns of bone density, uneven changes in joint space, and the temporal dynamics of bone degeneration. These features, deeply correlated with rehabilitation history data, formed an evolutionary trajectory model capable of predicting individual rehabilitation responses. This model not only reflects the objective state of bone degeneration but, more importantly, establishes a quantitative mapping relationship between bone status and rehabilitation capacity, enabling personalized precision in determining rehabilitation stages and projecting expected functional recovery. By accurately locating the current bone status within the evolutionary trajectory, the characteristics of the user's current rehabilitation window can be identified, providing fundamental data support for developing targeted load control plans and follow-up strategies.

[0063] In one optional implementation, the functional recovery magnitude is correlated with the degeneration rate curve to identify the nonlinear response relationship between bone degeneration rate and functional recovery capacity, and an individualized evolutionary trajectory reflecting the sensitivity of individual bone status to rehabilitation response is constructed, including:

[0064] The functional recovery range is segmented according to the time dimension to obtain the recovery rate change characteristics in different time periods. The recovery rate change characteristics are paired with the degradation rate values ​​in the corresponding time periods in the degradation rate curve to establish a paired dataset between degradation rate and recovery rate.

[0065] Nonlinear fitting analysis is performed on the paired dataset to identify the differences in response patterns of functional recovery ability when the degradation rate is in different numerical ranges. A sensitivity function reflecting the change of response sensitivity with degradation rate is extracted, and an individualized evolutionary trajectory is constructed based on the sensitivity function.

[0066] In the rehabilitation process of elderly patients with hip fractures, a complex nonlinear relationship exists between the rate of bone degeneration and functional recovery. In the early stages of rehabilitation, even with a relatively rapid rate of bone degeneration, some patients still exhibit good functional recovery; however, in the later stages, as the rate of bone degeneration levels off, the marginal effect of functional recovery diminishes. Identifying this nonlinear response relationship requires establishing a precise paired analysis mechanism.

[0067] First, the functional recovery magnitude is segmented over time, dividing the entire rehabilitation cycle from fracture to the present into multiple consecutive time periods. The length of each time period can be set according to the characteristics of the rehabilitation stage. In the first four weeks after fracture, the time period can be set to a 7-day unit to capture the rapid recovery characteristics of the acute phase. In the subacute phase from week 5 to week 12, the time period can be extended to 14 days to reflect the gradual slowing of the functional recovery rate. In the chronic rehabilitation phase after three months, the time period can be further extended to 21 or 30 days. Within each time period, the absolute magnitude of functional recovery for that time period is obtained by calculating the difference in functional scores between the start and end times of that period. Functional scores can be a comprehensive quantitative result of multiple dimensions such as hip joint range of motion, walking distance, and balance ability. Dividing the absolute magnitude of functional recovery by the time period length yields the average recovery rate within that time period. Further analysis of the changes in recovery rate between adjacent time periods extracts acceleration or deceleration characteristics of the recovery rate, forming a recovery rate change sequence.

[0068] After acquiring the characteristics of the recovery rate changes, they need to be precisely paired with the degradation rate curve, which reflects the decay trend of parameters such as bone mineral density and trabecular microstructure over time. For each segmented time period, the degradation rate value at the corresponding moment of that time period is extracted. The degradation rate can be calculated based on continuous medical imaging scans by comparing bone mineral density measurements at different times to calculate the amount of bone mineral density decrease per unit time. During the pairing process, the accuracy of time alignment is ensured, meaning that the time period corresponding to the recovery rate change characteristics is completely consistent with the sampling time period of the degradation rate values. The recovery rate of each time period is recorded as follows. The corresponding degradation rate is recorded as follows ,in Represents the time period sequence number. Create a collection containing multiple groups. The paired datasets of data pairs reflect the actual performance of patients' functional recovery rates under different rates of degeneration.

[0069] When performing nonlinear fitting analysis on paired datasets, the degradation rate is first divided into different numerical intervals. Based on clinical experience, degradation rates can be categorized into a low-speed degradation interval (e.g., monthly bone mineral density decline rate less than 0.5%), a medium-speed degradation interval (monthly decline rate between 0.5% and 1.5%), and a high-speed degradation interval (monthly decline rate exceeding 1.5%). The distribution characteristics of the recovery rate for all paired data points within each interval are statistically analyzed, and the mean, variance, and skewness are calculated. In the low-speed degradation interval, the recovery rate exhibits significant individual variability, with some patients showing rapid recovery while others recover slowly, and the variance is large. In the high-speed degradation interval, the recovery rate is generally suppressed, the mean decreases, and the variance narrows. By comparing the statistical characteristics of the recovery rate within different intervals, differences in response patterns are identified.

[0070] To quantitatively describe this difference, a sensitivity function is introduced. This function reflects the degree to which functional recovery capability responds to changes in the rate of degradation. The sensitivity function can be constructed as a piecewise function or a continuous nonlinear function. One implementation method is to use an exponential decay model, i.e. ,in This represents the baseline recovery capacity under ideal conditions (without bone degeneration). The attenuation coefficient reflects the strength of the inhibition of recovery ability by the degradation rate. The parameters are determined by least-squares fitting on the paired dataset. and The optimal value. Another implementation is to use a multinomial model, that is... This method captures the complex nonlinear characteristics of the response relationship through higher-order terms. During the fitting process, the sum of squared residuals between the fitted curve and the actual data points is calculated, and the parameter combination that minimizes the residuals is selected.

[0071] Based on the sensitivity function, an individualized evolutionary trajectory is further constructed. The core of this trajectory is to predict the expected path of functional recovery over future time periods, considering the ongoing impact of bone degeneration. Taking the current moment as the starting point, the degeneration rate is extrapolated based on the trend of the degeneration rate curve to predict the degeneration rate at various future time periods. Substitute the predicted degradation rate into the sensitivity function to calculate the corresponding sensitivity value. By combining the patient's current baseline functional status with the theoretical recovery potential offered by the rehabilitation training plan, the expected functional recovery increment is calculated. The expected recovery increment can be expressed as the product of the sensitivity value and the theoretical recovery potential; that is, under a specific rate of deterioration, the actual achievable recovery is equal to the theoretical potential modulated by the sensitivity. The expected recovery increments for each time period are summed to the functional status baseline to form a continuous functional recovery prediction curve, which is the core component of the individualized evolutionary trajectory.

[0072] In the evolutionary trajectory, key turning points in the rehabilitation stages also need to be marked. By analyzing the derivative of the sensitivity function, the critical value of the deterioration rate at which a significant change in the rate of sensitivity change occurs can be identified. When the deterioration rate exceeds a certain critical value, the sensitivity function drops sharply, indicating that functional recovery ability has entered a phase of accelerated decline. The time points corresponding to these critical values ​​are marked on the evolutionary trajectory as the basis for dividing the rehabilitation stages. For example, if the prediction shows that the deterioration rate will cross the critical value in the eighth week, this time point can be marked as the dividing point from the subacute phase to the chronic phase, indicating that intensive rehabilitation intervention is needed before then to delay the decline in sensitivity.

[0073] The construction of individualized evolutionary trajectories also needs to integrate the individual differences of patients. For patients with a history of osteoporosis, their baseline rate of degeneration is usually higher, and the attenuation coefficient of the sensitivity function... Significantly greater than in patients without osteoporosis. A hierarchical evolutionary trajectory model was established by fitting sensitivity functions to different patient subgroups. In practical applications, appropriate sensitivity function parameters are selected based on the patient's subgroup to generate a personalized evolutionary path for that patient. Furthermore, considering the influence of nutritional status, comorbidities, and other factors on bone degeneration and functional recovery, correction factors can be introduced into the sensitivity function. For example, patients with low serum calcium levels have a lower baseline recovery capacity in their sensitivity function. A certain percentage needs to be lowered to reflect the weakening effect of nutritional deficiencies on recovery potential.

[0074] The individualized evolutionary trajectory constructed using the above methods not only reflects the quantitative correlation between bone degeneration and functional recovery but also predicts the path of future rehabilitation progress, providing a scientific basis for the dynamic adjustment of rehabilitation plans. When actual rehabilitation progress deviates from the expected trajectory, comparative analysis can identify the causes of deviation, determining whether it is due to factors such as accelerated degeneration rate, insufficient rehabilitation training intensity, or decreased patient compliance, thereby enabling targeted interventions. The dynamic updating mechanism of the evolutionary trajectory ensures that the model can be continuously optimized with the accumulation of new data, improving predictive accuracy and clinical applicability.

[0075] In one optional implementation, the motion safety unit is further configured to:

[0076] A movement-pain correlation model is constructed. By analyzing the correspondence between load changes and user pain feedback in the movement monitoring data, the characteristics of movement patterns that exacerbate pain are identified. The movement safety boundary is dynamically adjusted based on the current rehabilitation stage in the individualized evolutionary trajectory. A load control scheme is generated based on the movement safety boundary, including:

[0077] The load change sequence is extracted from the exercise monitoring data. The load change sequence is time-aligned with the user pain feedback value at the corresponding time. The time-aligned load change sequence and pain feedback value are correlated and analyzed to identify the load change pattern that causes the pain feedback value to exceed the baseline level. The exercise intensity threshold and duration threshold in the load change pattern are extracted as pain triggering features. Based on the pain triggering features, an exercise-pain correlation model is constructed.

[0078] Obtain the current rehabilitation stage identifier in the individualized evolutionary trajectory, adjust the pain triggering features in the movement-pain association model according to the functional recovery potential corresponding to the current rehabilitation stage identifier, generate a movement safety boundary that matches the current rehabilitation stage, and generate a load control scheme based on the movement safety boundary.

[0079] For rehabilitation training of elderly patients with hip fractures, the collection of motion monitoring data covers multi-dimensional physiological and motor indicators. Wearable devices record in real-time the patient's hip joint range of motion, limb weight-bearing time, gait symmetry parameters, and peak acceleration during daily activities such as walking, standing, and bedside transfers. These parameters form the basic data source for the load change sequence. During data collection, the pain feedback values ​​reported by the patient using a pain scale (such as the visual analog scale) are simultaneously recorded. These values ​​are typically between 0 and 10, where 0 represents no pain and 10 represents severe pain. To ensure an accurate correspondence between the load change sequence and the pain feedback values, a timestamp alignment technique is used to match the load data recorded by the motion monitoring device with the pain scores entered by the patient on the smart terminal at the millisecond level, eliminating temporal discrepancies caused by data acquisition delays.

[0080] Analyzing the load variation sequence requires converting continuous motion parameters into quantifiable load indices. For example, the maximum flexion angle of the hip joint during a single walking cycle, the peak vertical ground reaction force on the weight-bearing limb, and the duration of continuous weight-bearing are integrated into a comprehensive load index. This index can be calculated using a weighted summation method, where the weight coefficient of each motion parameter is determined based on its influence on hip fracture healing. For patients who have just undergone internal fixation surgery, the weight coefficient for the peak weight-bearing force is typically set higher, while for patients who have entered the bony healing stage, the weight coefficient for the joint range of motion is correspondingly increased. After time alignment, a series of data records containing timestamps, the comprehensive load index, and corresponding pain scores are obtained.

[0081] The core of association analysis lies in identifying the causal relationship between load changes and pain feedback. This is achieved by using a sliding time window technique to detect abrupt changes in the load index within a load change sequence. For example, when a patient's average load index suddenly increases by more than 30% over a continuous 5-minute period, this moment is marked as a load mutation event. Subsequently, changes in the patient's reported pain score within 15 minutes following this mutation event are retrieved. If the pain score increases by 2 points or more compared to baseline (typically the patient's pain score at rest), an association is established between the load mutation event and increased pain. Statistical analysis of multiple such events identifies highly reproducible load change patterns. These patterns are characterized by specific exercise intensity thresholds, such as hip flexion exceeding 70 degrees for more than 3 minutes, or a vertical force on the weight-bearing limb exceeding 60% of body weight for more than 10 seconds.

[0082] Extracting pain trigger features requires refining key parameters from identified load change patterns. The exercise intensity threshold is defined as the minimum load index value that triggers a pain feedback score exceeding the baseline level. For example, retrospective analysis of 20 pain exacerbation events revealed that when the composite load index exceeded 45 units, it led to an increase in pain scores in 80% of cases; therefore, 45 units was set as the exercise intensity threshold. The duration threshold refers to the shortest time the load index is maintained at that intensity level. If statistics show that a load index above 45 units triggers pain after 2 minutes, while 1 minute does not result in significant pain exacerbation, then 2 minutes is set as the duration threshold. These two thresholds together constitute the pain trigger feature, serving as the core parameters of the exercise-pain association model.

[0083] The construction of the motion-pain association model is based on the mapping relationship between pain triggering characteristics and individual patient characteristics. This model can be expressed as: when the load index... Exceeding the exercise intensity threshold And duration Exceeding the duration threshold Pain feedback value at that time Predicted value It can be estimated in the following ways: ,in The patient's baseline pain score, This is the pain sensitivity coefficient, which varies depending on the patient's age, degree of osteoporosis, and pain tolerance. The model determines the specific values ​​of each parameter by fitting historical data, minimizing the mean square error between the predicted pain value and the actual pain feedback.

[0084] The current rehabilitation stage marker in the individualized evolutionary trajectory reflects the patient's functional recovery status. Rehabilitation stages are typically divided into an early protection phase, a mid-stage functional recovery phase, and a late-stage intensive phase. In the early protection phase, fracture healing is not yet stable, and the potential for functional recovery is limited. At this time, the load level must be strictly limited to avoid loosening of the internal fixation device or fracture displacement. In the mid-stage functional recovery phase, callus formation gradually increases, and the load can be moderately increased to promote bone remodeling and muscle strength recovery. The late-stage intensive phase allows for load training at near-normal activity levels. Based on the current rehabilitation stage marker, the pain trigger characteristics in the exercise-pain association model are dynamically adjusted. For example, in the early protection phase, the exercise intensity threshold is adjusted... Reduce by 20%, duration threshold The threshold is shortened to 50% of its original value to provide a more stringent pain warning. In the later intensive phase, the exercise intensity threshold can be increased by 15%, and the duration threshold extended to 150% of its original value, allowing patients to train within a wider range of motion.

[0085] The assessment of functional recovery potential integrates bone healing imaging scores, the percentage of joint range of motion recovery, and muscle strength test results. For example, X-rays are used to assess callus continuity; if the callus covers more than 75% of the fracture line, it is considered to be in the intermediate functional recovery stage. Combined with hip flexion range of motion recovering to 60% of the healthy side and quadriceps strength reaching MMT grade 4, the current functional recovery potential is comprehensively assessed as moderate. Based on this potential level, the pain sensitivity coefficient in the movement-pain correlation model is adjusted. For patients with high recovery potential, the coefficient can be appropriately reduced to make the predicted value of pain caused by the same load change lower, thus encouraging patients to increase their exercise.

[0086] The exercise safety boundary is generated based on adjusted pain triggering characteristics. The safety boundary is defined as the maximum allowable load index and longest duration combination without triggering pain feedback values ​​exceeding an acceptable threshold (typically set as the baseline pain score plus 3 points). By substituting the adjusted exercise intensity threshold and duration threshold into the exercise-pain correlation model, the solution is obtained... The load parameter range forms a two-dimensional safety zone. This zone is represented as a closed curve in the load index-duration coordinate system, and any combination of motions within the curve is considered a safe load.

[0087] The load control program develops specific rehabilitation training guidelines based on the exercise safety boundary. The program includes the upper limit of the load for a single training session, recommended types of training movements, and the total daily training duration. For example, for patients in the mid-stage of functional recovery, if the exercise safety boundary defines the upper limit of the load index as 55 units and the upper limit of the duration as 5 minutes, the load control program can stipulate that the walking speed for each walking training session should not exceed 60 steps per minute, the continuous walking time should not exceed 5 minutes, and the total daily training time should be controlled within 30 minutes. The program also needs to specify a load escalation strategy, such as increasing the upper limit of the load by 5% weekly based on the patient's training completion without increased pain, gradually expanding the exercise safety boundary to promote functional recovery. Simultaneously, the program incorporates a real-time monitoring mechanism. When the wearable device detects that the load index is approaching the safety boundary, it automatically prompts the patient to reduce the intensity of exercise through vibration or voice to avoid triggering a pain response.

[0088] In an optional implementation, the follow-up strategy unit is further configured to:

[0089] Based on the execution status of the load control scheme, the actual deviation from the exercise safety boundary is obtained. Combined with the expected progress of functional recovery, a follow-up urgency score is calculated. Based on the follow-up urgency score and the real-time availability of community medical resources, the timing and intensity of follow-up are determined, forming a differentiated follow-up execution strategy, including:

[0090] Monitor the actual execution data of the load control scheme, quantify the deviation between the motion intensity sequence in the actual execution data and the allowable range of the motion safety boundary, identify the time distribution pattern of the deviation and the cumulative trend of the deviation magnitude, and generate the actual deviation degree curve;

[0091] The expected recovery trajectory and the current actual recovery trajectory are obtained. The deviation area between the trajectories is calculated as the progress deviation. The cumulative trend characteristics of the actual deviation curve are nonlinearly mapped to the progress deviation. When the cumulative trend accelerates and the deviation exceeds the preset baseline, the urgency amplification mechanism is triggered to generate a dynamically adjusted follow-up urgency score.

[0092] Based on the follow-up urgency score, a resource matching priority sequence is defined. Follow-up resources are selected from the medical personnel closest to the available time and the medical equipment with the lowest load status according to the resource matching priority sequence. At the same time, the intervention intensity level is determined according to the numerical gradient of the follow-up urgency score. A differentiated follow-up execution strategy is formed based on the resource matching priority sequence and the intervention intensity level.

[0093] In the rehabilitation management of elderly patients with hip fractures, the quality of implementation of the load control program directly affects the stability of the rehabilitation process, while the timing and intensity of follow-up interventions determine the response efficiency to abnormal situations. To achieve precise control of follow-up strategies, a closed-loop management mechanism needs to be constructed from three levels: implementation monitoring, urgency assessment, and resource allocation.

[0094] Continuous monitoring of the actual execution data of the load control scheme is conducted, collecting exercise intensity sequences of target elderly users during rehabilitation training. These sequences include daily peak load, duration, and frequency. The collected exercise intensity sequences are compared with the allowable range defined by the exercise safety boundary, time-by-time. For each training period, the difference between the actual intensity and the safety upper limit is calculated. A positive difference indicates overload deviation, while a negative difference indicates the load is within the safe range. All deviation events are timestamped, and the temporal distribution of deviations is statistically analyzed to identify whether deviations are concentrated in specific time periods or scattered throughout the training cycles of the day. Simultaneously, changes in the deviation magnitude are tracked, and the deviation magnitudes over multiple consecutive days are arranged chronologically to observe whether the magnitude shows an increasing trend. If the deviation magnitude fluctuates but does not increase significantly, it indicates that the deviation is sporadic; if the deviation magnitude increases daily, it suggests a risk of systemic load runaway. Based on the time distribution pattern and the cumulative trend of the magnitude, the cumulative daily deviation magnitude is used as the ordinate, and time as the abscissa to plot the actual deviation degree curve. The slope of this curve reflects the rate of deviation accumulation, and the fluctuation of the curve reflects the regularity of the deviation.

[0095] The expected recovery trajectory, defined by functional recovery expectations, is extracted from the individualized evolutionary trajectory. This trajectory, with rehabilitation time as the independent variable and functional score as the dependent variable, depicts the trend of hip function recovery in elderly users under ideal conditions. Simultaneously, the actual recovery trajectory at the current stage is acquired. This trajectory is constructed by periodically measuring functional indicators such as hip joint range of motion, walking speed, and balance stability. The expected and actual recovery trajectories are superimposed on the same time coordinate system, and the vertical distance between the two trajectories is calculated. The vertical distance is integrated over all time points throughout the entire rehabilitation cycle; the resulting integral value is the area of ​​deviation between the trajectories. The larger the area, the more severe the deviation between the actual recovery progress and the expected goal. The area of ​​deviation is used as the deviation from the achieved progress, and a comprehensive evaluation is conducted in conjunction with the cumulative trend characteristics of the actual deviation curve.

[0096] The slope change rate within the recent time window is extracted from the actual deviation curve. If the slope change rate is greater than zero and continues to increase, the cumulative trend is determined to be accelerating. A preset baseline for the deviation in progress is set, determined based on the functional recovery tolerance at different rehabilitation stages. When the actual deviation exceeds this baseline value, an urgency amplification mechanism is activated. Under this mechanism, the accelerating upward characteristic of the cumulative trend is used as a weighting amplification factor to nonlinearly map the baseline urgency score. Specifically, the deviation in progress is normalized to the range of 0 to 1, the slope change rate of the cumulative trend is converted into an amplification coefficient, and the two are combined through a nonlinear function to generate a dynamically adjusted follow-up urgency score. The score range is set between 0 and 100, with a higher score indicating a more urgent need for follow-up intervention.

[0097] Based on the generated follow-up urgency score, a resource matching strategy is formulated. The follow-up urgency score is divided into multiple level intervals according to its numerical value, and each level interval corresponds to a different resource matching priority sequence. For users in the high-score interval, the highest priority is assigned, and they are given priority to use community medical resources. For users in the medium-score interval, the medium priority is assigned, and they are scheduled after the resource needs of high-priority users are met. For users in the low-score interval, the lower priority is assigned, and the regular follow-up rhythm is adopted. After determining the priority sequence, the real-time availability status of community medical resources is retrieved. This status includes the medical staff's shift schedule, workload saturation, and the usage status of medical equipment.

[0098] The system extracts the available time slots for all medical personnel, filters out the closest available time slots, and directly schedules these slots for high-priority users. For medium- and low-priority users, if the closest available time slot is already occupied, the next closest available time slot is scheduled. Simultaneously, the system assesses the load status of medical equipment by calculating the usage frequency and average usage time of each device in past time slots. The device with the lowest load status is selected for follow-up interventions, ensuring balanced utilization of equipment resources.

[0099] The intervention intensity level is determined based on the numerical gradient of the follow-up urgency score. The follow-up urgency score is divided into multiple gradient intervals, and each gradient interval is mapped to a specific intervention intensity level. For users with scores in the high gradient interval, a high-intensity intervention level is set, which includes measures such as increasing the frequency of follow-up visits, extending the duration of each follow-up visit, and introducing multidisciplinary joint assessment. For users with scores in the middle gradient interval, a medium-intensity intervention level is set, which includes measures such as maintaining the regular follow-up frequency but strengthening workload monitoring and increasing the number of remote guidance sessions. For users with scores in the low gradient interval, a low-intensity intervention level is set, which maintains the regular follow-up rhythm and only provides additional intervention when users actively report abnormalities.

[0100] The specific measures for each level of intervention intensity are broken down into actionable instructions. For example, under a high-intensity intervention level, the instructions specify three follow-up visits per week, each lasting at least 45 minutes, with rehabilitation therapists and orthopedic surgeons jointly participating in the assessment. Under a moderate-intensity intervention level, the instructions specify two follow-up visits per week, each lasting 30 minutes, with the assessment conducted solely by a rehabilitation therapist. Under a low-intensity intervention level, the instructions specify follow-up visits every two weeks, each lasting 20 minutes, using a standardized assessment process.

[0101] This strategy combines resource matching priorities with intervention intensity levels to create differentiated follow-up execution strategies. Personalized follow-up plans are generated for each user, clearly specifying the follow-up time, participants, equipment used, assessment content, and subsequent adjustment suggestions. For high-priority, high-intensity intervention users, the initial follow-up is scheduled during the nearest available time slot, covering comprehensive functional assessment, pain pattern analysis, and investigation of load execution deviations. Based on the assessment results, the parameters of the load control plan are adjusted immediately. For medium-priority, medium-intensity intervention users, follow-up is scheduled during the next nearest available time slot, focusing on tracking key functional indicators and verifying load execution. The need for fine-tuning the load control plan is determined based on the tracking results. For low-priority, low-intensity intervention users, follow-up is conducted within the regular follow-up cycle, using standardized assessment scales. The plan adjustment process is triggered only when assessment results show abnormalities. All follow-up execution strategies are synchronized to the community healthcare management system after generation. The system automatically sends task notifications to relevant medical personnel and reserves equipment usage time slots to ensure that follow-up interventions are implemented as planned. By quantitatively tracking the actual degree of deviation, dynamically assessing the urgency of follow-up visits, and prioritizing resource allocation, timely response and precise intervention can be achieved for abnormal situations in the rehabilitation process, ensuring the safety and efficiency of rehabilitation for elderly patients with hip fractures.

[0102] In one optional implementation, the cumulative trend characteristics of the actual deviation curve are nonlinearly mapped to the progress deviation. When the cumulative trend accelerates and the deviation exceeds a preset baseline, an urgency amplification mechanism is triggered to generate a dynamically adjusted follow-up urgency score, including:

[0103] Time series analysis is performed on the actual deviation curve to extract the rate of change sequence reflecting the change of deviation magnitude over time. The second derivative of the rate of change sequence is calculated to obtain the acceleration characteristic value. When the acceleration characteristic value is positive, it is determined that the cumulative trend is in an accelerating upward state.

[0104] The deviation from the achieved schedule is compared with the preset baseline to obtain the result of the deviation exceeding the limit. The cumulative trend feature and the deviation from the achieved schedule are input into a nonlinear mapping relationship. The nonlinear mapping relationship generates a mapping output value by applying an exponential transformation to the cumulative trend feature and using the deviation from the achieved schedule as a weight adjustment parameter.

[0105] When the acceleration characteristic value is positive and the deviation exceeds the preset baseline, the mapping output value is multiplied by the urgency amplification factor to trigger the urgency amplification mechanism, and a dynamically adjusted follow-up urgency score is generated based on the triggering result of the urgency amplification mechanism.

[0106] like Figure 2As shown, the method includes:

[0107] During the implementation of the rehabilitation plan, a detailed analysis of the temporal evolution characteristics of the deviation degree is required. The collected actual deviation degree data are arranged in chronological order to form a discrete time series. ,in Indicates the first The deviation value at each time point is calculated. A sliding time window is applied to the sequence, with a window length of 5 time points. Within each window, the difference between adjacent data points is calculated. The specific calculation method is as follows: The rate of change sequence can be obtained through this difference operation. This sequence reflects the rate of change of the deviation degree across different time periods. To further identify whether the deviation trend exhibits an accelerating characteristic, the rate of change sequence is differenced again to calculate the second-order difference value. The second-order difference value is the acceleration characteristic value. When three or more consecutive acceleration characteristic values ​​are positive, it is determined that the current cumulative trend is accelerating, indicating that the actual rehabilitation performance of elderly users is deteriorating rapidly and requires close attention.

[0108] When analyzing deviations from the expected progress, it is first necessary to clarify the method for determining the preset baseline. Based on clinical research data on hip fracture rehabilitation in the elderly, the functional recovery target is typically set at 60% of preoperative activity level in the first 6 weeks of rehabilitation, and 80% in weeks 6 to 12. Differentiated baseline thresholds are set for different rehabilitation stages: 15% for the early rehabilitation stage (within 4 weeks post-surgery), 10% for the mid-stage (4 to 8 weeks post-surgery), and 8% for the late rehabilitation stage (more than 8 weeks post-surgery). The difference between the actual functional recovery progress and the expected target is calculated. When this difference exceeds the baseline threshold for the corresponding rehabilitation stage, the deviation is marked as "exceeding the preset baseline." For example, an elderly user in the 6th week post-surgery should have achieved 55% functional recovery, but the actual measured progress is only 38%, a deviation of 17%, exceeding the 10% baseline for the mid-stage rehabilitation, thus being judged as exceeding the limit.

[0109] To achieve a non-linear mapping between cumulative trend characteristics and progress deviations, an adaptive mapping relationship needs to be constructed. This involves quantifying the cumulative trend characteristics into a trend strength index. The index is calculated by averaging the acceleration characteristic values ​​over the most recent seven time points and then normalizing them. The normalized value ranges from 0 to 1. An exponential transformation is applied to the trend strength index using the natural exponential function. Perform the conversion, where The adjustment factor is set between 1.5 and 2.5 based on the individual sensitivity of different elderly users. This will determine the progress deviation. This is converted into a weighted adjustment parameter, specifically by dividing the deviation by the baseline threshold of the current rehabilitation stage to obtain the normalized deviation. ,in This represents the baseline threshold corresponding to the rehabilitation stage. Mapping output value. The calculation is performed in product form. This formula can simultaneously reflect the accelerating characteristics of the deviation trend and the degree of deviation exceeding the limit. When both exist at the same time, it will produce a significant amplification effect.

[0110] In the determination process for triggering the urgency amplification mechanism, two core conditions must be met simultaneously. The first condition is a positive acceleration characteristic value, indicating that the rate of increase in deviation is accelerating. The second condition is that the deviation exceeds the preset baseline, indicating that the gap between the actual recovery progress and the expected goal has reached a level requiring intervention. Only when both conditions are met is the mapped output value multiplied by the urgency amplification factor. Urgency Amplification Factor The amplification factor is set based on the age, underlying health condition, and fall risk level of elderly users. Users aged 70-75 with no underlying diseases have an amplification factor of 1.2; users aged 75-80 or with one underlying disease have an amplification factor of 1.5; and users over 80 or with two or more underlying diseases have an amplification factor of 1.8. This is calculated... The original score for urgency was obtained after amplification.

[0111] The dynamically adjusted follow-up urgency score requires multi-dimensional adjustments based on the original score. Considering the adherence factors of elderly users, the completion rate of recent rehabilitation training is assessed. When the completion rate is below 75%, a correction coefficient of 0.15 is added to the original score; when the completion rate is between 75% and 85%, a correction coefficient of 0.08 is added; and when the completion rate is above 85%, no adjustment is made. Simultaneously, the trend of pain feedback is considered. Comparing the pain scores reported by users in the three most recent follow-ups, if the pain score shows a continuous upward trend, a further correction coefficient of 0.12 is added to the original score. The original score is then summed with all correction coefficients and normalized to a range of 0 to 10 to obtain the final follow-up urgency score. A score of 0 to 3 indicates low urgency, where regular follow-up frequency can be maintained; a score of 3 to 6 indicates moderate urgency, where it is recommended to shorten the follow-up interval to 70% of the original plan; a score of 6 to 8 indicates high urgency, requiring the follow-up interval to be shortened to 50% of the original plan and the frequency of telephone inquiries to be increased; a score of 8 or above indicates extremely high urgency, where community healthcare workers should be immediately arranged to conduct a home assessment or guide the user to a medical institution for professional examination.

[0112] In a practical application scenario, suppose a 78-year-old user is 7 weeks post-hip fracture surgery. Their deviation data for the most recent 7 time points are 12%, 15%, 19%, 24%, 31%, 39%, and 49%, respectively. Differential calculation yields a rate of change sequence of 3%, 4%, 5%, 7%, 8%, and 10%, while secondary differencing yields an acceleration characteristic value sequence of 1%, 1%, 2%, 1%, and 2%, all positive, indicating an accelerating cumulative trend. The trend strength index is calculated, averaged at 1.4%, and normalized to 0.7. The user's expected functional recovery progress should reach 65%, but the actual measurement is 48%, resulting in a deviation of 17%, exceeding the 10% baseline for the mid-stage rehabilitation. This deviation exceeds the preset baseline. The normalized deviation is 1.7, the adjustment coefficient is set to 2.0, and the calculated mapping output value is... Based on the user's age and the presence of an underlying medical condition, the urgency amplification factor was set to 1.5, resulting in an amplified original score of 16.44. Considering the user's recent training completion rate of 72%, a correction factor of 0.15 was added, and the pain score showed an upward trend, leading to a correction factor of 0.12. The total corrected score was 16.71. After normalization to the 0-10 range, the follow-up urgency score was 8.4, indicating extremely high urgency. The system immediately triggered the community healthcare home visit assessment process and generated a detailed deviation analysis report in the follow-up record for healthcare personnel's reference.

[0113] In an optional implementation, the model and scheme unit is further used for:

[0114] The process of comparing the rehabilitation effect data with the expected trajectory in the individualized evolutionary trajectory, and using the comparison results to update the boundary parameters of the exercise-pain correlation model and the adjustment strategy of the load control scheme, includes:

[0115] The functional recovery indicators in the rehabilitation effect data are compared point by point with the expected recovery indicators at the corresponding time points of the expected trajectory in the individualized evolution trajectory. The deviation sequence between the actual recovery indicators and the expected recovery indicators is calculated, and the deviation sequence is analyzed to identify the advanced or lagging state of the recovery process.

[0116] The direction of boundary parameter adjustment is determined based on the trend analysis results of the deviation sequence. When the deviation sequence shows a state of recovery process ahead, the intensity of the pain trigger feature restriction in the movement-pain association model is reduced to relax the movement safety boundary. When the deviation sequence shows a state of recovery process lag, the intensity of the pain trigger feature restriction is increased to tighten the movement safety boundary. The boundary parameter update value is generated based on the direction of boundary parameter adjustment.

[0117] The updated boundary parameter values ​​are applied to the exercise-pain correlation model to update the allowable range of exercise intensity and duration defined by the exercise safety boundary. Based on the updated allowable range of exercise intensity and duration, the exercise intensity progression rhythm and duration extension strategy in the load control scheme are regenerated, thus completing the update of the boundary parameters of the exercise-pain correlation model and the adjustment strategy of the load control scheme.

[0118] In the dynamic adjustment of the rehabilitation plan, a core step in achieving closed-loop optimization is to meticulously compare the collected rehabilitation effect data with the pre-established expected trajectory in the individualized evolutionary trajectory. The rehabilitation effect data includes multiple dimensions of functional recovery indicators, including hip joint range of motion, gait symmetry index, balance score, and visual analog scale (VAS) score for pain. These indicators are collected at different time intervals during the rehabilitation process, forming time-series data. The individualized evolutionary trajectory, based on initial assessment data and historical statistical models, predicts the expected recovery indicators at each time point; these expected indicators constitute the benchmark for evaluating the actual rehabilitation effect.

[0119] When performing point-by-point comparisons, it is essential to first ensure the accuracy of time alignment. Since the actual collected rehabilitation effect data may have a time offset, a timestamp matching mechanism is used to associate the actual collection time with the corresponding time in the expected trajectory. For the hip joint range of motion indicator, assuming the first... The actual buckling angle at each acquisition time is: The expected buckling angle at the corresponding moment in the expected trajectory is The deviation value at that moment is Repeat this calculation for all acquisition times to obtain the complete deviation sequence. ,in This represents the total number of data collections. The same method is applied to other functional recovery indicators to generate corresponding deviation sequences.

[0120] Trend analysis uses a moving window method to identify the status of the recovery process, with the window width set to [value missing]. For each data collection point, calculate the mean of the deviation values ​​within the window. With linear regression slope .when and When the recovery process is in an advanced state, it is determined that the actual recovery speed is faster than expected; when and When this occurs, it is determined to be a state of delayed recovery, indicating that the actual recovery speed is slower than expected. Threshold and Based on the clinical significance differences of different functional recovery indicators, a setting of 5 degrees can be used for hip joint range of motion, and a setting of 1 point can be used for pain scores. In practical applications, a weighted voting mechanism is used to determine the overall recovery progress status, taking into account the trend analysis results of multiple functional recovery indicators.

[0121] The direction of boundary parameter adjustment needs to be determined by adopting differentiated strategies based on the recovery process status. The intensity of pain triggering features in the motion-pain association model is limited by threshold parameters. This parameter defines the minimum pain score required to trigger a pain warning. When a state of accelerated recovery is identified, it indicates that the user's actual tolerance is better than expected, and the pain level can be appropriately lowered. The score was lowered from 3 to 2.5, relaxing the safety boundaries of exercise to allow for more aggressive rehabilitation training. Conversely, when a lag in the recovery process is identified, it indicates that the user's actual tolerance is lower than expected or there are potential risks, requiring improvement. The score was increased from 3 to 3.5, tightening the safety boundaries of exercise to avoid secondary injuries caused by excessive load.

[0122] Generating boundary parameter update values ​​requires quantifying the adjustment magnitude and defining adjustment coefficients. Its value is related to the statistical characteristics of the deviation series. Calculate the standard deviation of the deviation series. Deviation from the mean Adjustment coefficients are obtained through Confirmed, among which This is an empirical adjustment factor, typically ranging from 0.1 to 0.3. The updated boundary parameters are calculated as follows: ,in The function is used to determine the adjustment direction. This adaptive adjustment mechanism based on statistical characteristics can flexibly control the adjustment intensity according to the severity of the deviation, avoiding system oscillations caused by over-adjustment.

[0123] After applying the updated boundary parameters to the motion-pain correlation model, the permissible range of motion intensity defined by the motion safety boundary changes accordingly. The permissible range of motion intensity is determined by the maximum load intensity. With minimum effective strength The maximum load intensity is recalculated based on the adjusted pain trigger threshold. In the early stages of rehabilitation, the maximum load intensity is limited to 30% of body weight; this can be increased to 40% of body weight as the boundary parameters are relaxed. The allowable duration also needs adjustment, specifying the maximum duration of a single training session. Adjustments were made based on the updated boundary parameters, extending the original 20 minutes to 25 minutes or shortening it to 15 minutes to ensure effective training within the new safety boundaries.

[0124] The intensity progression rhythm in the load control scheme is re-planned based on the updated allowable intensity range. The progression rhythm is defined as the rate of increase in intensity within each training cycle. The original plan was to increase the workload by 5% per week. With the boundary parameters relaxed, the pace could be accelerated to 8% per week to match the user's advanced recovery capabilities. Conversely, when the boundary parameters tightened, the pace needed to be slowed to 3% per week to avoid exceeding the user's current capacity. The duration extension strategy was also adjusted accordingly. The original plan increased the duration of each training session by 5 minutes every two weeks; with the boundary relaxed, this could be adjusted to 5 minutes per week; and with the boundary tightened, it could be adjusted to 5 minutes every three weeks.

[0125] During the update process, it is also necessary to consider the coupling relationship between different functional recovery indicators. Rapid recovery of hip joint range of motion is accompanied by a lag in pain control, requiring a balance between the optimization objectives of both in the adjustment strategy. By establishing a multi-objective constrained optimization framework, the rate of range of motion recovery and the effect of pain control are used as joint optimization objectives, maximizing the rate of improvement in range of motion while ensuring that the pain score does not exceed a safe threshold. This multi-objective optimization mechanism ensures that the boundary parameters of the updated movement-pain correlation model and the load control scheme adjustment strategy can achieve a coordinated balance among various functional dimensions, avoiding an imbalance in the overall rehabilitation effect caused by optimizing a single indicator.

[0126] After the update, the new boundary parameters and adjustment strategies are recorded in the individualized evolutionary trajectory database, forming a closed-loop feedback loop. Subsequent rehabilitation training will be executed based on the updated parameters, continuously monitoring new rehabilitation effect data, and periodically repeating the above comparison and update process to achieve continuous adaptive optimization of the rehabilitation plan. This dynamic adjustment mechanism allows the rehabilitation plan to closely follow the user's actual recovery status, avoiding both delays in rehabilitation due to overly conservative approaches and the risk of injury caused by overly aggressive ones, providing personalized, safe, and efficient rehabilitation support for elderly patients with hip fractures.

[0127] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An AI-based question-and-answer system for adjusting rehabilitation plans for hip fractures in the elderly, characterized in that: include: The data analysis unit is used to acquire hip bone medical imaging data, motion monitoring data and rehabilitation history data of the target elderly user, analyze the hip bone medical imaging data to extract bone structure features, establish an individualized evolutionary trajectory in combination with the rehabilitation history data, and determine the current rehabilitation stage and functional recovery expectation based on the individualized evolutionary trajectory. The exercise safety unit is used to construct an exercise-pain correlation model. By analyzing the correspondence between load changes and user pain feedback in the exercise monitoring data, it identifies the characteristics of exercise patterns that aggravate pain, dynamically adjusts the exercise safety boundary according to the current rehabilitation stage in the individualized evolution trajectory, and generates a load control scheme based on the exercise safety boundary. The follow-up strategy unit is used to obtain the actual deviation of the exercise safety boundary based on the execution of the load control scheme, calculate the follow-up urgency score in combination with the expected progress of functional recovery, and determine the follow-up timing and intervention intensity based on the follow-up urgency score and the real-time availability of community medical resources, thereby forming a differentiated follow-up execution strategy. The model and scheme unit is used to execute the follow-up execution strategy and collect rehabilitation effect data after follow-up. The rehabilitation effect data is compared with the expected trajectory in the individualized evolution trajectory. The comparison results are used to update the boundary parameters of the exercise-pain association model and the adjustment strategy of the load control scheme.

2. The system according to claim 1, characterized in that, The data parsing unit is also used for: The hip bone medical imaging data is analyzed to extract bone structure features. An individualized evolutionary trajectory is established by combining this trajectory with the rehabilitation history data. Based on this individualized evolutionary trajectory, the current rehabilitation stage and expected functional recovery are determined, including: Bone morphology analysis was performed on hip bone medical imaging data to extract structural degeneration features reflecting changes in bone density distribution and joint space. The evolution rate of these structural degeneration features was then analyzed by time series comparison to generate a degeneration rate curve characterizing the dynamic process of bone degeneration. Extract the functional recovery range and recovery time corresponding to different rehabilitation stages from rehabilitation history data, correlate and map the functional recovery range with the degeneration rate curve, identify the nonlinear response relationship between bone degeneration rate and functional recovery ability, and construct an individualized evolutionary trajectory that reflects the sensitivity of individual bone status to rehabilitation response. The structural degradation features are located in the individualized evolutionary trajectory. Based on the location results, the rehabilitation stage identifier of the current bone status is determined. Based on the response relationship in the individualized evolutionary trajectory, the functional recovery potential and expected recovery period corresponding to the rehabilitation stage identifier are deduced.

3. The system according to claim 2, characterized in that, By mapping the functional recovery magnitude to the degradation rate curve, the nonlinear response relationship between bone degradation rate and functional recovery ability is identified, and an individualized evolutionary trajectory reflecting the sensitivity of an individual's bone status to rehabilitation response is constructed, including: The functional recovery range is segmented according to the time dimension to obtain the recovery rate change characteristics in different time periods. The recovery rate change characteristics are paired with the degradation rate values ​​in the corresponding time periods in the degradation rate curve to establish a paired dataset between degradation rate and recovery rate. Nonlinear fitting analysis is performed on the paired dataset to identify the differences in response patterns of functional recovery ability when the degradation rate is in different numerical ranges. A sensitivity function reflecting the change of response sensitivity with degradation rate is extracted, and an individualized evolutionary trajectory is constructed based on the sensitivity function.

4. The system according to claim 1, characterized in that, The motion safety unit is also used for: A movement-pain correlation model is constructed. By analyzing the correspondence between load changes and user pain feedback in the movement monitoring data, the characteristics of movement patterns that exacerbate pain are identified. The movement safety boundary is dynamically adjusted based on the current rehabilitation stage in the individualized evolutionary trajectory. A load control scheme is generated based on the movement safety boundary, including: The load change sequence is extracted from the exercise monitoring data. The load change sequence is time-aligned with the user pain feedback value at the corresponding time. The time-aligned load change sequence and pain feedback value are correlated and analyzed to identify the load change pattern that causes the pain feedback value to exceed the baseline level. The exercise intensity threshold and duration threshold in the load change pattern are extracted as pain triggering features. Based on the pain triggering features, an exercise-pain correlation model is constructed. Obtain the current rehabilitation stage identifier in the individualized evolutionary trajectory, adjust the pain triggering features in the movement-pain association model according to the functional recovery potential corresponding to the current rehabilitation stage identifier, generate a movement safety boundary that matches the current rehabilitation stage, and generate a load control scheme based on the movement safety boundary.

5. The system according to claim 1, characterized in that, The follow-up strategy unit is also used for: Based on the execution status of the load control scheme, the actual deviation from the exercise safety boundary is obtained. Combined with the expected progress of functional recovery, a follow-up urgency score is calculated. Based on the follow-up urgency score and the real-time availability of community medical resources, the timing and intensity of follow-up are determined, forming a differentiated follow-up execution strategy, including: Monitor the actual execution data of the load control scheme, quantify the deviation between the motion intensity sequence in the actual execution data and the allowable range of the motion safety boundary, identify the time distribution pattern of the deviation and the cumulative trend of the deviation magnitude, and generate the actual deviation degree curve; The expected recovery trajectory and the current actual recovery trajectory are obtained. The deviation area between the trajectories is calculated as the progress deviation. The cumulative trend characteristics of the actual deviation curve are nonlinearly mapped to the progress deviation. When the cumulative trend accelerates and the deviation exceeds the preset baseline, the urgency amplification mechanism is triggered to generate a dynamically adjusted follow-up urgency score. Based on the follow-up urgency score, a resource matching priority sequence is defined. Follow-up resources are selected from the medical personnel closest to the available time and the medical equipment with the lowest load status according to the resource matching priority sequence. At the same time, the intervention intensity level is determined according to the numerical gradient of the follow-up urgency score. A differentiated follow-up execution strategy is formed based on the resource matching priority sequence and the intervention intensity level.

6. The system according to claim 5, characterized in that, The cumulative trend characteristics of the actual deviation curve are nonlinearly mapped to the progress deviation. When the cumulative trend accelerates and the deviation exceeds a preset baseline, an urgency amplification mechanism is triggered to generate a dynamically adjusted follow-up urgency score, including: Time series analysis is performed on the actual deviation curve to extract the rate of change sequence reflecting the change of deviation magnitude over time. The second derivative of the rate of change sequence is calculated to obtain the acceleration characteristic value. When the acceleration characteristic value is positive, it is determined that the cumulative trend is in an accelerating upward state. The deviation from the achieved schedule is compared with the preset baseline to obtain the result of the deviation exceeding the limit. The cumulative trend feature and the deviation from the achieved schedule are input into a nonlinear mapping relationship. The nonlinear mapping relationship generates a mapping output value by applying an exponential transformation to the cumulative trend feature and using the deviation from the achieved schedule as a weight adjustment parameter. When the acceleration characteristic value is positive and the deviation exceeds the preset baseline, the mapping output value is multiplied by the urgency amplification factor to trigger the urgency amplification mechanism, and a dynamically adjusted follow-up urgency score is generated based on the triggering result of the urgency amplification mechanism.

7. The system according to claim 1, characterized in that, The model and scheme unit is also used for: The process of comparing the rehabilitation effect data with the expected trajectory in the individualized evolutionary trajectory, and using the comparison results to update the boundary parameters of the exercise-pain correlation model and the adjustment strategy of the load control scheme, includes: The functional recovery indicators in the rehabilitation effect data are compared point by point with the expected recovery indicators at the corresponding time points of the expected trajectory in the individualized evolution trajectory. The deviation sequence between the actual recovery indicators and the expected recovery indicators is calculated, and the deviation sequence is analyzed to identify the advanced or lagging state of the recovery process. The direction of boundary parameter adjustment is determined based on the trend analysis results of the deviation sequence. When the deviation sequence shows a state of recovery process ahead, the intensity of the pain trigger feature restriction in the movement-pain association model is reduced to relax the movement safety boundary. When the deviation sequence shows a state of recovery process lag, the intensity of the pain trigger feature restriction is increased to tighten the movement safety boundary. The boundary parameter update value is generated based on the direction of boundary parameter adjustment. The updated boundary parameter values ​​are applied to the exercise-pain correlation model to update the allowable range of exercise intensity and duration defined by the exercise safety boundary. Based on the updated allowable range of exercise intensity and duration, the exercise intensity progression rhythm and duration extension strategy in the load control scheme are regenerated, thus completing the update of the boundary parameters of the exercise-pain correlation model and the adjustment strategy of the load control scheme.

Images