Stretching force adaptive adjustment method and system based on sleep data recognition

Abstract

The application relates to a stretching force self-adaptive adjusting method and system based on sleep data recognition, and relates to the technical field of body state intelligent intervention and sleep data linkage. First, sleep data packets and body movement data are acquired, local body movement data is used as a correction factor to perform weighted fusion correction on original sleep staging labels, and standardized sleep stage judgment values of four types of stages are obtained; then, according to user age information, matched age group parameter intervals are called from a matching database, and further based on the standardized sleep stage judgment values, target air pressure values and target inflation interval cycles associated with the current stage are matched; finally, a main control pressure regulating execution module is driven to execute segmented pulse pressure regulation of a lower limb stretching air bag, segmented pressure relief or segmented inflation is executed according to the high-low relationship between actual air pressure and target air pressure, and thus, the stretching force adapted to the sleep stage is output. The application improves the use experience, use compliance and adaptability of the stretching equipment.

Description

Technical Field

[0001] This application relates to the technical field of intelligent intervention of body posture and linkage with sleep data, and in particular to a method and system for adaptive adjustment of stretching force based on sleep data recognition. Background Technology

[0002] In recent years, with the increasing awareness of family health management and the growing demand for home fitness, home stretching equipment has gradually become a common tool for stretching and relaxing lower limb muscles at night. These devices typically apply periodic tension to the lower limbs through airbags or mechanical mechanisms, completing preset stretching programs in conjunction with nighttime sleep. Because users' tolerance to external tension and their body movement response characteristics differ at different sleep stages, whether the stretching intensity can match their real-time sleep state directly affects the user's sleep continuity and device usage compliance.

[0003] In related technologies, home stretching devices generally adopt a fixed tension mode or a time-based segmented preset mode, that is, the user sets the total time and level in advance and then executes according to a fixed curve; some solutions add sensors to the device locally and adjust the level according to the physical signals collected by the sensors; other solutions connect to a sleep physiological data acquisition device to read the sleep state information output by the device and then drive the device to execute.

[0004] However, the above solutions are not ideal in actual use. There is often a discrepancy between the device's operating status and the user's actual physical condition at night, and the user experience and user compliance need to be improved. Summary of the Invention

[0005] This application provides a method and system for adaptive adjustment of stretching force based on sleep data recognition, which is used to adapt the operating state of the stretching device to the user's actual physical state at night, thereby improving the user experience and user compliance.

[0006] In a first aspect, this application provides a method for adaptive adjustment of stretching force based on sleep data recognition, the method comprising: Acquire sleep data packets output by the sleep physiological data acquisition device, collect body movement data output by the local sensor of the stretching device, and correct the original sleep stage labels in the sleep data packets based on the weighted fusion of the body movement data to obtain standardized sleep stage determination values. Based on the user's age information, the matching age group parameter range is retrieved from the pre-built age-specific sleep intensity matching database. Each age group in the age-specific sleep intensity matching database is configured with air pressure values ​​and inflation intervals associated with four sleep stages: wakefulness, light sleep, deep sleep, and REM sleep. Based on the standardized sleep stage determination value, the target air pressure value and target inflation interval are matched within the age group parameter range to obtain the current sleep stage. Based on the target air pressure value and the target inflation interval, the lower limb stretching airbag is driven by the main control pressure regulation execution module to perform segmented pulse pressure regulation. When the actual air pressure of the lower limb stretching airbag is higher than the target air pressure value, it is depressurized in segments to the target air pressure value. When the actual air pressure of the lower limb stretching airbag is lower than the target air pressure value, it is inflated in segments to the target air pressure value, thereby obtaining a stretching force that matches the standardized sleep stage determination value.

[0007] In the above embodiments, this application first acquires the sleep data package output by the sleep physiological data acquisition device and the body movement data collected by the local sensor of the stretching device in parallel. The local body movement data is used as a correction factor to perform weighted fusion correction on the original sleep stage label to obtain a standardized sleep stage judgment value covering four stages: wakefulness, light sleep, deep sleep, and REM sleep. Secondly, according to the user's age information, the matching age group parameter range is retrieved from the pre-constructed age-specific sleep intensity matching database. Furthermore, based on the standardized sleep stage judgment value, the target air pressure value and target inflation interval associated with the current stage are matched. Finally, the lower limb stretching airbag is driven by the main control pressure regulation execution module to perform segmented pulse pressure regulation. The airbag is depressurized or inflated in segments according to the relationship between the actual air pressure and the target air pressure, thereby outputting a stretching force that is adapted to the sleep stage. This application also uses the age-specific parameter library and segmented pulse pressure regulation to make the stretching force change smoothly with the differences in sleep stage and age group, reducing the proportion of nighttime awakenings caused by sudden changes in air pressure, and improving the user experience, user compliance, and adaptability of stretching force and usage status.

[0008] In some embodiments, the sleep data package includes heart rate, heart rate variability, number of times the patient turns over per unit of time, and original sleep stage labels.

[0009] In the above embodiments, this application standardizes the data format requirements for cross-brand wearable devices to access this method, providing a clear field naming basis for subsequent heart rate abnormality determination and fusion correction, and improving cross-device compatibility.

[0010] In some embodiments, the step of correcting the original sleep stage labels in the sleep data packet based on the weighted fusion of the body movement data to obtain a standardized sleep stage determination value includes: Extract limb vibration amplitude and continuous motion identifiers based on the limb vibration amplitude from the body motion data; Using the limb vibration amplitude and the continuous movement identifier as correction factors, the original sleep stage labels in the sleep data package are corrected according to a preset weighting coefficient to obtain the standardized sleep stage determination value. The standardized sleep stage determination value is taken as the label value corresponding to wakefulness, light sleep, deep sleep or REM sleep.

[0011] In the above embodiments, this application simultaneously extracts the instantaneous amplitude feature of limb vibration amplitude and the temporal pattern feature of continuous movement identification based on the amplitude from the body movement data collected by local sensors. The two are used as correction factors to correct the original sleep stage labels output by the wearable device according to preset weighting coefficients, and the correction results are uniformly mapped to four types of label values: wakefulness, light sleep, deep sleep, and REM sleep. The parallel fusion of instantaneous and temporal signals improves the accuracy of sleep stage identification, while ensuring that the output results are consistent with the value naming of the downstream age parameter library, which facilitates subsequent matching.

[0012] In some embodiments, obtaining the standardized sleep stage determination value specifically includes: The result obtained by correcting the original sleep stage label according to the preset weighting coefficient is used as the intermediate judgment value; Read the number of times the patient turns over within a unit of time in the sleep data packet; When the number of times the body turns over is not higher than the first turning over threshold, the intermediate judgment value is determined as the standardized sleep stage judgment value; When the number of times the body turns over is higher than the first turning over threshold but not higher than the second turning over threshold, if the intermediate judgment value indicates deep sleep, then the intermediate judgment value is lowered to the identification value corresponding to light sleep and determined as the standardized sleep stage judgment value. When the number of times the patient turns over is higher than the second turning over threshold and the intermediate judgment value indicates light sleep or deep sleep, the intermediate judgment value is adjusted up to the identifier value corresponding to wakefulness and then determined as the standardized sleep stage judgment value.

[0013] In the above embodiments, this application introduces the number of times the body turns over per unit time in the sleep data package as a secondary verification basis for weighted fusion correction. The number of turns over is divided into three levels for differentiated processing through a first turning over threshold and a second turning over threshold: when the number of turns over is low, the intermediate judgment value remains unchanged; when the number of turns over is in the middle range, the intermediate judgment value indicating deep sleep is suppressed in reverse to avoid misjudging the moderate body movement stage as deep sleep due to the lag of the original sleep stage label; when the number of turns over is high, the intermediate judgment value is directly adjusted to the label value corresponding to wakefulness to form a wakefulness warning. This not only improves the ability of the standardized sleep stage judgment value to distinguish between light sleep, REM sleep and wakefulness aura stages, but also enables the weighted fusion correction process to cover three dimensions: instantaneous body movement signal, temporal movement pattern and turning over frequency, further reducing the deviation between stretching force and real-time body state.

[0014] In some embodiments, prior to the weighted fusion correction, the method further includes: Abnormal data points in the sleep data package are filtered out. The abnormal data points include signal loss data points and jump data points where the heart rate exceeds the preset normal physiological range. The abnormal data points that are continuously filtered are accumulated to obtain an abnormal count; When the abnormal count reaches a preset threshold, it is determined that the sleep physiological data acquisition device is currently disconnected or data is unavailable, and it switches to the local recognition mode based on the body movement data. The local recognition mode outputs the backup sleep stage determination value. The standby sleep stage determination value is taken as the identifier value corresponding to wakefulness, light sleep, deep sleep or REM sleep, and the standby sleep stage determination value replaces the standardized sleep stage determination value to continue to perform the matching of the target air pressure value and the target inflation interval cycle and the segmented pulse pressure regulation.

[0015] In the above embodiments, wearable devices are susceptible to data loss or heart rate fluctuations due to factors such as battery level, loose fit, and Bluetooth jitter at night. If instantaneous disconnection or single abnormality is used as the trigger condition for switching local recognition, frequent erroneous switching under brief disturbances can easily occur, thus reducing system stability. This application filters out signal loss points and heart rate fluctuations exceeding the preset normal physiological range in the sleep data packets before weighted fusion correction, and accumulates and counts the continuously filtered abnormal data points. When the abnormal count reaches a preset threshold, the wearable device is determined to be in a disconnected or unavailable data state and switches to the local recognition mode based on body movement data. The output backup sleep stage judgment value adopts the same naming as the standardized sleep stage judgment value and directly continues the subsequent parameter matching and voltage adjustment process. This not only avoids erroneous switching caused by instantaneous disturbances, but also makes the dual-path recognition switching seamlessly connected at the algorithm layer, improving the system robustness in wearable disconnection scenarios.

[0016] In some embodiments, the age-based sleep intensity matching database includes multiple age groups, each of which corresponds to the air pressure value and inflation interval cycle associated with four sleep stages: wakefulness, light sleep, deep sleep, and REM sleep. In any of the age groups, the barometric pressure associated with deep sleep is greater than that associated with light sleep, which in turn is greater than that associated with wakefulness. In any of the age groups, the inter-sleep intervals associated with deep sleep are shorter than those associated with light sleep, which in turn are shorter than those associated with wakefulness. In any of the age groups, the air pressure associated with REM sleep is not higher than that associated with light sleep, and the air intermittent period associated with REM sleep is not less than that associated with light sleep.

[0017] In the above embodiments, since the lower limb muscle groups and fascia of users of different ages have different tolerances, if the same set of air pressure and pressure regulation rhythm parameters are used for all age groups, it is easy to cause excessive force to younger users or insufficient force to older users. This application divides the age-specific sleep intensity matching database into multiple age groups, and configures the air pressure values ​​and inflation intervals associated with the four stages of wakefulness, light sleep, deep sleep, and REM sleep for each age group. The parameter database is constrained by a structural inequality relationship of "deep sleep air pressure is greater than light sleep air pressure is greater than wakefulness air pressure" and "deep sleep interval is less than light sleep interval is less than wakefulness interval". At the same time, the air pressure value associated with REM sleep is constrained to be no higher than the light sleep air pressure value, and the inflation interval is constrained to be no less than the light sleep inflation interval. This not only makes the stretching force and age group-sleep stage form a two-dimensional differentiated match, but also reduces the risk of decreased comfort caused by REM dream limb activities, and improves the adaptability of stretching force and usage status.

[0018] In some embodiments, during segmented pulse pressure regulation, the target inflation interval period is maintained between two adjacent segmented inflation or deflation.

[0019] In the above embodiments, if segmented pulse pressure regulation uses a fixed segment interval, it cannot be coordinated with the pressure regulation rhythm required for different sleep stages. For example, during deep sleep, users have little body movement and can tolerate relatively compact pressure changes, while during light sleep, more gentle changes are needed to avoid waking up. This application uses the target inflation interval period matched from the age-specific parameter library as the time interval between two adjacent segmented inflations or deflations, so that the rhythm of pressure changes is linked with the sleep stage and age group. It synchronously controls both the pressure amplitude and the rhythm of change, making the pressure regulation response during deep sleep more compact and the pressure regulation transition during light sleep more gentle, reducing the proportion of nighttime awakenings caused by pressure changes.

[0020] In some embodiments, the method further includes at least one of the following: When the duration of continuous wakefulness as determined by the standardized sleep stage judgment value reaches the preset wakefulness threshold, the lower limb stretching airbag is depressurized to the preset minimum protective air pressure and enters a shutdown standby state. When the cumulative running time after the segmented pulse pressure regulation start reaches the preset maximum running threshold, the lower limb stretching airbag is depressurized to the preset minimum protective air pressure and enters the shutdown standby state.

[0021] In the above embodiments, this application sets two independent shutdown protection conditions: the first is based on the duration of continuous wakefulness determined by the standardized sleep stage judgment value, corresponding to the user's short-term awakening scenario, so that the shutdown decision is synchronized with the user's actual state; the second is based on the cumulative running time after the segmented pulse pressure regulation is started, corresponding to the user's overuse scenario of forgetting to turn off the device all night, avoiding the decrease in comfort caused by cumulative overuse; both types of protection adopt a two-stage action of first depressurizing to the preset minimum protection air pressure and then entering shutdown standby, avoiding sudden changes in the air path caused by one-time depressurization, improving the user experience and safety.

[0022] In some embodiments, the method further includes: A hardware pressure upper limit threshold is preset. When the target pressure value exceeds the hardware pressure upper limit threshold, the segmented pulse pressure regulation is performed instead of the target pressure value by using the hardware pressure upper limit threshold.

[0023] In the above embodiments, this application pre-sets a hardware air pressure upper limit threshold. When the target air pressure value output by the algorithm exceeds the hardware air pressure upper limit threshold, the hardware air pressure upper limit threshold is used to replace the target air pressure value to perform segmented pulse pressure regulation. The arbitration function of the hardware air pressure upper limit threshold is independent of the configuration correctness of the age-based sleep intensity matching database. It can effectively deal with abnormal runtime scenarios such as the age-based sleep intensity matching database being issued incorrect parameters through remote upgrades, user age information being mistakenly entered, and the target air pressure value exceeding the safe range due to numerical overflow during calculation. Together with the parameter upper limit constraint in the database construction stage, it forms a dual protection at the algorithm layer and hardware layer, reducing the risk of overpressure drive caused by parameter database abnormalities or input errors, and improving the safety of product use.

[0024] Secondly, this application provides a stretching force adaptive adjustment system based on sleep data recognition. The system includes a sleep physiological data acquisition device, a data transmission module, a sleep data cleaning, correction and fusion module, an age-specific sleep force matching database, a main control pressure adjustment execution module, a lower limb stretching airbag and a local body motion sensing backup module. The sleep physiological data acquisition device is configured to collect the user's sleep physiological parameters and output a sleep data package, the sleep data package containing original sleep stage labels; The data transmission module is configured to establish a data transmission link between the sleep physiological data acquisition device and the sleep data cleaning, correction and fusion module, and to transmit the sleep data packet to the sleep data cleaning, correction and fusion module via the data transmission link; The local motion sensing backup module is configured to collect motion data output by the local sensor of the stretching device and output the motion data to the sleep data cleaning, correction and fusion module; The sleep data cleaning, correction and fusion module is configured to correct the original sleep stage labels in the sleep data package based on the weighted fusion of the body movement data to obtain standardized sleep stage determination values. The age-specific sleep intensity matching database is configured to store age-specific sleep-stretching intensity matching parameters. In the age-specific sleep-stretching intensity matching parameters, each age group is configured with the air pressure value and inflation interval cycle associated with four sleep stages: wakefulness, light sleep, deep sleep, and REM sleep. The main control pressure regulation execution module is configured to retrieve the matching age group parameter range from the age-specific sleep intensity matching database according to the user's age information, match the target air pressure value and target inflation interval period associated with the current sleep stage in the age group parameter range based on the standardized sleep stage judgment value, and drive the lower limb stretching airbag to perform segmented pulse pressure regulation according to the target air pressure value and target inflation interval period. The lower limb stretching airbag is configured to, under the drive of the main control pressure regulating execution module, depressurize in stages to the target pressure value when the actual air pressure is higher than the target air pressure value, and inflate in stages to the target air pressure value when the actual air pressure is lower than the target air pressure value, so as to obtain a stretching force that matches the standardized sleep stage determination value.

[0025] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. By using the original sleep stage labels output by the sleep physiological data acquisition device and the limb vibration amplitude and continuous movement identifiers collected by the local sensors of the stretching device as correction factors for weighted fusion correction, and mapping the correction results to standardized judgment values ​​covering four sleep stages, the recognition accuracy and subsequent control link compatibility in cross-brand wearable device access scenarios are improved, so that the operating status of the stretching device is adapted to the user's actual physical state at night, improving the user experience and user compliance of the device. 2. By using the filtering results of abnormal sleep data packets as the trigger for switching local recognition modes through continuous cumulative counting, and ensuring that the value of the backup sleep stage judgment output after switching is consistent with the value of the main path, the erroneous switching caused by instantaneous communication jitter or single heart rate jump is avoided, the continuity of the algorithm link in the wearable disconnection scenario is improved, and the robustness of the system in long-term use at night is enhanced. 3. By constraining the air pressure corresponding to different ages using an age-specific parameter library, and simultaneously reusing the target inflation interval cycle in the parameter library as the time interval between adjacent segments of segmented pulse pressure regulation, the air pressure amplitude and change rhythm change in conjunction with the sleep stage and age group, reducing the proportion of nighttime awakenings caused by sudden changes in air pressure, and improving the differentiated adaptation between the stretching force and the user's actual physical condition at night. Attached Figure Description

[0026] Figure 1 This is an overall architecture block diagram of a stretching force adaptive adjustment system based on sleep data recognition in an embodiment of this application; Figure 2 This is a flowchart illustrating an adaptive adjustment method for stretching force based on sleep data recognition in an embodiment of this application. Figure 3 This is an age-specific sleep intensity matching diagram according to an embodiment of this application; Figure 4 This is a schematic diagram comparing the timing of air pressure stretching in four types of sleep cycle airbags in the embodiments of this application. Detailed Implementation

[0027] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification of this application, the singular expressions “a,” “an,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to any or all possible combinations including one or more of the listed items.

[0028] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0029] To facilitate understanding, the application scenarios of the embodiments of this application are described below.

[0030] In the field of sleep data linkage control for home stretching devices, home stretching devices in the home setting are gradually becoming a common tool for stretching and relaxing the lower limb muscles at night. Their core use case is to apply periodic micro-stretching to the lower limbs during the nighttime sleep period.

[0031] In related technologies, home stretching devices generally adopt a fixed tension mode or a time-based segmented preset mode. Some solutions also add sensors to the device to collect physical signals and adjust the intensity level, or connect to wearable devices to read sleep status information and drive the device to perform the operation. However, the actual operating state of these solutions often deviates from the user's actual physical state at night, resulting in room for improvement in user experience and user compliance.

[0032] This application is primarily applied to scenarios involving lower limb stretching during nighttime sleep, where the stretching intensity needs to be adapted to the user's real-time sleep state and age. In these applications, a key challenge is how to dynamically match the stretching intensity with the real-time sleep stage and age without disrupting sleep continuity. To address this technical problem, this application provides a method for adaptively adjusting stretching intensity based on sleep data recognition. The following description, using an embodiment and in conjunction with the accompanying drawings, illustrates such a method.

[0033] Please see Figure 1 , Figure 1 This application provides an overall architecture block diagram of a stretching force adaptive adjustment system based on sleep data recognition, as provided in an embodiment of the present application. The system includes a sleep physiological data acquisition device, a data transmission module, a sleep data cleaning, correction and fusion module, an age-specific sleep force matching database, a main control pressure adjustment execution module, a lower limb stretching airbag, and a local body motion sensing backup module.

[0034] The sleep physiological data acquisition device can be a third-party wearable device used to collect the user's sleep physiological parameters and output a sleep data package when the user wears it at night. These sleep physiological parameters include heart rate, heart rate variability, number of turns per unit time, and original sleep stage labels. Alternatively, it can be other data acquisition devices capable of performing the same function, such as sleep monitoring mattresses based on piezoelectric films or impact mapping principles, or non-wearable sleep monitoring devices based on millimeter waves or ultra-wideband sleep radar. This application does not limit the specific form of the sleep physiological data acquisition device; any device capable of outputting one or more of the aforementioned sleep physiological parameters and normalizing them into a sleep data package falls within the scope of the sleep physiological data acquisition device described in this application. The data transmission module is used to establish a connection between the sleep physiological data acquisition device and the sleep data cleaning module. The system includes a data transmission link between calibration and fusion modules; a local motion sensing backup module for collecting motion data output from the local sensors of the stretching device and outputting the motion data to the sleep data cleaning, calibration, and fusion module; a sleep data cleaning, calibration, and fusion module for weighted fusion calibration of the original sleep stage labels in the sleep data package based on motion data and outputting standardized sleep stage judgment values; an age-specific sleep intensity matching database for storing age-specific sleep-stretching intensity matching parameters; a main control pressure regulation execution module for retrieving the target air pressure value and target inflation interval from the age-specific sleep intensity matching database based on the user's age information and standardized sleep stage judgment values, and driving the lower limb stretching airbag to perform segmented pulse pressure regulation according to the above results; and a lower limb stretching airbag for changing the actual air pressure according to the segmented pulse method under the drive of the main control pressure regulation execution module.

[0035] Please see Figure 2 , Figure 2This is a flowchart illustrating an embodiment of the present application. The method generally includes steps S201 to S204, which are detailed below.

[0036] S201. Acquire the sleep data package output by the sleep physiological data acquisition device, acquire the body movement data output by the local sensor of the stretching device, and correct the original sleep stage label in the sleep data package based on the weighted fusion of the body movement data to obtain the standardized sleep stage judgment value.

[0037] Among them, sleep data package refers to a structured data set containing the user's nighttime sleep physiological parameters, which is output by the sleep physiological data acquisition device through the data transmission module; body movement data refers to physical signals that reflect the user's nighttime limb movement characteristics, which are collected by the local sensors of the stretching device; and standardized sleep stage judgment value refers to a unified identification value that covers the four stages of wakefulness, light sleep, deep sleep and REM sleep, which is output by the sleep data cleaning, correction and fusion module.

[0038] Specifically, the system first initiates a data synchronization request to the sleep physiological data acquisition device via the data transmission module; then it receives the sleep data package output by the sleep physiological data acquisition device, which includes heart rate, heart rate variability, number of turns per unit time, and original sleep stage labels; next, the local motion sensing backup module collects the motion data output by the local sensor of the stretching device; finally, the sleep data cleaning, correction and fusion module performs weighted fusion correction on the original sleep stage labels in the sleep data package based on the motion data to obtain standardized sleep stage determination values.

[0039] It should be noted that the number of turns within a unit of time in the sleep data packet is preferably the number of turns within 5 minutes in this embodiment, but it can also be adjusted to the number of turns within 3 minutes or 10 minutes depending on the output granularity of the wearable device. This application does not limit this. The data transmission module can realize data synchronization request and data reception through any of the following parallel methods: First, establish a local direct connection channel between the sleep physiological data acquisition device and the main control terminal of the stretching device through Bluetooth Low Energy, and pull the sleep data packet through the local direct connection channel; Second, the sleep physiological data acquisition device pre-synchronizes the sleep data to the cloud server, and the stretching device calls the cloud server interface through the wireless network to pull the sleep data packet. The above two transmission methods can be implemented selectively or deployed in parallel in this step. For example, the Bluetooth channel can be the main one, and the cloud interface can be used as a supplementary link when the Bluetooth link is interrupted. This application does not limit this.

[0040] In this embodiment, the weighted fusion correction is performed by the sleep data cleaning, correction, and fusion module. Specifically, the sleep data cleaning, correction, and fusion module first extracts limb vibration amplitude and continuous motion identifiers based on limb vibration amplitude recognition from the body movement data. The limb vibration amplitude reflects the instantaneous amplitude characteristics of the body movement signal, while the continuous motion identifier is used to characterize the temporal pattern characteristics of the body movement signal, such as short single movements or continuous long-term movements. Then, using the limb vibration amplitude and continuous motion identifiers as correction factors, the original sleep stage labels in the sleep data package are corrected according to a preset weighting coefficient. Next, the correction result is mapped to a unified identifier value namespace. The standardized sleep stage determination value S takes the values ​​of identifier 0 for wakefulness, identifier 1 for light sleep, identifier 2 for deep sleep, or identifier 3 for REM sleep. Finally, the standardized sleep stage determination value S is output to the downstream main control voltage regulation execution module.

[0041] In this embodiment, the weighted fusion correction adopts a two-stage hierarchical determination specific implementation rule: The first stage involves preliminary identification of REM sleep. The sleep data cleaning, correction, and fusion module determines whether the user is currently in REM sleep based on the original sleep stage labels and the continuous motion identifiers. When the original sleep stage label indicates REM sleep or the continuous motion identifier indicates a brief high-frequency micro-motion pattern, the intermediate judgment value is directly determined as the identifier value 3 corresponding to REM sleep; otherwise, the second stage begins. The second stage involves weighted fusion of the three sleep stages—wakefulness, light sleep, and deep sleep—based on monotonic sleep depth. First, the original sleep stage labels are mapped to label depth values ​​L∈{0,1,2} according to the monotonic encoding rules for sleep depth, where wakefulness is mapped to 0, light sleep to 1, and deep sleep to 2. Then, based on the normalization of limb vibration amplitude to amplitude depth values ​​A∈{0,1,2}—A is 0 when the limb vibration amplitude is in the high range, 1 when it is in the middle range, and 2 when it is in the low range; based on the continuous motion identifiers… The action identifier is normalized to a temporal depth value D∈{0,1,2}—D is 0 when continuous large-amplitude action is indicated, 1 when intermittent medium-amplitude action is indicated, and 2 when no action or brief micro-movement is indicated; then, a weighted summation is performed according to the following formula: S''=α·L+β·A+γ·D, where α, β, and γ are preset weighting coefficients and α+β+γ=1; finally, S'' is rounded and limited to the {0,1,2} interval, and mapped back to the identifier values ​​corresponding to wakefulness, light sleep, or deep sleep as intermediate judgment values.

[0042] Through the above two-stage stratification, REM sleep, a shallow active state accompanied by limb movements, is separated from the continuous weighting process in advance, avoiding the ambiguity in physical meaning that would arise when it is included in the continuous weighting process that is monotonically encoded according to sleep depth.

[0043] It should be noted that the preset weighting coefficients can be pre-calibrated by those skilled in the art based on the confidence level of the wearable device's native sleep staging algorithm. For example, for wearable brands with high confidence, the weight of the wearable native tag can be set to 0.7 and the weight of the body movement correction factor can be set to 0.3; for wearable brands with low confidence or whose algorithms are not publicly disclosed, the weights of the two can be set to 0.5 and 0.5 respectively. This application does not limit the specific coefficient values. In this embodiment, the continuous motion identifier can be determined based on whether the cumulative value of limb vibration amplitude within the sliding time window exceeds a preset amplitude threshold, or it can be identified using a state machine method based on threshold segmentation. This application does not limit this method.

[0044] Furthermore, the number of times a person turns over per unit time in the sleep data packet is used as an auxiliary correction factor in the weighted fusion correction process, specifically for secondary verification and bias compensation of the correction results of the original sleep stage labels.

[0045] Specifically, after obtaining the intermediate judgment value based on the weighted correction of limb vibration amplitude and continuous movement identifier, the sleep data cleaning, correction and fusion module further reads the number of times the body turns over within a unit of time in the sleep data packet and compares this number of times with a pre-set turning frequency classification threshold: When the number of turning over is not higher than the first turning over threshold, the user is determined to have sparse limb movement, and the intermediate judgment value remains unchanged. When the number of turning over is higher than the first turning over threshold but not higher than the second turning over threshold, the user is determined to be in a moderate body movement state corresponding to light sleep or REM sleep. If the intermediate judgment value indicates deep sleep, the intermediate judgment value is lowered to the label value corresponding to light sleep to avoid misjudging the moderate body movement stage as deep sleep due to the lag of the original sleep stage label. When the number of turning over is higher than the second turning over threshold, the user is determined to be in a state of pre-awakening or awake. If the intermediate judgment value indicates light sleep or deep sleep, the intermediate judgment value is raised to the label value corresponding to awakening, and the main control pressure regulation execution module is triggered to depressurize the lower limb stretching airbag to the preset minimum protective air pressure in advance, so as to unload the stretching tension before the user is fully awake and avoid the continuous tension of the airbag causing additional interference to the user's short-term awakening process.

[0046] It should be noted that, in this embodiment, the first turning-over threshold and the second turning-over threshold are preferably 2 times within 5 minutes and 4 times within 5 minutes, respectively. Those skilled in the art can also adjust the above thresholds according to the turning-over detection sensitivity of the wearable device and individual differences in users, and this application does not limit this. The number of turning over plays a three-fold role in this process: firstly, as a reverse inhibitory factor against misjudging deep sleep, preventing the wearable device from misjudging the moderate body movement stage as deep sleep due to the lag of the original algorithm; secondly, as an auxiliary basis for identifying light sleep and REM sleep stages, improving the stage discrimination of the standardized sleep stage judgment value; and thirdly, as a wake-up warning signal, working in conjunction with the wakefulness termination protection logic to achieve early pressure reduction, further reducing the proportion of nighttime awakenings caused by sudden changes in air pressure.

[0047] Furthermore, prior to the weighted fusion correction, the sleep data cleaning, correction, and fusion module also performs anomaly filtering and fault-tolerance switching processes. These are detailed below: The sleep data cleaning, correction, and fusion module first filters out abnormal data points in the sleep data package. Abnormal data points include data points with lost signals and data points with abrupt changes in heart rate exceeding the preset normal physiological range. Then, it accumulates the continuously filtered abnormal data points to obtain an abnormality count. Next, when the abnormality count reaches a preset threshold, it is determined that the sleep physiological data acquisition device is currently in a disconnected or unavailable state. The control system switches from the main recognition path to a local recognition mode based on body movement data, and the local body movement sensor backup module outputs a backup sleep stage determination value. Finally, the backup sleep stage determination value is output downstream according to the same value namespace as the standardized sleep stage determination value, replacing the standardized sleep stage determination value to continue participating in subsequent parameter matching and voltage regulation.

[0048] It should be noted that, in this embodiment, the preset normal physiological range is preferably a heart rate of not less than 40 beats / minute and not more than 160 beats / minute. Heart rate readings exceeding this range are considered jump data points and are directly discarded. The preset threshold is preferably an abnormal count of not less than 3 times in this embodiment. Those skilled in the art can also adjust it to 2 or 5 times according to the actual output stability of the wearable device. This application does not limit this. In this embodiment, the local recognition mode is independently completed by the local body motion sensing backup module to identify the four types of sleep stages. Based on the amplitude of limb vibration, continuous action markers and their cumulative change patterns, the local body motion sensing backup module can adopt the following judgment rules: when the continuous action marker indicates continuous large movements and the amplitude of limb vibration is in the high range, it is judged as the marker value corresponding to wakefulness; when the continuous action marker indicates intermittent medium amplitude movements, it is judged as the marker value corresponding to light sleep; when the continuous action marker indicates no movement and the amplitude of limb vibration is in the low range for a long time, it is judged as the marker value corresponding to deep sleep; when the continuous action marker indicates brief high-frequency micro-movements and the amplitude of limb vibration is in the low to medium range, it is judged as the marker value corresponding to REM sleep. Since the output set of the local identification mode is consistent with the standardized sleep stage judgment value, the logic of the downstream main control voltage regulation execution module does not need to be modified for any adaptation action.

[0049] S202. Retrieve the matching age group parameter range from the pre-built age-based sleep intensity matching database based on the user's age information.

[0050] Among them, user age information refers to the age parameters that the user or guardian enters in advance when the stretching equipment is initially configured; the age-specific sleep intensity matching database refers to the data set that stores age-specific sleep-stretching intensity matching parameters, which are organized into air pressure values ​​and inflation intervals according to two dimensions: age group and sleep stage.

[0051] Specifically, the system first reads the user's age information; then it matches the user's age information with the pre-divided age groups in the age-based sleep intensity matching database; next, it retrieves the age group parameter range that matches the user's age information; finally, it loads the age group parameter range into the running memory of the main control voltage regulation execution module for use by downstream steps.

[0052] It should be noted that the age-specific sleep intensity matching database includes multiple age groups. Each age group corresponds to the air pressure values ​​and air intervals associated with four sleep stages: wakefulness, light sleep, deep sleep, and REM sleep. The air pressure values ​​are in kPa (kilopascals), and the air intervals are in seconds (s). Within any age group, the air pressure value associated with deep sleep is greater than that associated with light sleep, and the air pressure value associated with light sleep is greater than that associated with wakefulness. Within any age group, the air interval associated with deep sleep is less than that associated with light sleep, and the air interval associated with light sleep is less than that associated with wakefulness. Within any age group, the air pressure value associated with REM sleep is no higher than that associated with light sleep, and the air interval associated with REM sleep is no less than that associated with light sleep.

[0053] In a preferred embodiment, the age-specific sleep intensity matching database is divided into five age groups based on age differences in the lower limb muscle and fascia endurance of users. Specific parameter configurations are as follows: Figure 3 As shown, the detailed parameters for each age group are as follows.

[0054] The first age group, aged 3 to 6 years, has an associated air pressure of 2 kPa and an inflation interval of 120 s for wakefulness, 3 kPa and 90 s for light sleep, 5 kPa and 45 s for deep sleep, and 3 kPa and 90 s for REM sleep. The second age group, aged 7 to 9 years, has an associated air pressure of 2.5 kPa and an inflation interval of 120 s for wakefulness, 4 kPa and 90 s for light sleep, 6 kPa and 45 s for deep sleep, and 4 kPa and 90 s for REM sleep. The third age group, aged 10 to 12 years, has an associated air pressure of 3 kPa and an inflation interval of 100 s for wakefulness. The four age groups are: 1. Light sleep (5 kPa, 80 s intervals), 2. Deep sleep (7 kPa, 40 s intervals), and 3.5 kPa (80 s intervals). The fourth age group (13-15 years) has the following associated air pressure values: 3.5 kPa (100 s intervals), 6 kPa (80 s intervals), 8 kPa (40 s intervals), and 6 kPa (80 s intervals). The fifth age group (16-18 years) has the following associated air pressure values: 4 kPa (90 s intervals), 7 kPa (70 s intervals), and 1.5 kPa (80 s intervals). The pressure value associated with REM sleep is 7 kPa and the air-inflation interval is 30 s.

[0055] In another implementation, the air pressure value associated with REM sleep can also be lower than that of light sleep. For example, for the third age group, the air pressure value associated with REM sleep can be 4.5 kPa (5 kPa lower than the light sleep air pressure value of this age group), and the air-inflation interval can be 90 s (greater than the light sleep air-inflation interval of this age group, which is 80 s). This allows for a lower air pressure amplitude and a longer pressure regulation interval to match the dream limb movement characteristics of this stage during REM sleep. As long as the constraint that "REM sleep air pressure value is not higher than light sleep air pressure value and air-inflation interval is not less than light sleep air-inflation interval" is met, it falls within the scope of the age-specific sleep intensity matching database of this application.

[0056] Optionally, those skilled in the art can iteratively optimize the above parameters based on subsequently collected usage sample data, for example, by dividing the age group into six or eight segments. As long as the aforementioned structural constraints of "deep sleep pressure is greater than light sleep pressure, which is greater than wakefulness pressure," "deep sleep interval is less than light sleep interval, which is less than wakefulness interval," and "rapid eye movement sleep pressure is not higher than light sleep pressure, and the air-inflation interval is not less than the light sleep air-inflation interval" are met, they all fall within the scope of the age-specific sleep intensity matching database of this application. In this embodiment, the age-specific sleep intensity matching database is constructed using a pre-established static mapping table, which can complete the mapping retrieval without introducing a machine learning model, resulting in low implementation costs. In implementations where hardware resources permit, the mapping table can also be deployed in the form of a lookup function or a lightweight decision tree, which is not limited in this application.

[0057] S203. Based on standardized sleep stage determination values, the target air pressure value and target inflation interval associated with the current sleep stage are matched within the age group parameter range.

[0058] The target air pressure value refers to the steady-state air pressure amplitude that the lower limb stretching airbag should reach in the current sleep stage; the target inflation interval period refers to the time interval benchmark between two adjacent segmented pressure adjustments in the current sleep stage.

[0059] Specifically, the system first reads the standardized sleep stage determination value S; then performs a lookup operation in the loaded age group parameter range using S as the index; next, it reads the target air pressure value and the target inflation interval from the lookup results; finally, it outputs the target air pressure value and the target inflation interval to the pressure regulation drive subunit of the main control pressure regulation execution module.

[0060] S204. Based on the target air pressure value and the target inflation interval, the lower limb stretching airbag is driven to perform segmented pulse pressure regulation through the main control pressure regulation execution module.

[0061] Among them, segmented pulse pressure regulation refers to a pressure regulation method in which the main control pressure regulation execution module drives the lower limb stretching airbag pressure to change segment by segment with a preset segment step size as the minimum pressure regulation unit and the target inflation interval as the segment interval until the target pressure value is reached.

[0062] Specifically, the system first uses a pressure feedback sensor to collect the actual air pressure of the lower limb stretching airbag; then it compares the actual air pressure with the target air pressure value; next, when the actual air pressure is higher than the target air pressure value, it depressurizes in stages with a preset step size to the target air pressure value; when the actual air pressure is lower than the target air pressure value, it inflates in stages with a preset step size to the target air pressure value; finally, it maintains the target inflation interval period between two adjacent stages of inflation or depressurization before executing the next stage of pressure adjustment.

[0063] It should be noted that the preset segment step size in this embodiment is preferably 0.25 kPa or 0.5 kPa. Those skilled in the art can adjust the specific step size value according to the airway response speed and the characteristics of the airbag material, and this application does not limit this. The target inflation interval period serves two functions: first, as the time interval between two adjacent segment inflations or deflations, ensuring that the pressure change occurs in a slow, stepwise manner; second, as the steady-state holding time after the completion of a complete pressure regulation cycle. Thus, the same target inflation interval period parameter controls both the rhythm of pressure amplitude change and the pressure regulation response frequency during the sleep phase, achieving coupled control of two physical quantities by a single parameter.

[0064] Please see Figure 4 , Figure 4 A schematic diagram showing the time sequence comparison of the stretching air pressure of four types of sleep cycle airbags provided in the embodiments of this application. Figure 4 Taking one age group as an example, with time as the horizontal axis and airbag stretching pressure as the vertical axis, and a preset segment step size of 0.25 kPa, the dynamic trajectory of the lower limb stretching airbag pressure under segmented pulse pressure regulation is visually shown as the user progresses from the awake stage to the light sleep stage to the deep sleep stage to the REM sleep stage. As can be seen, during the waking period, only a low protective air pressure (low tension) is maintained; after entering the light sleep period, the target air pressure value rises to the medium tension level, while the target inflation interval takes a relatively large value, so that the air pressure curve shows a slow step-like upward shape; after entering the deep sleep period, the target air pressure value rises to the high safe micro-tension level, and the target inflation interval takes a relatively small value, so that the air pressure curve shows a more compact step-like upward shape, thus outputting a stretching force with a more compact rhythm that matches the sparse body movement characteristics of the deep sleep stage; after entering the REM sleep period, the target air pressure value and the target inflation interval drop back to values ​​that are no higher than the air pressure value of the light sleep period and no less than the inflation interval of the light sleep period, in order to reduce the risk of decreased comfort caused by dream-related limb movements.

[0065] After completing a voltage regulation closed loop from S201 to S204, the system returns to S201 according to the preset synchronization cycle to continue the next round of data synchronization, anomaly filtering, fusion correction, parameter matching and segmented pulse voltage regulation.

[0066] Furthermore, based on completing the main processes S201 to S204 mentioned above, the method can also deploy three independent security protection logics, corresponding to user active awakening scenarios, system parameter over-limit scenarios, and excessive exposure scenarios throughout the night.

[0067] Specifically, the main control pressure regulation module monitors the standardized sleep stage judgment value in real time. When the duration of continuous wakefulness (S=0) in the standardized sleep stage judgment value reaches the preset wakefulness threshold, the lower limb stretching airbag is depressurized to the preset minimum protective air pressure, and the system enters a shutdown standby state. In this embodiment, the preset wakefulness threshold is preferably 10 minutes; the preset minimum protective air pressure is preferably a pressure no higher than the wakefulness air pressure value. The above shutdown action depressurizes first and then enters standby to avoid sudden changes in the airway caused by a one-time depressurization.

[0068] Before executing segmented pulse pressure regulation, the main control pressure regulation execution module arbitrates the target air pressure value—a hardware air pressure upper limit threshold is pre-set. When the target air pressure value exceeds the hardware air pressure upper limit threshold, the hardware air pressure upper limit threshold is used instead of the target air pressure value for segmented pulse pressure regulation. In this embodiment, the hardware air pressure upper limit threshold is preferably 10 kPa, meaning that the maximum target air pressure value for any age group does not exceed 10 kPa. The arbitration function of the hardware air pressure upper limit threshold differs from the parameter upper limit constraint during the age-specific sleep intensity matching database construction phase—the former acts on every pressure regulation drive command of the main control pressure regulation execution module, independent of the configuration correctness of the age-specific sleep intensity matching database. It can effectively cope with runtime abnormal scenarios such as incorrect parameters being issued to the age-specific sleep intensity matching database through remote upgrades, incorrect entry of user age information, and the target air pressure value exceeding the safe range due to numerical overflow during calculation. Together with the parameter upper limit constraint during the database construction phase, it forms a dual protection at the algorithm and hardware levels.

[0069] The main control pressure regulation execution module accumulates the running time after the segmented pulse pressure regulation is started. When the accumulated running time reaches the preset maximum running threshold, the lower limb stretching airbag is depressurized to the preset minimum protective air pressure and enters the shutdown standby state. In this embodiment, the preset maximum running threshold is preferably 8 hours to avoid the user's discomfort caused by excessive cumulative stretching due to forgetting to turn off the device.

[0070] It is understandable that the above three types of security protection logic are independent of each other, and one can be deployed or all of them can be superimposed.

[0071] Please refer to it again. Figure 1 Another embodiment of this application provides a stretching force adaptive adjustment system based on sleep data recognition. The system includes a sleep physiological data acquisition device, a data transmission module, a sleep data cleaning, correction and fusion module, an age-specific sleep force matching database, a main control pressure adjustment execution module, a lower limb stretching airbag, and a local body motion sensing backup module.

[0072] A sleep physiological data acquisition device is configured to collect a user's sleep physiological parameters and output a sleep data package, which includes an original sleep stage label. In this embodiment, the sleep physiological data acquisition device can be a mature wearable product such as a smartwatch or smart bracelet that supports sleep monitoring functions. It has built-in hardware such as a heart rate photoelectric sensor and an accelerometer to collect sleep physiological parameters such as heart rate, heart rate variability, and number of times the user turns over, and outputs an original sleep stage label. This application does not limit the specific brand and model of the wearable device.

[0073] The data transmission module is configured to establish a data transmission link between the sleep physiological data acquisition device and the sleep data cleaning, correction, and fusion module, and to transmit sleep data packets to the sleep data cleaning, correction, and fusion module via this data transmission link. The data transmission link can be the aforementioned local direct connection channel based on Bluetooth Low Energy, or a wireless network channel based on a cloud server interface, or a combination of both.

[0074] A local motion sensing backup module is configured to collect motion data output from the local sensor of the stretching device and output the motion data to the sleep data cleaning, correction and fusion module. The local sensor of the stretching device can be selected from conventional motion sensing elements such as accelerometers or piezoelectric vibration sensors; this application does not limit its selection.

[0075] The sleep data cleaning, correction, and fusion module is configured to correct the original sleep stage labels in the sleep data package based on weighted fusion of body movement data to obtain standardized sleep stage determination values. In this embodiment, the sleep data cleaning, correction, and fusion module can also undertake the aforementioned derivative functions such as abnormal data filtering, abnormal counting, and dual-path fault-tolerant switching. The specific execution process is consistent with the weighted fusion correction and abnormal filtering described in S201 above, and will not be repeated here.

[0076] The age-specific sleep intensity matching database is configured to store age-specific sleep-stretching intensity matching parameters. Each age group in the age-specific sleep-stretching intensity matching parameters is configured with the air pressure value and inflation interval cycle associated with the four sleep stages of wakefulness, light sleep, deep sleep and REM sleep. The specific parameter organization format is consistent with the age-specific parameter table described in S202 above.

[0077] The main control pressure regulation module is configured to retrieve a matching age group parameter range from an age-specific sleep intensity matching database based on the user's age information. Based on standardized sleep stage determination values, it matches the target air pressure value and target inflation interval associated with the current sleep stage within the age group parameter range. Then, it drives the lower limb stretching airbag to perform segmented pulse pressure regulation according to the target air pressure value and target inflation interval. Optionally, the main control pressure regulation module can also perform the aforementioned safety protection functions such as awake state termination protection, air pressure upper limit hard lock, and excessively long operation protection.

[0078] The lower limb stretching airbag is configured to, under the drive of the main control pressure regulating execution module, depressurize in stages to the target pressure value when the actual air pressure is higher than the target air pressure value, and inflate in stages to the target air pressure value when the actual air pressure is lower than the target air pressure value, thereby obtaining a stretching force adapted to the standardized sleep stage judgment value. The lower limb stretching airbag can be a single-chamber airbag or a multi-chamber independently controlled pressure airbag; this application does not limit this.

[0079] The modules mentioned above are connected in series via data streams, and the data stream interaction logic between them is as follows: Figure 1 As shown, the specific execution process is consistent with the aforementioned S201 to S204 processes, and will not be repeated here.

[0080] In this embodiment, due to the adoption of a combined technical solution of cross-device sleep data acquisition and local body movement data weighted fusion correction, age-specific sleep intensity matching database retrieval based on structured inequality constraints, and segmented pulse pressure regulation with the synchronous reuse of the target inflation interval cycle as the segmented pressure regulation time interval, the stretching intensity can undergo a two-dimensional differentiated smooth change according to the user's real-time sleep stage and age group, and the air pressure change rate is also adjusted in conjunction with the sleep stage; effectively solving the problems of deviation between the operating state of existing stretching solutions and the user's actual physical state, and the need to improve user experience and compliance; thus achieving improved sleep stage recognition accuracy, reduced nighttime awakening rate, and differentiated adaptation between stretching intensity and the user's actual nighttime physical state.

Claims

1. A stretch force adaptive adjustment method based on sleep data recognition, characterized in that, The method includes: Acquire sleep data packets output by the sleep physiological data acquisition device, collect body movement data output by the local sensor of the stretching device, and correct the original sleep stage labels in the sleep data packets based on the weighted fusion of the body movement data to obtain standardized sleep stage determination values. Based on the user's age information, the matching age group parameter range is retrieved from the pre-built age-specific sleep intensity matching database. Each age group in the age-specific sleep intensity matching database is configured with air pressure values ​​and inflation intervals associated with four sleep stages: wakefulness, light sleep, deep sleep, and REM sleep. Based on the standardized sleep stage determination value, the target air pressure value and target inflation interval are matched within the age group parameter range to obtain the current sleep stage. Based on the target air pressure value and the target inflation interval, the lower limb stretching airbag is driven by the main control pressure regulation execution module to perform segmented pulse pressure regulation. When the actual air pressure of the lower limb stretching airbag is higher than the target air pressure value, it is depressurized in segments to the target air pressure value. When the actual air pressure of the lower limb stretching airbag is lower than the target air pressure value, it is inflated in segments to the target air pressure value, thereby obtaining a stretching force that matches the standardized sleep stage determination value.

2. The method of claim 1, wherein, The sleep data package includes heart rate, heart rate variability, number of times the patient turns over per unit of time, and original sleep stage labels.

3. The method according to claim 1, characterized in that, The step of correcting the original sleep stage labels in the sleep data packet based on the weighted fusion of the body movement data to obtain standardized sleep stage determination values ​​includes: Extract limb vibration amplitude and continuous motion identifiers based on the limb vibration amplitude from the body motion data; Using the limb vibration amplitude and the continuous movement identifier as correction factors, the original sleep stage labels in the sleep data package are corrected according to a preset weighting coefficient to obtain the standardized sleep stage determination value. The standardized sleep stage determination value is taken as the label value corresponding to wakefulness, light sleep, deep sleep or REM sleep.

4. The method according to claim 3, characterized in that, Obtaining the standardized sleep stage determination value specifically includes: The result obtained by correcting the original sleep stage label according to the preset weighting coefficient is used as the intermediate judgment value; Read the number of times the patient turns over within a unit of time in the sleep data packet; When the number of times the body turns over is not higher than the first turning over threshold, the intermediate judgment value is determined as the standardized sleep stage judgment value; When the number of times the body turns over is higher than the first turning over threshold but not higher than the second turning over threshold, if the intermediate judgment value indicates deep sleep, then the intermediate judgment value is lowered to the identification value corresponding to light sleep and determined as the standardized sleep stage judgment value. When the number of times the patient turns over is higher than the second turning over threshold and the intermediate judgment value indicates light sleep or deep sleep, the intermediate judgment value is adjusted up to the identifier value corresponding to wakefulness and then determined as the standardized sleep stage judgment value.

5. The method according to claim 3, characterized in that, Prior to the weighted fusion correction, the method further includes: Abnormal data points in the sleep data package are filtered out. The abnormal data points include signal loss data points and jump data points where the heart rate exceeds the preset normal physiological range. The abnormal data points that are continuously filtered are accumulated to obtain an abnormal count; When the abnormal count reaches a preset threshold, it is determined that the sleep physiological data acquisition device is currently in a disconnected or unavailable state, and it switches to the local recognition mode based on the body movement data. The local recognition mode outputs the backup sleep stage determination value. The standby sleep stage determination value is taken as the identifier value corresponding to wakefulness, light sleep, deep sleep or REM sleep, and the standby sleep stage determination value replaces the standardized sleep stage determination value to continue to perform the matching of the target air pressure value and the target inflation interval cycle and the segmented pulse pressure regulation.

6. The method according to claim 1, characterized in that, The age-specific sleep intensity matching database includes multiple age groups, each of which corresponds to the air pressure value and inflation interval cycle associated with four sleep stages: wakefulness, light sleep, deep sleep, and REM sleep. In any of the age groups, the barometric pressure associated with deep sleep is greater than that associated with light sleep, which in turn is greater than that associated with wakefulness. In any of the age groups, the inter-sleep intervals associated with deep sleep are shorter than those associated with light sleep, which in turn are shorter than those associated with wakefulness. In any of the age groups, the air pressure associated with REM sleep is not higher than that associated with light sleep, and the air intermittent period associated with REM sleep is not less than that associated with light sleep.

7. The method according to claim 1, characterized in that, In the segmented pulse pressure regulation, the target inflation interval period is maintained between two adjacent segmented inflation or deflation.

8. The method according to claim 1, characterized in that, The method further includes at least one of the following: When the duration of continuous wakefulness as determined by the standardized sleep stage judgment value reaches the preset wakefulness threshold, the lower limb stretching airbag is depressurized to the preset minimum protective air pressure and enters a shutdown standby state. When the cumulative running time after the segmented pulse pressure regulation start reaches the preset maximum running threshold, the lower limb stretching airbag is depressurized to the preset minimum protective air pressure and enters the shutdown standby state.

9. The method according to claim 1, characterized in that, The method further includes: A hardware pressure upper limit threshold is preset. When the target pressure value exceeds the hardware pressure upper limit threshold, the segmented pulse pressure regulation is performed instead of the target pressure value by using the hardware pressure upper limit threshold.

10. A stretching force adaptive adjustment system based on sleep data recognition, characterized in that, The system includes a sleep physiological data acquisition device, a data transmission module, a sleep data cleaning, correction and fusion module, an age-based sleep intensity matching database, a main control pressure regulation execution module, a lower limb stretching airbag, and a local body motion sensing backup module. The sleep physiological data acquisition device is configured to collect the user's sleep physiological parameters and output a sleep data package, the sleep data package containing original sleep stage labels; The data transmission module is configured to establish a data transmission link between the sleep physiological data acquisition device and the sleep data cleaning, correction and fusion module, and to transmit the sleep data packet to the sleep data cleaning, correction and fusion module via the data transmission link; The local motion sensing backup module is configured to collect motion data output by the local sensor of the stretching device and output the motion data to the sleep data cleaning, correction and fusion module; The sleep data cleaning, correction and fusion module is configured to correct the original sleep stage labels in the sleep data package based on the weighted fusion of the body movement data to obtain standardized sleep stage determination values. The age-specific sleep intensity matching database is configured to store age-specific sleep-stretching intensity matching parameters. In the age-specific sleep-stretching intensity matching parameters, each age group is configured with the air pressure value and inflation interval cycle associated with four sleep stages: wakefulness, light sleep, deep sleep, and REM sleep. The main control pressure regulation execution module is configured to retrieve the matching age group parameter range from the age-specific sleep intensity matching database according to the user's age information, match the target air pressure value and target inflation interval period associated with the current sleep stage in the age group parameter range based on the standardized sleep stage judgment value, and drive the lower limb stretching airbag to perform segmented pulse pressure regulation according to the target air pressure value and target inflation interval period. The lower limb stretching airbag is configured to, under the drive of the main control pressure regulating execution module, depressurize in stages to the target pressure value when the actual air pressure is higher than the target air pressure value, and inflate in stages to the target air pressure value when the actual air pressure is lower than the target air pressure value, so as to obtain a stretching force that matches the standardized sleep stage determination value.

Images