Method and system for assessing compensatory regurgitation of exercise benefit in the weight loss plateau of obesity

Abstract

The application discloses an obesity weight loss plateau exercise benefit compensatory regurgitation evaluation method and system, relates to the technical field of health data processing, and provides the following scheme, which comprises the following steps: acquiring weight monitoring data collected according to time, planned exercise event records and bypass monitoring data synchronously collected in the whole process of weight loss intervention of a target object, and the bypass monitoring data comprises non-exercise activity data and metabolism-related data. The same type of exercise response of the same target object in the weight loss active period is taken as a counterfactual reference template, a change rate and a maintenance time length are applied to the pseudo-filtering of an instant favorable deviation amount in an instant window, and a return response section paired with a rapid reversible transient deviation in a delay window is flexibly deducted by a return weight reduction coefficient, so that the processing mismatch caused by the fact that the existing weight single rate indicates progress, the same type of exercise response is homogenously integrated into unified statistics in different stages and the compensatory strength cannot be identified is solved.

Description

Technical Field

[0001] This invention relates to the field of health data processing technology, and more specifically, this application relates to a method and system for assessing the compensatory rebound of exercise benefits during the weight loss plateau in obesity patients. Background Technology

[0002] Obesity is a chronic metabolic disease that requires long-term management. Weight loss interventions typically last for several months or more, during which patients' weight, daily activities, and related metabolic indicators are collected continuously over time to assess intervention progress and adjust the treatment plan accordingly. In such long-term management, patients generally enter a plateau phase where their weight no longer decreases significantly after a period of weight loss. How to perform information-based analysis on the longitudinal data collected during this phase to support the assessment is a routine requirement for long-term obesity management.

[0003] Existing methods can achieve basic assessment when data is relatively complete. The basic approach is to indicate progress by the rate of weight change or the duration of no further weight loss, and to import the concurrent activity, heart rate, blood glucose and other indicators according to their respective statistics. This approach assumes that weight change is sufficient to characterize progress, considers the stage where weight loss stops as a state where no more information is available, and assumes that the body's response to exercise is roughly the same throughout the process. Therefore, it homogeneously imports the responses after the same type of exercise in different stages into a unified statistic, without retaining the patient's response to the same type of exercise during the stage of continuous weight loss for internal comparison in subsequent stages.

[0004] However, as the intervention continues, the body gradually compensates for the energy deficit caused by the same amount of exercise through both behavioral and metabolic means. This compensation increases with the weight loss process. Therefore, under the seemingly identical phenomenon of "no weight loss" at the plateau stage, there are actually two opposite situations: those where compensation is still weak and benefits are still accumulating and can continue to be promoted, and those where compensation is strong and the benefits of exercise are being significantly offset, and further exercise will also be offset. Existing methods, based on whether weight has decreased and homogeneous statistics of each response, cannot distinguish between these two situations. As a result, those with strong compensation are misinterpreted as ordinary stagnation, leading to an overly optimistic assessment for patients who most need compensation and missing the opportunity to adjust the direction of intervention in a timely manner. It is evident that relying on weight changes and homogeneous statistics of each response is insufficient to reflect the degree to which the benefits of exercise are offset by the body's compensation at the plateau stage. Therefore, a method and system for assessing the compensatory return of exercise benefits during the weight loss plateau stage in obesity is proposed to solve this problem. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method and system for assessing the compensatory retracement of exercise benefits during the weight loss plateau phase in obese patients. This technical solution resolves the issues raised in the background section.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows:

[0007] In a first aspect, this application provides a method for assessing compensatory retracement of exercise benefits during the weight loss plateau phase in obese patients, the method comprising:

[0008] Acquire weight monitoring data, planned exercise event records, and simultaneously collected bypass monitoring data collected throughout the weight loss intervention process for the target subjects. The bypass monitoring data includes non-exercise activity data and metabolic-related data.

[0009] The platform window, which is identified by weight monitoring data and whose net change approaches zero and lasts for a preset duration, is used to determine the platform baseline during the period within which it is not affected by planned exercise events. The planned exercise events are used to locate the disturbance anchor point, and an instant window and a delayed window are constructed with the anchor point as the phase zero point. The disturbance intensity and isolation degree are respectively marked by the movement quantity of the disturbance anchor point and the superposition degree of other disturbances in adjacent periods.

[0010] During the active weight loss period when the target object is in the platform window and its weight is continuously decreasing, similar planned exercise events with perturbation intensity matching each perturbation anchor point are retrieved. Their non-exercise activity volume and metabolism-related response trajectory are extracted according to the same instant window and delayed window as the counterfactual reference template of the perturbation anchor point. The matching quality is obtained by the matching degree of the number and intensity of the retrieved similar planned exercise events.

[0011] Within the immediate window, the deviation of metabolic data from the platform baseline toward the direction of weight loss benefits is extracted to obtain the immediate beneficial deviation. Within the delayed window, the deviation components of non-exercise activity and metabolic responses along their respective corresponding weight loss benefit directions are taken. The corresponding deviation components measured in the same window are subtracted from the counterfactual reference template to obtain the behavioral and metabolic feedback residuals, which are then added together to synthesize the net feedback.

[0012] The net return amount is normalized by the immediate favorable deviation to obtain the return ratio, which represents the degree of compensation return of exercise gains. The return ratios of each disturbance anchor point within the platform window are continuously weighted and aggregated according to isolation degree, matching quality and directional consistency of the two return residuals to obtain the plateau period compensation return parameter, which is used as the decision support parameter output for weight loss intervention assessment.

[0013] Secondly, this application provides a system for assessing the compensatory return of exercise benefits during the weight loss plateau phase in obesity patients, used to implement the aforementioned method for assessing the compensatory return of exercise benefits during the weight loss plateau phase in obesity patients, including:

[0014] The data acquisition module is used to acquire weight monitoring data, planned exercise event records, and simultaneously acquired bypass monitoring data collected over time throughout the weight loss intervention of the target subject. The bypass monitoring data includes non-exercise activity data and metabolic-related data.

[0015] The platform anchoring module is used to identify a platform window where the net change approaches zero and continues for a preset duration based on weight monitoring data. The platform baseline is determined within the time period that is not affected by planned exercise events. The planned exercise events are used to locate disturbance anchor points, and an instant window and a delayed window are constructed with these anchor points as the phase zero points. The disturbance intensity and isolation degree are respectively calibrated based on the movement quantity of the disturbance anchor point and the superposition degree of other disturbances in adjacent time periods.

[0016] The reference construction module is used to retrieve similar planned exercise events with perturbation intensity matching each perturbation anchor point during the active weight loss period when the target object is in front of the platform window and its weight is continuously decreasing. The non-exercise activity volume and metabolism-related response trajectory are extracted according to the same instant window and delayed window as the counterfactual reference template of the perturbation anchor point. The matching quality is obtained by the matching degree of the number and intensity of the retrieved similar planned exercise events.

[0017] The residual synthesis module is used to extract the immediate beneficial deviation from the deviation of metabolic data relative to the platform baseline in the direction of weight loss benefit within the immediate window, and to extract the deviation components of non-exercise activity and metabolic response along their respective weight loss benefit directions within the delayed window. The corresponding deviation components measured in the same window are subtracted from the counterfactual reference template to obtain the behavioral and metabolic feedback residuals and then added together to synthesize the net feedback amount.

[0018] The parameter output module is used to normalize the net return amount with the immediate favorable deviation to obtain the return ratio, which represents the degree of compensation return of exercise gains. The return ratio of each disturbance anchor point within the platform window is continuously weighted and aggregated according to isolation degree, matching quality and directional consistency of the two return residuals to obtain the plateau period compensation return parameter, which is used as the decision support parameter output for weight loss intervention assessment.

[0019] Thirdly, this application provides a computer device, the computer device including a memory and a processor, the memory storing code, the processor being configured to acquire the code and execute the above-described method for evaluating the compensatory return of exercise benefits during the weight loss plateau in obesity.

[0020] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for evaluating the compensatory return of exercise benefits during the weight loss plateau in obesity.

[0021] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0022] This application uses the same type of exercise response during the active weight loss period of the same target subject as a counterfactual reference template, applies a false filtering of the rate of change and duration of maintenance to the immediate beneficial deviation within the immediate window, and flexibly subtracts the return response segment paired with the rapid reversible transient deviation within the delayed window with a return weighting coefficient. This solves the processing mismatch of existing methods that use a single rate of weight to indicate progress and homogeneously merge the same type of exercise response across stages into a unified statistic, making it impossible to identify the strength of compensation. This allows both the exercise benefit and the component that is returned to be reconstructed as an intermediate quantity that has been corrected by echoing the previous and subsequent steps, thereby separating the true component that can be attributed to compensation under the same dose and morphology control within the individual.

[0023] This application uses the acceptance signature obtained by jointly extracting motion response data and metabolic related data within an instant window as a unified standard, and reuses it in three areas: authenticity calibration of perturbation intensity, dose normalization of behavioral feedback residuals, and morphological verification of matching quality. This solves the implicit biases of existing methods that use registered motion measurements as the sole characterization of motion dose, mislabel perfunctory execution as valid perturbations, and undifferentiated equal weighting of reference hit events in terms of acceptance morphology. This allows the previously un-explicitly modeled step of whether a perturbation is physiologically accepted to share the same signature dimension in the verification, quantification, and matching stages, avoiding the standard drift caused by taking temporary features in each step.

[0024] This application solves the problems of existing methods that only observe compensation in a single path, ignore the energy gap that can be independently offset by both behavioral and metabolic pathways, and rely on hard threshold selection, by symmetrically synthesizing the compensatory resting amount after behavioral-side exercise and the regurgitation residual of metabolic-side exercise as additive independent paths to form the net regurgitation amount. It also solves the problems of signal loss or contamination caused by relying on hard threshold selection, by using the consistency coefficient of the two paths to continuously de-weight the perturbation anchor points with only single-path deviation, and by using isolation degree and matching quality to synergistically weight and aggregate them as plateau period compensation regurgitation parameters. This allows the output to be reconstructed from a single numerical feedback into intermediate decision support parameters that can trace the evidence strength of each anchor point, so that the downstream scheme adjustment process can identify the main compensation path and select the corresponding intervention direction. Attached Figure Description

[0025] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. Wherein:

[0026] Figure 1 This is a flowchart of the method for evaluating the compensatory rebound of exercise benefits during the weight loss plateau period in obesity patients, as proposed in this invention.

[0027] Figure 2 This is a structural block diagram of the obesity weight loss plateau phase exercise benefit compensatory rebound assessment system proposed in this invention. Detailed Implementation

[0028] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0029] Reference Figure 1 As shown, this application proposes an overall processing flow for assessing the compensatory rebound of exercise benefits during the weight loss plateau in obesity. This application is applicable to research / follow-up scenarios where weight loss interventions have been implemented on target subjects for several months or more, and multimodal health data has been collected longitudinally over time. Typical scenarios include weight loss follow-up cohorts based on electronic health records, long-term management projects with wearable devices, and follow-up data archiving at hospital weight management clinics. This method only performs information analysis and decision support on the collected longitudinal data, without involving in vivo motion data collection and sensing device implementation. The final output is intermediate decision support parameters for weight loss intervention assessment, which researchers or management teams can use to adjust the next stage of the intervention plan, such as adjusting the intensity ratio of exercise prescriptions, introducing anti-compensatory sedentary interventions, or reassessing the energy deficit setting.

[0030] For ease of understanding, the following text uses a consistent research scenario: In a weight loss follow-up cohort in a certain region, an obese adult patient who had been undergoing intervention for approximately 14 weeks was included as the target subject. Their weight monitoring data was collected daily (accurate to 0.1 kg) from cloud records of a home electronic scale; non-exercise activity data was statistically analyzed hourly (in steps / h and metabolic equivalent minutes MET·min, which can be understood as the cumulative activity within that hour) from a wrist-worn activity recording device; metabolic-related data included heart rate tracking (HR, bpm) at the second level and continuous glucose monitoring data (CGM, in mmol / L, with a 5-minute sampling interval); planned exercise events were recorded from the start and end times and exercise intensity (intensity level × duration converted to MET·min) registered by the target subject in the accompanying mini-program according to the procedures. The target subject entered a stage where weight loss no longer became significant from week 10 onwards.

[0031] Acquire weight monitoring data, planned exercise event records, and simultaneously collected bypass monitoring data collected throughout the weight loss intervention process for the target subjects. The bypass monitoring data includes non-exercise activity data and metabolic-related data.

[0032] It should be noted that all acquired data comes from the target subject's complete electronic record from the start of the intervention to the current assessment time, uniformly aligned to the same time benchmark and indexed by minute-level time according to a unified time zone; weight monitoring data is merged daily, taking valid records from the same time period (e.g., fasting in the morning), missing days are filled with the most recent day and marked with an interpolation flag; non-exercise activity data is aggregated hourly into two indicators: valid activity MET·min and low activity (i.e., cumulative duration of steps less than 100 steps / h); heart rate trajectory in metabolic-related data is averaged minute by minute for platform baseline and delay window statistics, and downsampled by 10-second average to retain the trajectory during exercise for signature extraction; continuous glucose monitoring data is retained in its original sample and accompanied by a reliability flag; the planned exercise event record fields include at least the event identifier and start time. Duration With exercise measurement (The exercise measure is obtained by multiplying the MET value corresponding to the exercise type by the duration, and the unit is MET·min). When importing event records from different sources, duplicates are removed based on the overlap between the timestamp and the duration. If the overlap is higher than 0.8, they are considered to be the same event, and the record with the higher measure is retained.

[0033] It should be noted that a major problem in the data acquisition phase is the heterogeneous sampling rate, time zone jumps, and short-term data loss caused by multi-source acquisition. To address this, this embodiment uses a minute-level time index as the benchmark for aligning and downsampling / interpolating the data from each source. A confidence bypass bit is set independently for each source: if there is no valid weight record for the day, it is set to 0 and interpolation is allowed for up to 3 consecutive days; windows with a loss of more than 2 hours in non-exercise activity are marked as unavailable; samples exceeding the physiologically reasonable range (e.g., HR < 30 bpm or > 220 bpm, CGM < 2.2 mmol / L or > 22.0 mmol / L) in heart rate and continuous glucose monitoring are removed and filled back using linear interpolation within a window of no more than 15 minutes. Data exceeding this window is truncated for the current analysis within that window.

[0034] The platform window, which is identified by weight monitoring data and whose net change approaches zero and lasts for a preset duration, is used to determine the platform baseline during the period within which it is not affected by planned exercise events. The planned exercise events are used to locate the disturbance anchor point, and an instant window and a delayed window are constructed with the anchor point as the phase zero point. The disturbance intensity and isolation degree are respectively marked by the movement quantity of the disturbance anchor point and the superposition degree of other disturbances in adjacent time periods.

[0035] It should be noted that the weight monitoring data is used. Piecewise descent rate obtained from fitting Attenuation to a preset near-zero range (typically) (week) and maintain at least the preset duration (Typically 28 days) is used as a candidate condition, and the platform window is obtained by determining the compensation starting boundary of the candidate segment. ;

[0036] Platform baseline Robust estimation of metabolic-related indicators is performed by taking quantiles from the "quiet period" within the platform window, where the interval between the start and end times of any planned exercise event is greater than the duration of a delay window. (Typically, the median is taken from the hourly quantiles; if necessary, binning is performed by time of day to eliminate systematic biases in the diurnal rhythm). Its field composition includes at least the following: These serve as baseline reference levels for heart rate, continuous glucose, and non-exercise activity, respectively.

[0037] Disturbance anchor point The event is determined directly by the planned motion event falling within the platform window, where the index... K is the index of the perturbation anchor points within the platform window, and K is the total number of perturbation anchor points within the platform window (in this application, all physical quantities marked with the subscript k correspond one-to-one with the kth perturbation anchor point); subsequently, physical quantities marked with the subscript j uniformly refer to similar planned exercise events during the weight loss active period hit in the counterfactual reference template retrieval stage for the kth perturbation anchor point. The phase zero point is taken as the event's... ;

[0038] Instant window calculated from the zero point of this phase Used to characterize the immediate physiological and metabolic responses produced during exercise (typical) (At least covering the period during exercise and the early recovery period after exercise); Delay window Used to characterize the timing of compensatory behavior and metabolic reversion (typical) For events with high levels of activity, the timeframe can be extended to 48 hours; disturbance intensity. Measurement of the motion of the event Calibrate together with heart rate response data:

[0039] For example, the expression for calculating the disturbance intensity is:

[0040] ;

[0041] in, The peak heart rate at the instantaneous window (i.e., at the time when the heart rate is at the peak heart rate) The phase zero point, with a length of Instant window access The maximum instantaneous value is obtained by the following formula:

[0042] ;

[0043] The median of exercise measurements recorded during the target individual's active weight loss period (for intra-individual normalization). The predicted upper limit of heart rate reserve is estimated based on age for the target individual (used as a reference for normalizing the individual's heart rate response). To assign weights;

[0044] Since exercise measurement reflects the planned "expected" disturbance dose, while heart rate response reflects the actual "expected" disturbance intensity, both must be considered simultaneously to avoid perfunctory exercise being mislabeled as high-intensity disturbance. Therefore, this embodiment uses... This gives heart rate response a slightly larger weight in the calibration, typically The value falls within the interval [0, 2.0].

[0045] isolation Characterizing whether there is superimposed contamination from other motion events in adjacent time periods at the disturbance anchor point, defined as... Expand one forward and one backward from the center. Within the neighborhood of the event, the negative normalized value of the sum of motion metrics of all planned motion events other than the current event, taking values ​​in [0, 1]. The closer to 1, the more isolated the disturbance is, and the more likely it is to be assigned a higher confidence weight when the plateau period compensation backfeed parameter is aggregated as an independent disturbance unit.

[0046] During the active weight loss period when the target object is in the platform window and its weight is continuously decreasing, similar planned exercise events with perturbation intensity matching each perturbation anchor point are retrieved. Their non-exercise activity volume and metabolism-related response trajectory are extracted according to the same instant window and delayed window as the counterfactual reference template of the perturbation anchor point. The matching quality is obtained by the matching degree of the number and intensity of the retrieved similar planned exercise events.

[0047] It should be noted that the active weight loss period is defined as the time segment before the plateau window, during which weight monitoring data shows a continuous downward trend (typically, a weekly weight loss rate of at least 0.3 kg / week for at least 6 consecutive weeks), denoted as . This period is considered as the self-reference baseline of the same target object before compensation is fully established, thus constituting an intra-individual counterfactual, that is, answering the question of "what kind of "response trajectory" would be obtained if the same intensity of this disturbance occurred during its own decline phase without introducing data from other target objects; the practical basis for this treatment is that compensatory retardation gradually increases with the weight loss process, so the non-motor activity and metabolic response of the same target object to the same amount of exercise during the decline phase can be regarded as a "weak compensation" reference, and the difference between the response and the plateau phase corresponds to the retardation component caused by compensation, thus avoiding the individual heterogeneity error introduced by cross-population reference.

[0048] For each plateau period disturbance anchor point The search criteria are: disturbance intensity is within Interval (typical) ), type of exercise and Consistent (e.g., both aerobic or resistance training), isolation degree not less than 80%; hit events are counted by the number of hits. For each hit event, the non-motor activity and metabolic response trajectory are extracted using the same immediate and delayed windows. Robust aggregation (typically using the median trajectory with interquartile range) is performed, indexed by minutes, to obtain a counterfactual reference template. Its field composition includes at least ;

[0049] in, The counterfactual reference trajectory is obtained by robustly aggregating the non-motor activity response trajectories of all hit events retrieved for the k-th perturbation anchor point within the same instantaneous and delayed windows. The counterfactual reference trajectory is obtained by robustly aggregating the heart rate response trajectories within the same window for the same retrieval set. The counterfactual reference trajectory is obtained by robustly aggregating the response trajectories of continuous glucose monitoring data within the same retrieval set in the same window. All three reference trajectories are indexed by minutes and aligned with the measured trajectory during the plateau period at the same phase zero point. They are used for subsequent calculation of the backflow residuals of behavioral and metabolic pathways (including the two metabolic side pathways of heart rate and continuous glucose).

[0050] like (Insufficient hits lead to insufficient template robustness), so a downgrade is applied: If the isolation threshold is relaxed to 0.25 and lowered to 0.6, and there are still fewer than 3 references, the anchor points of this disturbance will be marked as "insufficient references" and will participate in subsequent aggregations with a matching quality approaching zero.

[0051] Matching quality It is determined by both the number of hits and the degree of intensity matching:

[0052] ;

[0053] in, This is a typical saturation hit number (typically 5). For the first The intensity of the perturbation in the hit event; the higher the matching quality, the higher the statistical reliability of the counterfactual reference, and the more reliable the attribution of the backpropagation of this perturbation.

[0054] Within the immediate window, the deviation of metabolic data from the platform baseline toward the direction of weight loss benefits is extracted to obtain the immediate beneficial deviation. Within the delayed window, the deviation components of non-exercise activity and metabolic responses along their respective corresponding weight loss benefit directions are taken. The corresponding deviation components measured in the same window are subtracted from the counterfactual reference template to obtain the behavioral and metabolic feedback residuals, which are then added together to synthesize the net feedback.

[0055] It should be noted that "in the direction of weight loss benefits" has different physical meanings for different indicators: for non-exercise activity, the direction of weight loss benefits is an increasing direction (more non-exercise activity is beneficial for maintaining the energy deficit); for metabolic-related data such as heart rate and continuous glucose, the direction of weight loss benefits is a baseline stabilization accompanied by a phased decline to a lower level (reflecting an orderly response of resting energy expenditure and glucose metabolism burden), and cannot be simply identified by a unilateral rise or fall; here, we will first point out according to the definition: in Within, metabolic-related data will be compared to the platform baseline. The cumulative deviation area in the direction of weight loss benefits is used as The initial foundation.

[0056] The two-way backfeedback residuals use a counterfactual reference template as the "expected response" and the measured response during the plateau period as the "actual response." The difference is the backfeed component caused by compensation. Specifically, the behavioral backfeedback residuals... Pick Relative amount of internal non-motor activity Insufficient amount (the amount of behavioral activity that should have been achieved but was not), metabolic regurgitation residual. Take metabolic-related data from classmates. or The corresponding deviation component that was not realized (the amount of favorable deviation that should have been present on the metabolic side that was not maintained) is combined with the net regurgitation amount using the following formula: ;

[0057] Since behavioral pathways (such as decreased non-motor activity and prolonged low activity duration after exercise) and metabolic pathways (such as regression of metabolic-related baseline and narrowing of response amplitude) physiologically correspond to two independent subsystems, namely voluntary activity regulation and neuro-endocrine response, and both pathways can independently offset the energy deficit caused by the exercise, this embodiment combines the two in parallel rather than selectively treating them to ensure that the regurgitation that occurs in either pathway is taken into account.

[0058] Additionally, in one implementation, metabolic data includes continuous glucose monitoring data. Accordingly, the extraction of metabolic reflux residuals specifically includes: extracting the glucose baseline relative to the plateau baseline within a delay window. The uncompleted closure residual, as the metabolic component of the regurgitation, forms a symmetrical dual outlet with the post-motor compensatory resting quantity at the behavioral outlet.

[0059] Specifically, this can be implemented as follows: within the delay window, take the moving median of continuous glucose monitoring data on a minute-by-minute basis to obtain a smooth trajectory, and use the glucose decline curve corresponding to the counterfactual reference template as the "expected decline trajectory". The difference between the measured glucose trajectory during the plateau phase and the plateau baseline failed to close. The cumulative area at the appropriate level should be used as the glucose baseline closure residual. The normalized intensity of the response to this disturbance is then included in the calculation. The basis of this lower-level embodiment is that the orderly decline of the glucose baseline after exercise is a sign that the glucose metabolism response is normal in accepting exercise perturbations. However, once the body's compensation is established, the decline amplitude of the glucose baseline relative to its own decline phase after a perturbation of the same intensity will show "failure to close". This closure residual is a fine-grained characterization of metabolic backflow. This lower-level embodiment is only an optional implementation of the metabolic exit. The independent method is not limited to this. Equivalent closure characterization of heart rate response or other metabolic-related data combinations can also be used. Choosing one or using both is within the scope of this embodiment.

[0060] The net return amount is normalized by the immediate favorable deviation to obtain the return ratio, which represents the degree of compensation return of exercise gains. The return ratios of each disturbance anchor point within the platform window are continuously weighted and aggregated according to isolation degree, matching quality and directional consistency of the two return residuals to obtain the plateau period compensation return parameter, which is used as the decision support parameter output for weight loss intervention assessment.

[0061] retracement ratio The normalized form is:

[0062] ;

[0063] in, Minimal positive value (typical) This is the arithmetic mean of the immediate beneficial deviations of each disturbance anchor point within the target platform window, in order to avoid... (When approaching zero, the return ratio diverges). Since the return ratio is the internal ratio of "how much exercise benefit is returned" to "how much exercise benefit should have been", the dimensions naturally cancel each other out. Therefore, this embodiment selects this normalized form so that perturbations of different intensities can be compared horizontally within individuals. Similarly, a return ratio of 0.4 indicates the same level of compensation pressure regardless of whether the perturbation is low-intensity aerobic or medium-intensity resistance, avoiding confusion between "compensation intensity" and "exercise dosage" using absolute return amount. The typical range of values ​​is [0, 1.2]. Values ​​exceeding 1.0 indicate that the gains from the disturbance are completely reversed or even result in a net loss.

[0064] The return ratios of each disturbance anchor point are aggregated by weight to form the plateau period compensation return parameters. :

[0065] ;

[0066] in, The directional consistency coefficient of the two-way backfeed residuals is defined as follows: and The combined value of the degree to which the signs of the two are in the same direction and the degree of proportional balance when they are in the same direction falls within [0,1]: when both are in the direction of compensation (i.e., both are positive insufficient quantities) and their magnitudes are close. Approaching 1, when one path is positive and the other is negative, or one path is much larger than the other. Significantly decreased, approaching 0.

[0067] For example, the expression for calculating the directional consistency coefficient is as follows:

[0068] ;

[0069] in, The coefficient is set to 1 when the variable is positive and 0 otherwise, and is directly set to zero when either of the two is not positive. and The smaller and larger values ​​of the two residuals are taken respectively. The ratio of these two values ​​characterizes the degree of balance in the compensation of the two pathways in the same direction. The closer the two values ​​are, the closer the ratio is to 1; the greater the difference between them, the closer the ratio is to 0. Since if the behavioral pathway and the metabolic pathway can deviate from the compensation direction simultaneously and by a similar magnitude, the attribution of the compensatory backflow of the perturbation is more consistent, therefore... The credibility of the anchor point is continuously weighted so that individual suspected bias anchor points that only show backflow in a single path are naturally downweighted, without having to be removed by a hard threshold.

[0070] final The field composition includes at least: the main value of the platform period compensation backoff parameter (scalar, value range [0, 1.2]), the number of perturbation anchors participating in the aggregation, the average matching quality, and the proportion of behavior path backoff. The percentage of metabolic pathway relapse, and the start and end times of the platform window used for this assessment; this field serves as an intermediate decision support parameter for weight loss intervention assessment, and is adjusted by downstream program adjustments according to a preset threshold (typically...). , , The option may be to maintain the original plan, adjust the exercise dosage ratio, or introduce anti-compensation interventions (such as reducing the intensity of a single exercise session and increasing the frequency, or increasing non-exercise activity prompts, etc.), but this parameter itself does not constitute a conclusive judgment on the health status of the target subject.

[0071] Through the above technical solution, this embodiment uses a four-step process for processing the longitudinal data of the target object: "intra-individual counterfactual reference - symmetrical synthesis of two-way backfeeding residuals - backfeeding ratio normalization - continuous weighted aggregation". This differs from the existing method that uses the rate of weight change or the duration of no further weight loss to indicate progress and homogeneously aggregates similar exercise responses from each stage into a unified statistic. The key difference is that this solution uses similar exercise responses from the same target object during the weight loss phase as a counterfactual reference and symmetrically characterizes the compensatory backfeeding in both behavioral and metabolic pathways as summable residuals. This allows for the differentiation of two opposite situations—weak compensation and strong compensation—under the same appearance of "no weight loss" in the plateau phase.

[0072] The segmented rate of decline is calculated from the weight monitoring data, and the segmented rate of decline is attenuated to a preset near-zero range. The period in which the density of planned exercise events and the amount of exercise do not decrease synchronously is determined as the candidate plateau period.

[0073] It should be noted that the segmented descent rate The calculation method is as follows: weight monitoring data A weighted linear regression is performed using a preset local time window (typically a forward 14-day window), and the slope is the slope of that day. Unit: kg / week; near-zero range preset as The weekly (typical value) covers normal fluid fluctuations while excluding true continuous weight loss. The density of planned exercise events during the same period is counted by "weekly event count" and the exercise volume is counted by "weekly cumulative MET·min". Both are compared with the average of the previous 4 weeks of sliding window and the weight loss active period before the plateau window. If the density of events is not lower than 80% of the average of the active period and the exercise volume is not lower than 80% of the average of the active period, it is considered that "the density of planned exercise events and the exercise volume during the same period have not decreased synchronously", thus excluding "weight plateau caused by training stoppage rather than compensation". Since another legitimate explanation for no weight loss is that the target subject actively reduced the amount of exercise, if this situation is not excluded, the training stoppage plateau will be misjudged as a compensation plateau, leading to incorrect attribution of subsequent rebound parameters. Therefore, verifying both near-zero weight and no weight loss on the exercise side is a double necessary condition.

[0074] Within the reference segment where weight continued to decline during and before the candidate platform period, the ratio of the magnitude of the weight recovery segment to the adjacent weight decline segment was extracted according to a preset local time window to obtain the recovery and decline magnitude ratio.

[0075] It should be noted that the preset local time window is typically 7 days. A weight gain segment is defined as a continuous sub-segment within this window where weight monitoring data shows a non-negative change relative to the previous day, with the amplitude taken as the difference between the start and end weights of this sub-segment. An adjacent weight loss segment is defined as a continuous sub-segment of negative change immediately preceding the gain segment, with the amplitude taken as the absolute value. The ratio of the gain / loss amplitude... This is the ratio of the cumulative magnitude of the upward phase to the cumulative magnitude of the downward phase within the same 7-day window, and its value is within... Typical during the active period The value fluctuates slightly between 0.1 and 0.3 after entering the platform. It will rise significantly, approaching or exceeding 1.0.

[0076] The starting point where the rise and fall within the candidate platform period is continuously higher than the relative reference segment is determined as the compensation starting boundary, and the portion from the compensation starting boundary to the end of the candidate platform period is determined as the platform window.

[0077] Specifically, the candidate platform time period Sequence and reference segment (i.e., the active period window of at least 4 weeks prior to the candidate platform period) The difference between the means is obtained. ;when If the value remains positive for at least 7 consecutive days and its cumulative increase is not less than a preset threshold (typically 1.5 times the standard deviation of the reference segment), then its starting time is... This is determined as the starting boundary for compensation, from which the timeframe ends at the candidate platform. The range is the platform window. The key points for this boundary determination are: the characteristics of a compensatory plateau are not only that it no longer declines, but also that "the proportion of small rebounds relative to the decline rate continues to rise." This increase in the ratio occurs earlier than the visually perceptible inflection point of the weight curve. Therefore, using it as the starting point of the plateau window can more accurately delineate the time period dominated by compensation, avoiding including the mixed period of the transition phase in the aggregation. If the above conditions are not met, or if the platform appears in the middle of the candidate platform period, the former will mark the current candidate period as "unconfirmed platform" and exit the subsequent processing, while the latter will be truncated according to the actual take-off time and will not be pushed forward.

[0078] Through the above technical solution, this embodiment, by performing a dual verification of the near-zero rate of weight monitoring data segmentation and the asynchronous decrease on the exercise side, and by locating the compensation starting boundary based on the ratio of the recovery and decrease amplitudes, differs from the existing method of identifying the platform solely based on the duration of no weight loss. The key difference lies in using "compensation dominance" as the intrinsic judgment mark of the platform window, rather than using the near-zero rate as the only external appearance. The resulting platform window boundary provides a time range for locking in the subsequent delineation of disturbance anchor points and the determination of "whether it falls into the compensation dominance period" for counterfactual reference.

[0079] The bypass monitoring data also includes exercise response data collected synchronously with the planned exercise event, and the exercise response data includes at least one of heart rate response data and activity intensity data during the planned exercise.

[0080] It should be noted that the heart rate response data is taken from the continuous heart rate trajectory during exercise on the wrist-worn device (sampled per second, then averaged and downsampled over 10 seconds), while the activity intensity data is taken from the MET estimation of the accelerometer on the same device or from the cadence / stroke rate / instantaneous power during exercise (averaged per minute). One or both can be selected: the combination of heart rate response as the primary factor and activity intensity as the secondary factor can simultaneously characterize the "physiological load borne" and the "exercise dose at the output end." The two are complementary in assessing "perfunctory execution." For example, low exercise intensity but high heart rate response may indicate low cardiovascular load, while high exercise intensity but insufficient heart rate response may indicate poor device wearing or abnormal measurement.

[0081] Within the instantaneous window of the disturbance anchor point, extract the magnitude of the jump in motion response data relative to the level before the disturbance anchor point, as well as the start delay of the jump relative to the moment of the disturbance anchor point;

[0082] It should be noted that the magnitude of the jump Defined as the difference between the peak value of the motion response data within the instantaneous window (typically the maximum 1-minute average within that window) and the average value within a preset silent window (typically the 30 minutes before exercise) prior to the disturbance anchor point; the unit is consistent with the motion response data, with heart rate in bpm and activity intensity in MET; start delay. Defined as the difference between the moment when the motion response data first crosses a preset jump threshold (typically twice the mean of the silent window plus its standard deviation) from the moment of the disturbance anchor point, and the moment of the disturbance anchor point itself, in minutes. If the time falls within the 0-5 minute range, and exceeds the preset upper limit (typically 10 minutes), it is judged as "abnormal connection".

[0083] Within this instant window, the initial slope of the metabolic data relative to the platform baseline is extracted, and the jump magnitude, initial delay and initial slope of the response are combined to form the acceptance signature of the disturbance anchor point;

[0084] It should be noted that the initial slope of the response Defined as metabolic-related data within 30 minutes prior to the perturbation anchor point relative to the plateau baseline. The least squares slope of the deviation sequence, in bpm / min for heart rate and mmol / (L·min) for continuous glucose; the signature is combined into ordered tuples in three components. Before subsequent use, each component is Z-standardized within the individual according to the same component distribution during the active weight loss period of the target object, so that the three components have consistent dimensions and can be treated in parallel in the fit comparison.

[0085] Additionally, it should be noted that the reason for selecting the three components of "jump amplitude, initial delay, and response initiation slope" for the signature, rather than any other combination of features, is as follows: In this case, a planned movement event is used as an energy disturbance applied to the body. Whether the movement gain is actually generated depends on whether the disturbance is physiologically received. The receiving process can be physically decomposed into three independent observable aspects: First, the jump amplitude characterizes whether the movement response caused by the disturbance reaches an intensity sufficient to constitute an effective physiological load, used to distinguish perfunctory execution (although the measurement record exists, the movement response does not reach the jump judgment threshold, for example, registering 30). The three components are: first, the initial delay, which characterizes whether the body's response to the disturbance starts within a physiologically reasonable timeframe, and is used to distinguish between true acceptance (rapid autonomic nervous system response, typically with a heart rate jump observable within 0-5 minutes) and accidental fluctuations (such as short-term non-exercise heart rate increases caused by accidental device touch or emotional momentary jumps, which often cannot form a stable jump sequence); second, the response initiation slope, which characterizes the immediate response speed of the metabolic side to the disturbance, and describes the steepness of the "acceptance slope" from the start of the disturbance to the entry of metabolic-related data into the response trajectory, and is used to distinguish the efficiency of the metabolic system's response to the disturbance; these three components correspond to the "dose authenticity, temporal authenticity, and response authenticity" of the same process, and are orthogonal rather than redundant, thus constituting the minimum sufficient characterization of the process of "whether the disturbance is accepted".

[0086] The signature was reused in three places in this case: for verifying the authenticity of the jump amplitude as a measure of motion and for the intensity of the disturbance. The calibration; the actual strength of the response to this disturbance, used as the dose for behavioral backfeed residuals; this design of "verification, quantification, and matching" using the same signature caliber ensures that the characterization of motion disturbance reception throughout the process has inherent consistency, rather than using unrelated temporary features in different places.

[0087] Through the above technical solution, this embodiment extracts and standardizes the three components of motion response data and metabolic related data within the instant window, which is different from the existing method that only uses the registered motion measurement as the sole characterization of motion dosage. The key difference is that the link of "whether the perturbation is physiologically received", which was not explicitly modeled, is placed on the observable three components, and the signature is used in the subsequent perturbation intensity calibration, behavior backflow residual normalization, and counterfactual reference matching verification. The resulting acceptance signature is used as the common input for the perturbation intensity correction coefficient, backflow residual dose normalization benchmark, and matching quality verification dimension.

[0088] Among the deviations of metabolic-related data from the platform baseline toward weight loss benefits, those that remain for more than a preset period of time and show a narrowing of the baseline fluctuation range or a phased decline in the metabolic-related data during the same period are determined as valid deviations.

[0089] It should be noted that the weight loss benefits are specifically oriented according to different metabolic-related data as follows: in terms of heart rate, it is the phased decline of the resting heart rate after exercise from the plateau baseline; in terms of continuous glucose, it is the narrowing of the fluctuation range of the glucose baseline or the phased decline of the baseline. Both reflect that the metabolic burden is shifting in a more orderly direction and closer to the healthy trajectory, rather than a simple one-sided rise or fall.

[0090] For example, the preset duration for "maintaining for a preset duration or longer" Typically, 15 minutes is used, meaning deviations must occur within the immediate window. Internal continuous maintenance at least To avoid being triggered by transient noise; "Bottomline fluctuation amplitude narrowing" is determined by the ratio of the moving standard deviation of metabolic-related data within the same window to the standard deviation of the silent window before the perturbation anchor point being lower than a preset proportion (typically 0.8); "Phase shift decline" is determined by the difference between the median within the same window and the plateau baseline being significantly greater than zero in the direction of weight loss gains (typically not less than 0.5 times the standard deviation of the silent window); meeting any of the above accompanying conditions and lasting for a long period of time The deviation segment is determined as the valid deviation segment, and its cumulative magnitude is obtained by integrating over time.

[0091] The portion of the deviation whose rate of change exceeds the preset sustainable rate of change range and returns briefly within the instantaneous window is marked as a fast reversible transient deviation.

[0092] It should be noted that the upper limit of the rate of change Preset parameters for different metabolic data: heart rate is set at 5 bpm / min (steeper changes caused by non-exercise factors are often related to emotions / posture rather than the true metabolic response), and continuous glucose is set at 0.1 mmol / (L·min); "short return" means that the time between the start of the deviation segment and the return to the pre-deviation level does not exceed the preset return time. (Typical 30 minutes) and the deviation segment does not trigger repeatedly as the criterion; satisfying "the rate of change exceeds "and" in The deviation segment under the two conditions of "internal return" is marked as a fast reversible transient deviation, and its deviation amplitude is obtained by time integration. The time period of occurrence is recorded as .

[0093] Record the deviation magnitude and occurrence time of rapid reversible transient deviations, summarize the magnitudes of effective deviations, and obtain the immediate beneficial deviation amount;

[0094] Immediate favorable deviation We obtain the following formula:

[0095] ;

[0096] in, This is the set of all valid deviation segments within the instantaneous window corresponding to the current disturbance anchor point (excluding fast reversible transient deviations). This is a trajectory of metabolic-related data (taken from heart rate or continuous glucose monitoring data). To and Platform baseline for corresponding branch ( or ); Fast reversible transient deviation ( Archived separately, not included This serves as a location clue for the delayed window return response segment.

[0097] Additionally, it should be noted that "deviation in the direction of weight loss gains" is not equivalent to a simple, short-term decrease in weight. The definition of true metabolic gain needs further clarification using metabolic sub-parameters. Specifically, this can be implemented as follows: only when the deviation remains at least [a certain value] within the immediate window... And the rate of change does not exceed the preset upper limit of sustainable lipomobilization rate. Only when metabolic-related sub-parameters improve in the same direction simultaneously is the immediate beneficial deviation included; among which, the upper limit of sustainable lipomobilization rate... Weight monitoring data are typically taken as 1.0 kg / week (i.e., not exceeding approximately 0.14 kg / day), the rate of decline of continuous glucose baseline is typically taken as 0.05 mmol / (L·h), and the rate of decline of resting heart rate is typically taken as 0.5 bpm / h;

[0098] Concurrent conditions include, but are not limited to: synchronous narrowing of baseline fluctuation amplitude (the sliding standard deviation of the peer window decreases by no less than 20% compared to the silent window), phased deviation decline (the median of the peer window shifts towards the weight loss gain direction by no less than 0.5 times the silent window standard deviation), and narrowing of the area under the curve of continuous glucose over 24 hours compared to the area under the curve corresponding to the plateau baseline; deviations with excessively rapid or large rates of change and short-term reversals within the immediate window are marked as rapid and reversible transient deviations and are not included in the effective gains; the basis for this spurious judgment is that the significant weight loss that occurs in a short period of time after exercise mainly comes from the transient depletion of glycogen-bound water (each gram of glycogen binds approximately 3-4 times the amount of water), rather than an effective improvement in true fat and the metabolic system. Glycogen depletion will be replenished within several hours to 48 hours with food intake and metabolic recovery, therefore, this type of rapid and reversible "false gain" is excluded. This avoids artificially lowering the denominator of the subsequent return ratio due to false gains, thus preventing misjudgment as "the movement gains are inherently strong and the return ratio is low" when the compensation is already strong; the independent method of this lower-level decision layer does not forcibly limit all sub-parameter combinations, and can select several to execute based on data availability.

[0099] Through the above technical solution, this embodiment uses a triple filtering method of "maintenance duration + accompanying sub-parameters in the same direction + upper limit of sustainable rate of change" to filter deviations in metabolic-related data within the immediate window, as well as independent labeling and temporary storage of rapid and reversible transient deviations. This differs from the existing method of directly incorporating the area of ​​unilateral deviation into the benefit statistics. The key difference lies in the addition of a glycogen-water depletion removal mechanism on the benefit side, making the immediate beneficial deviation amount closer to the true metabolic benefit.

[0100] Within the delay window, based on the deviation magnitude and occurrence time of the rapid reversible transient deviation, the response segment that has a magnitude correspondence and a time sequence correspondence with it and shows a reversal trend that deviates from the direction of weight loss benefits is identified as the reversal response segment.

[0101] It should be noted that the magnitude correspondence is defined as: the cumulative magnitude of the candidate response segment within the delay window along the direction of divergence from the weight reduction gain. The magnitude of the deviation from the instantaneous rapid reversible transient within the instant window ratio Falling into the preset range (Typical [0.6, 1.3]), meaning that the return segment is "matched" in magnitude with the transient deviation within the instantaneous window; the temporal correspondence is defined as: the difference between the center time of the candidate response segment and the end time of the fast reversible transient deviation falls within the preset time delay interval. (typical This refers to the reasonable time window for glycogen and water replenishment after physiological depletion.

[0102] The regression trend is determined as follows: the displacement direction of metabolic-related data relative to the platform baseline within the candidate response segment is opposite to the direction of the rapid reversible transient deviation within the instantaneous window; response segments that simultaneously meet the three conditions of amplitude correspondence, temporal correspondence, and opposite direction are identified as regression response segments. The practical basis for this treatment is that the transient deviation caused by glycogen-water depletion and its subsequent fill regression are considered as a pair of events. They are neither independent beneficial deviations nor true metabolic compensatory regressions. If they are not identified and downweighted in the delay window, the fill regression will be mistakenly counted as metabolic regression residuals, thus artificially inflating the value. and .

[0103] The return weighting coefficient is determined based on the magnitude correspondence and the time sequence correspondence, and the weight of the return response segment in the calculation of the corresponding behavioral return residual or metabolic return residual is reduced according to the return weighting coefficient.

[0104] Return weighting coefficient It is obtained by multiplying the amplitude matching degree and the temporal matching degree:

[0105] ;

[0106] in, With a peak value of 1.0, in A membership function that takes the value 1.0 and then linearly decreases to 0 upon exceeding that value. In order to be in A trapezoidal membership function that takes a value of 1.0 within the interval and then linearly decreases to 0 at both ends after exceeding the interval (the typical width of the linear descent region is 2h). For timing difference; If it falls within [0, 1], the closer it is to 0, the more it matches the glycogen-water depletion pairing pattern and the more it should be removed from the feedback residual;

[0107] When specifically included, record If the delay window period is the k-th perturbation anchor point, then the corrected metabolic backflow residual is:

[0108] ;

[0109] in, The metabolic backflow residual after weighting by the backflow response segment. It is the union of the time periods occupied by all the return response segments identified within the delay window. The result of the set difference operation is the remaining non-return time period after deducting all return response segments in the delay window; For the metabolic side, the opposite fact reference template corresponds to the trajectory (i.e. or one), The measured trajectory represents the metabolic-related data from the same pathway. Only the metabolic side should include the insufficient components for unrealized responses; the behavioral side... Similarly, To reduce the weight of the behavior backoff residual after the return response segment, the integrand function is... Replace with ;

[0110] The necessity of this mechanism lies in the fact that if all return response segments are hard-removed directly, the true return components will be lost when the pairing relationship is not clear; if they are not removed, pseudo-returns filled with glycogen-water will be introduced. That is, by using continuous weighting coefficients for soft processing, the return response segments can be approximately removed where the pairing is clear and most of the residual signal can be retained where the pairing is ambiguous.

[0111] Through the above technical solution, this embodiment identifies the rapid reversible transient deviation within the instantaneous window and the return response segment within the delayed window in a three-dimensional pairing of "amplitude-time sequence-direction", and continuously and flexibly subtracts the paired return segment in the return residual integral using a return weighting coefficient. The key is to treat the elimination of glycogen-water transient loss within the instantaneous window and the elimination of the corresponding filling return within the delayed window as two ends of the same spurious removal logic in time, so that "false gains" and "false returns" are identified in pairs rather than being treated in isolation. The resulting corrected metabolic return residual and behavioral return residual replace the two additive components of the uncorrected net return amount, and then enter the molecule of the return ratio.

[0112] Within the delay window, the amount of non-motor activity that is insufficient relative to the trajectory corresponding to the counterfactual reference template is taken as the activity deficiency amount, and the amount of continuous duration of non-motor activity that is lower than the preset low activity level that is extended relative to the continuous duration corresponding to the counterfactual reference template is taken as the sedentary extension amount. The activity deficiency amount and the sedentary extension amount are combined into the post-exercise compensatory rest amount.

[0113] Activity missing quantity We obtain the following formula:

[0114] ;

[0115] in, The trajectory of non-motor activity measured during the plateau period. The counterfactual reference template is used for classmates, and the unit is MET·min; only the "insufficient" side (the part that is actually lower than the counterfactual reference template) is taken, and the excess side is not taken. Since this residual describes the compensatory reduction rather than the active doubling, the excess is not mirrored.

[0116] Low activity level is typically defined as less than 100 steps / hour or less than 90 METs / min. "Continuous duration" is counted after exceeding the preset minimum duration (typically 10 minutes). Sedentary duration extension is also considered. The positive part of the difference between the measured continuous duration of low activity during the plateau period and the corresponding duration of the counterfactual reference template is used as the compensatory resting quantity after exercise. Weighted sum of the two:

[0117] ;

[0118] Since the amount of activity deficit and the amount of prolonged sedentary behavior physically reflect "the absence of expected activity" and "the extension of existing inactivity," respectively, and they are complementary rather than repetitive in compensatory inactivity, this embodiment takes... As a baseline equalizer, and allows for offsetting under different activity data availability conditions (e.g., when step data is unavailable). It can be reduced to 0 and leading).

[0119] The actual strength of the response to this disturbance is determined based on the jump amplitude in the received signature, and the post-motion compensatory static quantity is normalized with the actual strength of the response to obtain the behavioral backflow residual.

[0120] It should be noted that the actual strength of the response The increase in the number of signatures received Obtained by intra-individual Z-standardization transformation (typically...) ),in, The median increase in the rate of increase for similar exercises during the target subject's active weight loss period. (The lower bound is typically 0.2 to prevent division by zero during normalization); behavior backs up residuals. for: ;

[0121] Since direct normalization using registered motion measurements cannot eliminate cases of perfunctory execution (the measurement record may be high, but the actual bearing is weak, and the corresponding post-motion static state is relatively high, which is normal and should not be entirely counted as compensatory backflow), this embodiment uses the jump amplitude as a proxy for "actual bearing strength" for normalization, so that the high static quantity under perfunctory execution is reasonably discounted and avoids being wrongly attributed to strong compensation.

[0122] Through the above technical solution, this embodiment synthesizes the "activity loss" and "prolonged sedentary time" of non-motor activity within the delay window, and uses the jump amplitude of the acceptance signature as the dose benchmark to normalize the resting quantity within the individual. This is different from the existing method of separately statistically analyzing the activity response based on the average activity value or low activity ratio within the delay window. The key difference is that "the expected activity was not realized" and "the existing rest was prolonged" are treated as two complementary aspects of compensatory rest and synthesized, and the behavioral residual and the acceptance signature share the same dose benchmark.

[0123] Based on the construction method of the succession signature, extract the succession signatures of similar planned exercise events during the retrieved weight loss active period;

[0124] It should be noted that for each similar planned movement event during the active weight loss period detected in the counterfactual reference template retrieval phase, its successor signature is extracted using a unified construction method comprising three components: "jump amplitude, start delay, and response start slope," and denoted as... ,in, The jump in motion response data of the j-th hit event within its immediate window relative to the mean of the silent window before its perturbation anchor point. The starting time delay from the moment when the motion response data of the hit event first crosses the jump determination threshold is defined as the time delay from the moment of its disturbance anchor point. The initial slope of the metabolic data within the immediate window of the hit event relative to its plateau baseline; the three components are also Z-standardized within the individual according to the same component distribution during the active weight loss period of this target subject, and the connection signature with the perturbation anchor point during the plateau period. They are in the same standardized coordinate system.

[0125] The degree of similarity between the acceptance signature of the disturbance anchor point and the acceptance signature of the retrieved similar planned motion events in terms of rise amplitude, start time delay, and response start slope is used as the acceptance signature fit degree, and the matching quality is corrected based on the acceptance signature fit degree.

[0126] It should be noted that the signature fit is acceptable. Depend on With the set of event signatures The attenuation function of the average distance over the three components is obtained as follows:

[0127] ;

[0128] in, It is a three-dimensional Euclidean distance (the three components have the same dimensions in a normalized coordinate system). The value falls within (0,1]. The closer it is to 1, the more consistent the plateau disturbance anchor point is with the retrieved hit event in terms of the morphological level of "how the disturbance is received"; the corrected matching quality is:

[0129] ;

[0130] Can be directly from Alternative The necessity of entering continuous weighted aggregation lies in the fact that if only the perturbation intensity range and exercise type are used as the hit conditions, events with "similar doses but significantly different reception patterns" may be included (for example, the current perturbation is an effective reception that quickly enters a high heart rate state, while the hit event is an inefficient reception that slowly climbs the heart rate). Using the two as a reference will introduce implicit bias. Using the reception signature fit as a soft check can continuously correct the credibility of the event as a counterfactual reference based on the degree of morphological similarity without discarding the hit event.

[0131] Through the above technical solution, this embodiment reuses the received signature as the morphological consistency dimension of the counterfactual reference hit event and continuously maps the fit to the correction coefficient for the matching quality using the Euclidean distance decay function. The key is that, in addition to the dose level, a secondary check is added at the "morphological aspect of the perturbation received" level, making the counterfactual reference template closer to the ideal comparison of "the same perturbation received in the same way during the descent phase of the same target object".

[0132] See Figure 2As shown, this solution proposes a system for assessing the compensatory return of exercise benefits during the weight loss plateau phase in obese patients. This system is used to implement the aforementioned method for assessing the compensatory return of exercise benefits during the weight loss plateau phase in obese patients, including:

[0133] The data acquisition module is used to acquire weight monitoring data, planned exercise event records, and simultaneously acquired bypass monitoring data collected over time throughout the weight loss intervention of the target subject. The bypass monitoring data includes non-exercise activity data and metabolic-related data.

[0134] The platform anchoring module is used to identify a platform window where the net change approaches zero and continues for a preset duration based on weight monitoring data. The platform baseline is determined within the time period that is not affected by planned exercise events. The planned exercise events are used to locate disturbance anchor points, and an instant window and a delayed window are constructed with these anchor points as the phase zero points. The disturbance intensity and isolation degree are respectively calibrated based on the movement quantity of the disturbance anchor point and the superposition degree of other disturbances in adjacent time periods.

[0135] The reference construction module is used to retrieve similar planned exercise events with perturbation intensity matching each perturbation anchor point during the active weight loss period when the target object is in front of the platform window and its weight is continuously decreasing. The non-exercise activity volume and metabolism-related response trajectory are extracted according to the same instant window and delayed window as the counterfactual reference template of the perturbation anchor point. The matching quality is obtained by the matching degree of the number and intensity of the retrieved similar planned exercise events.

[0136] The residual synthesis module is used to extract the immediate beneficial deviation from the deviation of metabolic data relative to the platform baseline in the direction of weight loss benefit within the immediate window, and to extract the deviation components of non-exercise activity and metabolic response along their respective weight loss benefit directions within the delayed window. The corresponding deviation components measured in the same window are subtracted from the counterfactual reference template to obtain the behavioral and metabolic feedback residuals and then added together to synthesize the net feedback amount.

[0137] The parameter output module is used to normalize the net return amount with the immediate favorable deviation to obtain the return ratio, which represents the degree of compensation return of exercise gains. The return ratio of each disturbance anchor point within the platform window is continuously weighted and aggregated according to isolation degree, matching quality and directional consistency of the two return residuals to obtain the plateau period compensation return parameter, which is used as the decision support parameter output for weight loss intervention assessment.

[0138] In another embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above embodiments.

[0139] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps described above.

[0140] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and executes the computer instructions, causing the computer device to perform the steps described above.

[0141] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. A method for assessing compensatory retracement of exercise benefits during the weight loss plateau in obese patients, characterized in that... The method includes: Acquire weight monitoring data, planned exercise event records, and simultaneously collected bypass monitoring data collected throughout the weight loss intervention process for the target subjects. The bypass monitoring data includes non-exercise activity data and metabolic-related data. The platform window, which is identified by weight monitoring data and whose net change approaches zero and lasts for a preset duration, is used to determine the platform baseline during the period within which it is not affected by planned exercise events. The planned exercise events are used to locate the disturbance anchor point, and an instant window and a delayed window are constructed with the anchor point as the phase zero point. The disturbance intensity and isolation degree are respectively marked by the movement quantity of the disturbance anchor point and the superposition degree of other disturbances in adjacent periods. During the active weight loss period when the target object is in the platform window and its weight is continuously decreasing, similar planned exercise events with perturbation intensity matching each perturbation anchor point are retrieved. Their non-exercise activity volume and metabolism-related response trajectory are extracted according to the same instant window and delayed window as the counterfactual reference template of the perturbation anchor point. The matching quality is obtained by the matching degree of the number and intensity of the retrieved similar planned exercise events. Within the immediate window, the deviation of metabolic data from the platform baseline toward the direction of weight loss benefits is extracted to obtain the immediate beneficial deviation. Within the delayed window, the deviation components of non-exercise activity and metabolic responses along their respective corresponding weight loss benefit directions are taken. The corresponding deviation components measured in the same window are subtracted from the counterfactual reference template to obtain the behavioral and metabolic feedback residuals, which are then added together to synthesize the net feedback. The net return amount is normalized by the immediate favorable deviation to obtain the return ratio, which represents the degree of compensation return of exercise gains. The return ratios of each disturbance anchor point within the platform window are continuously weighted and aggregated according to isolation degree, matching quality and directional consistency of the two return residuals to obtain the plateau period compensation return parameter, which is used as the decision support parameter output for weight loss intervention assessment.

2. The method according to claim 1, characterized in that, A platform window that identifies net changes in weight monitoring data that approach zero and remain so for a preset duration, specifically including: The segmented rate of decline is calculated from the weight monitoring data, and the segmented rate of decline is attenuated to a preset near-zero range. The period in which the density of planned exercise events and the amount of exercise do not decrease synchronously is determined as the candidate plateau period. Within the reference segment where weight continued to decline during and before the candidate platform period, the ratio of the magnitude of the weight recovery segment to the adjacent weight loss segment was extracted according to a preset local time window to obtain the recovery-decline ratio. The starting point where the rise and fall within the candidate platform period is continuously higher than that of the relative reference segment is determined as the compensation starting boundary, and the portion from the compensation starting boundary to the end of the candidate platform period is determined as the platform window.

3. The method according to claim 1, characterized in that, After locating the disturbance anchor point from the planned motion event and constructing an instant window and a delayed window with that point as the phase zero, the process also includes the construction of the signature, specifically including: The bypass monitoring data also includes exercise response data collected synchronously with the planned exercise event, and the exercise response data includes at least one of heart rate response data and activity intensity data during the planned exercise; Within the instantaneous window of the disturbance anchor point, extract the magnitude of the jump in motion response data relative to the level before the disturbance anchor point, as well as the start delay of the jump relative to the moment of the disturbance anchor point; Within this instant window, the initial slope of the metabolic data relative to the platform baseline is extracted, and the jump magnitude, initial delay, and initial slope of the response are combined to form the acceptance signature of the disturbance anchor point.

4. The method according to claim 1, characterized in that, Within the immediate window, the deviation of metabolic-related data from the platform baseline in the direction of weight loss benefits is used to extract the immediate favorable deviation, specifically including: Among the deviations of metabolic-related data from the platform baseline toward weight loss benefits, those that remain for more than a preset period of time and show a narrowing of the baseline fluctuation range or a phased decline in the metabolic-related data during the same period are identified as valid deviations. The portion of the deviation whose rate of change exceeds the preset sustainable rate of change range and returns briefly within the instantaneous window is marked as a fast reversible transient deviation. Record the magnitude and timing of rapid reversible transient deviations, summarize the magnitudes of effective deviations, and obtain the immediate favorable deviation amount.

5. The method according to claim 4, characterized in that, The process of obtaining and summing the behavioral and metabolic feedback residuals to synthesize the net feedback amount also includes the weighting of the feedback response segment within the delay window, specifically including: Within the delay window, based on the deviation magnitude and occurrence time of the rapid reversible transient deviation, the response segment that has a magnitude correspondence and a time sequence correspondence with it and shows a reversal trend that deviates from the direction of weight loss benefits is identified as the reversal response segment. The return weighting coefficient is determined based on the magnitude correspondence and the time sequence correspondence, and the weight of the return response segment in the calculation of the corresponding behavioral return residual or metabolic return residual is reduced according to the return weighting coefficient.

6. The method according to claim 3, characterized in that, The extraction of the behavioral backflow residual specifically includes: Within the delay window, the amount of non-motor activity that is insufficient relative to the trajectory corresponding to the counterfactual reference template is taken as the activity deficiency amount, and the amount of continuous duration of non-motor activity that is lower than the preset low activity level that is extended relative to the continuous duration corresponding to the counterfactual reference template is taken as the sedentary extension amount. The activity deficiency amount and the sedentary extension amount are combined into the post-exercise compensatory rest amount. The actual strength of the response to this disturbance is determined based on the jump amplitude in the received signature, and the post-motion compensatory static quantity is normalized using the actual strength of the response to obtain the behavioral backflow residual.

7. The method according to claim 3, characterized in that, The process of determining the matching quality by analyzing the quantity and intensity of similar planned sports events retrieved also includes verifying the signature matching, specifically including: Extract the signatures of similar planned exercise events during the active weight loss period based on the construction method of the signatures. The degree of similarity between the acceptance signature of the disturbance anchor point and the acceptance signature of the retrieved similar planned motion events in terms of rise amplitude, start time delay, and response start slope is used as the acceptance signature fit degree, and the matching quality is corrected based on the acceptance signature fit degree.

8. A system for assessing the compensatory return of exercise benefits during the weight loss plateau in obese patients, characterized in that: The method for assessing compensatory retracement of exercise gains during the weight loss plateau phase in obese patients as described in any one of claims 1-7 includes: The data acquisition module is used to acquire weight monitoring data, planned exercise event records, and synchronously acquired bypass monitoring data collected over time throughout the weight loss intervention of the target subject. The bypass monitoring data includes non-exercise activity data and metabolic-related data. The platform anchoring module is used to identify a platform window where the net change approaches zero and continues for a preset duration based on weight monitoring data. The platform baseline is determined within the time period that is not affected by planned exercise events. The planned exercise events are used to locate disturbance anchor points, and an instant window and a delayed window are constructed with these anchor points as the phase zero points. The disturbance intensity and isolation degree are respectively calibrated based on the movement quantity of the disturbance anchor point and the superposition degree of other disturbances in adjacent time periods. The reference construction module is used to retrieve similar planned exercise events with perturbation intensity matching each perturbation anchor point during the active weight loss period when the target object is in front of the platform window and its weight is continuously decreasing. The non-exercise activity volume and metabolism-related response trajectory are extracted according to the same instant window and delayed window as the counterfactual reference template of the perturbation anchor point. The matching quality is obtained by the matching degree of the number and intensity of the retrieved similar planned exercise events. The residual synthesis module is used to extract the immediate beneficial deviation from the deviation of metabolic data relative to the platform baseline in the direction of weight loss benefit within the immediate window, and to extract the deviation components of non-exercise activity and metabolic response along their respective weight loss benefit directions within the delayed window. The corresponding deviation components measured in the same window are subtracted from the counterfactual reference template to obtain the behavioral and metabolic feedback residuals and then added together to synthesize the net feedback amount. The parameter output module is used to normalize the net return amount with the immediate favorable deviation to obtain the return ratio, which represents the degree of compensation return of exercise gains. The return ratio of each disturbance anchor point within the platform window is continuously weighted and aggregated according to isolation degree, matching quality and directional consistency of the two return residuals to obtain the plateau period compensation return parameter, which is used as the decision support parameter output for weight loss intervention assessment.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method described in any one of claims 1-7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-7.

Images