A sitting posture quantification correction guidance method and device based on multi-source seat surface perception

Abstract

The application discloses a sitting posture quantification correction guidance method and device based on multi-source seat surface perception, comprising: acquiring multi-source seat surface perception data of a target cushion; fusing the multi-source seat surface perception data, performing user seat state recognition, determining an effective sitting period corresponding to a sitting state after recognizing that the user is in the sitting state; intercepting the multi-source seat surface perception data in the effective sitting period to obtain a period analysis data set; performing time sequence statistics and sitting posture feature extraction processing on the period analysis data set to generate sitting posture quantification result data; and generating target sitting posture correction guidance information matched with the sitting posture quantification result data based on a preset mapping rule and in combination with a user's long sitting health behavior image. The application can fuse multi-source perception data, improve the accuracy of sitting state determination, realize the scene of sitting posture correction guidance, meet the demand of fine health management, and can be widely applied to the technical field of intelligent health monitoring.

Description

Technical Field

[0001] This invention relates to the field of intelligent health monitoring technology, and in particular to a method and device for quantitative correction guidance of sitting posture based on multi-source seat surface sensing. Background Technology

[0002] With the popularization of smart office and occupational health management concepts, monitoring and intervening in the sedentary behavior of office workers using sensor technology has become a research hotspot. By integrating sensing units into office chairs to collect human posture data in real time and analyzing it using algorithm models, the aim is to prevent various occupational diseases caused by prolonged poor posture and improve office health.

[0003] Existing sedentary monitoring solutions mainly include software-based timed reminders, activity monitoring based on wearable devices (such as smart bracelets), and seated detection based on simple pressure pads. These solutions remind users to get up and move around or record the duration of sedentary time by setting fixed time interval pop-ups, detecting limb activity, or judging whether there is pressure on the seat surface.

[0004] However, existing solutions generally suffer from insufficient monitoring accuracy, failing to accurately identify the user's actual sitting status, and their functions are relatively limited, only able to perform simple statistics on sedentary time, making it difficult to meet the needs of refined health management. Summary of the Invention

[0005] This invention provides a method and device for quantitative correction guidance of sitting posture based on multi-source seat surface sensing, which solves the problems of insufficient accuracy and limited functionality of existing sitting posture monitoring.

[0006] In a first aspect, the present invention provides a method for quantitative correction guidance of sitting posture based on multi-source seat surface perception, comprising: Acquire multi-source seat surface perception data of the target seat cushion; By integrating the multi-source seat surface perception data, user seat status recognition is performed. After recognizing that the user is in a seated state, the effective seating period corresponding to the seated state is determined. Multi-source seat surface sensing data within the effective seating cycle are extracted to obtain a cycle analysis dataset; The periodic analysis dataset is subjected to time-series statistics and sitting posture feature extraction processing to generate sitting posture quantification result data; Based on preset mapping rules and combined with the user's sedentary health behavior profile, target posture correction guidance information is generated by matching the quantitative results of the sitting posture.

[0007] In some embodiments, fusing the multi-source seat surface perception data to perform user seat state recognition, and determining the effective seating period corresponding to the seated state after recognizing that the user is in a seated state, includes: By fusing the multi-source seat surface perception data, the user's seat status is identified based on the fusion result; If the user's seat status is "sitting down", then record the start time of the "sitting down" status. If the user's seat status changes from the seated state to the unseated state, the corresponding state change time is recorded, and the effective seated period is determined based on the state start time and the state change time.

[0008] In some embodiments, fusing the multi-source seat surface perception data and identifying the user's seat status based on the fusion result includes: Extract the total seat pressure, seat temperature, and seat pressure distribution features from the multi-source seat sensing data, and perform multi-dimensional joint fusion judgment; If the total pressure on the seat surface is greater than the first pressure threshold, and the seat surface temperature is higher than the human body temperature threshold, and the seat surface pressure distribution meets the human body sitting pressure distribution characteristics, then the user's seat state is determined to be a sitting state. If the total pressure on the seat surface is less than the second pressure threshold and the duration exceeds the preset duration, or if the seat surface temperature is lower than the ambient temperature threshold, then the user's seat status is determined to be an unseat state.

[0009] In some embodiments, the step of performing time-series statistics and posture feature extraction processing on the periodic analysis dataset to generate posture quantification result data includes: Perform time-series continuity statistics on the periodic analysis dataset to calculate the cumulative sitting time within the effective sitting period; Extract the periodic pressure distribution characteristics from the periodic analysis dataset; The periodic pressure distribution characteristics are input into a pre-constructed sitting posture quantitative assessment model. After sitting posture classification and quantitative scoring, the target sitting posture type and sitting posture health score are output. The cumulative sitting time, the target sitting posture type, and the sitting posture health score are integrated to generate quantitative sitting posture result data.

[0010] In some embodiments, the construction of the user's sedentary health behavior profile includes: Obtain user health behavior dataset; Extract historical intervention behavior data, historical authorized device feedback data, and historical seating cycle record data from the user health behavior dataset; Multi-dimensional feature statistics are performed on the historical intervention behavior data, the historical authorized device feedback data, and the historical sitting cycle record data to determine user intervention preference characteristics and sitting posture behavior characteristics. Time-series statistical analysis was performed on the user health behavior dataset to determine the distribution characteristics of sedentary time periods; Based on the user intervention preference features and the sitting posture behavior features, combined with the sedentary time period distribution features, multi-dimensional feature integration and attribute label mapping processing are performed to construct a user sedentary health behavior profile.

[0011] In some embodiments, generating target posture correction guidance information based on preset mapping rules and combined with a user's sedentary health behavior profile, and matching the quantitative results of the sitting posture data, includes: Extract intervention preference features, physical characteristics, and sedentary time distribution features from the user's sedentary health behavior profile; Extract the cumulative sitting time and target sitting posture type from the sitting posture quantification results data; According to the preset mapping rules, the posture adjustment scheme based on the cumulative sitting time and the targeted posture correction scheme based on the target sitting posture type are matched; The posture adjustment scheme and the targeted posture correction scheme are combined to generate preliminary sitting posture correction guidance information; Based on the intervention preference characteristics, body posture characteristics, and sedentary time distribution characteristics, the preliminary posture correction guidance information is filtered and the layout of the plan is optimized to generate the corresponding target posture correction guidance information.

[0012] In some embodiments, it also includes: Obtain status data from external authorized devices and schedule data from the office system; The current work scenario is identified by matching the status data of the external authorized devices and the schedule data of the office system with scene features. Extract the cumulative sitting time from the sitting posture quantification result data, and match the cumulative sitting time with the target duration threshold range; A target-level early warning strategy that matches the target duration threshold range; Based on the current working scenario, the target hierarchical early warning strategy is subjected to threshold calibration and modal combination matching processing to generate multimodal reminder execution instructions.

[0013] In some embodiments, it also includes: Acquire user intervention behavior data in response to the multimodal reminder execution command; After binding the periodic analysis dataset, the posture quantification result data, the multimodal reminder execution command, and the intervention behavior data, the sitting period record data corresponding to the effective sitting period is generated. The periodic record data is stored in the user's health behavior dataset; At preset time intervals, a health analysis model is used to perform health analysis on the user's health behavior dataset, generating a user periodic health analysis report.

[0014] In some embodiments, it also includes: Extract the user activity status data returned by the external authorized device from the intervention behavior data, corresponding to the multimodal reminder execution command; According to the preset intervention effectiveness determination rules, the effectiveness of the intervention response to the user activity status data is verified, an intervention response verification result is generated and bound to the periodic record data; If the response status in the intervention response verification result is "valid standing up", then the cumulative sedentary time is reset.

[0015] Secondly, the present invention provides a posture quantification and correction guidance device based on multi-source seat surface sensing, comprising: The data acquisition module is used to acquire multi-source seat surface sensing data of the target seat cushion; The seating determination module is used to integrate the multi-source seat surface perception data, perform user seat status recognition, and determine the effective seating period corresponding to the seating status after recognizing that the user is in a seated state. The periodic interception module is used to intercept the multi-source seat surface sensing data within the effective seating period to obtain a periodic analysis dataset. The sitting posture quantification module is used to perform time-series statistics and sitting posture feature extraction processing on the periodic analysis dataset to generate sitting posture quantification result data. The correction matching module is used to generate target sitting posture correction guidance information that matches the quantitative sitting posture result data based on preset mapping rules and combined with the user's sedentary health behavior profile.

[0016] As can be seen from the above technical solutions, the present invention has the following advantages: The present invention provides a sitting posture quantitative correction guidance method and device based on multi-source seat surface perception, which, by acquiring multi-source seat surface perception data of the target seat cushion, overcomes the limitations of single data acquisition and provides a comprehensive and reliable data foundation for accurate sitting judgment and sitting posture analysis; by fusing multi-source seat surface perception data, it identifies the sitting state and determines the effective sitting cycle, accurately dividing the effective sitting cycle and avoiding the defects of large errors in sitting and leaving judgment and inaccurate timing in traditional methods; by extracting multi-source seat surface perception data within the effective sitting cycle... The system obtains a periodic analysis dataset and removes invalid time-series data interference. By performing time-series statistics and extracting sitting posture features from the periodic analysis dataset, it generates quantitative sitting posture results data, realizing the quantification of sitting posture quality. This overcomes the shortcomings of traditional devices that can only simply time but cannot quantify sitting posture quality. By combining the user's sedentary health behavior profile, it generates target sitting posture correction guidance information that matches the quantitative sitting posture results data. This realizes the scenario-based application of sitting posture correction guidance, breaks the traditional fixed and uniform guidance model, and greatly improves the adaptability, practicality, and implementation of sitting posture correction guidance, meeting the needs of refined health management. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart of a sitting posture quantitative correction guidance method based on multi-source seat surface perception provided by the present invention; Figure 2 A schematic diagram of the structure of a posture quantification and correction guidance device based on multi-source seat surface sensing provided by the present invention; Figure 3 This is a schematic diagram of the hardware structure of an electronic device provided by the present invention. Detailed Implementation

[0019] This invention provides a method and device for quantitative correction guidance of sitting posture based on multi-source seat surface sensing, which solves the technical problems of insufficient accuracy and limited functionality of existing sitting posture monitoring.

[0020] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0021] Please see Figure 1 , Figure 1 An optional flowchart of a sitting posture quantitative correction guidance method based on multi-source seat surface perception provided in an embodiment of the present invention includes steps 101 to 105.

[0022] Step 101: Obtain multi-source seat surface perception data of the target seat cushion; Multi-source seat surface sensing data refers to a collection of multi-dimensional sensing data related to the seat surface contact state, collected by multiple types of sensing units deployed inside the seat cushion. Example data includes total seat surface pressure, seat surface temperature, and seat surface pressure distribution.

[0023] A flexible piezoresistive sensor array refers to a matrix of piezoresistive sensing units arranged on a flexible substrate. The resistance value of the sensing unit changes accordingly with the magnitude of the applied pressure, and the pressure value at each point can be obtained by resistance conversion.

[0024] Optionally, the process for acquiring multi-source seat surface sensing data in step 101 is as follows: First, the target seat cushion has multiple types of sensing units built in. Each sensing unit synchronously collects the original sensing signals according to a preset sampling rate. Among them, the flexible piezoresistive sensor array collects the original pressure signals at each point at a sampling rate of 10Hz, and the multiple temperature sensors collect the temperature signals at different positions on the seat surface according to the corresponding sampling rate. Subsequently, the original pressure values ​​at all points of the flexible piezoresistive sensor array are summed to obtain the total pressure on the seat surface. At the same time, the pressure values ​​at each point and their spatial arrangement are retained to generate seat surface pressure distribution data. Finally, the various types of sensor data, such as total seat pressure, seat temperature, and seat pressure distribution, are aligned according to the acquisition timestamp and packaged to generate multi-source seat sensing data for the corresponding time.

[0025] Step 102: Integrate multi-source seat surface perception data, perform user seat status recognition, and after recognizing that the user is in a seated state, determine the effective seated period corresponding to the seated state. The effective sitting period refers to the complete time interval corresponding to a user's single continuous sitting state (such as the start and end duration of a single sitting session). Within the period, the user's slight adjustment of sitting posture will not interrupt the continuity of the period.

[0026] The pressure distribution characteristics of human body when sitting refer to the typical spatial distribution pattern of seat surface pressure when a human body sits down normally (such as symmetrical double support areas on the left and right, and high pressure concentration areas at the ischial support position), which is used to distinguish between human body sitting and heavy object placement scenarios.

[0027] The pressure judgment threshold includes a first pressure threshold for determining whether a person is seated and a second pressure threshold for determining whether a person is alighting, which are the critical pressure values ​​corresponding to the switching between the seated and alighting states, respectively.

[0028] Optionally, step 102 uses multi-source seat surface sensing data as input to perform seat status recognition and period determination, as follows: First, the multi-source seat surface sensing data is fused and processed to identify the user's current seat status: when the total seat surface pressure is greater than the first pressure threshold, the seat surface temperature is higher than the human body temperature threshold, and the seat surface pressure distribution meets the characteristics of human body sitting pressure distribution, the user's seat status is determined to be a seated state; when the total seat surface pressure is less than the second pressure threshold and the duration exceeds the preset duration, or the seat surface temperature is lower than the ambient temperature threshold, the user's seat status is determined to be an unseated state. If the user is detected to be in a seated state for the first time, the moment is recorded as the start time of the seated state. During the duration of the seated state, pressure distribution fluctuations caused by the user's slight adjustments to their posture do not change the seat state determination and do not interrupt the cycle timer. If the user's seat status is detected to change from seated to unseated, the moment of this change is recorded as the corresponding state change moment. The time interval between the state start moment and the state change moment is the effective seat period corresponding to this seating.

[0029] In some embodiments, step 102 includes the following sub-steps: By fusing multi-source seat surface perception data, the user's seat status is identified based on the fusion result; If the user's seat status is "sitting down", then record the start time of the sitting down status. If a user's seat status changes from seated to unseated, the corresponding state change time is recorded, and the valid seated period is determined based on the state start time and state change time.

[0030] Optionally, the above-mentioned fusion of multi-source seat surface perception data, based on the fusion result, identifies the user's seat status, including: Extract the total seat pressure, seat temperature, and seat pressure distribution features from multi-source seat sensing data, and perform multi-dimensional joint fusion judgment; If the total pressure on the seat surface is greater than the first pressure threshold, and the seat surface temperature is higher than the human body temperature threshold, and the seat surface pressure distribution meets the characteristics of human body sitting pressure distribution, then the user's seat state is determined to be a sitting state. If the total pressure on the seat is less than the second pressure threshold and the duration exceeds the preset duration, or if the seat temperature is lower than the ambient temperature threshold, the user's seat status will be determined as an unseated state.

[0031] For example, assuming the first pressure threshold is 100N, the second pressure threshold is 50N, the human body temperature threshold is 30℃, the ambient temperature threshold is 25℃, and the preset duration for determining low pressure is 5 seconds.

[0032] At 09:00:00 on the same day, multi-source seat surface sensing data was acquired. Among them, the total seat surface pressure was 652N, the seat surface temperature was 32.5℃, and the seat surface pressure distribution showed a double-support high-pressure zone with left and right symmetry. All the conditions for sitting were met, and the user was determined to have entered the sitting state. This moment was recorded as the state start moment, and the cycle timer was started. At 09:15:30 on the same day, the seat pressure distribution fluctuated briefly, but the total seat pressure was still higher than the first pressure threshold. It was determined that the user was fine-tuning his sitting posture without changing the seat status, and the periodic timing remained continuous. At 10:30:00 on the same day, the total pressure on the seat surface was collected as 5N, and this value lasted for more than 5 seconds, which met the conditions for leaving the seat. The user was determined to have switched to the state of leaving the seat, and this moment was recorded as the state switching moment. Starting from 09:00:00 and ending at 10:30:00, the effective sitting cycle duration was determined to be 90 minutes, and the corresponding sitting session numbered Session_001 was stored in the dataset.

[0033] Step 103: Extract multi-source seat surface sensing data within the effective seating cycle to obtain the cycle analysis dataset; The periodic analysis dataset refers to the collection of sensor data (such as time-aligned pressure and temperature sequence data sets) that can be directly used for sitting posture analysis after standardization and preprocessing within the effective sitting period. It serves as a unified input data source for subsequent quantitative analysis of sitting posture.

[0034] Specifically, step 103 includes the following sub-steps: Extract the start time of the valid seating cycle; Multi-source seat-surface sensing data between the start time and the current analysis time are extracted to obtain a preliminary analysis dataset; The preliminary analysis dataset is time-aligned and outlier filtered to generate a periodic analysis dataset.

[0035] Time alignment refers to matching multiple sensor data with different sampling rates to a uniform time granularity, thereby eliminating timing deviations caused by differences in sampling frequencies among multiple sensors.

[0036] Outlier filtering refers to removing abrupt changes in sensor data caused by transient interference (such as sudden pressure increases caused by instantaneous impacts from external forces) to avoid abnormal data interfering with the accuracy of subsequent analysis.

[0037] Optionally, step 103 uses the effective seating cycle as the time basis to extract and preprocess the multi-source seat surface sensing data. The specific process is as follows: First, extract the state start time of the identified valid sitting cycle. Use this start time as the starting point of the data extraction time and the current analysis time as the ending point of the data extraction time. Extract all the original multi-source seat surface perception data within the two time intervals to obtain the preliminary analysis dataset. The current analysis time can correspond to the real-time intervention trigger time or the effective sitting cycle termination time after the user leaves the seat. Subsequently, the preliminary analysis dataset was preprocessed: first, the multi-sensor data such as pressure and temperature with different sampling rates were time-aligned according to a uniform time granularity to eliminate time sequence misalignment; then, transient jump outliers in the data were removed by threshold judgment to complete outlier filtering. Finally, the standardized time-series data obtained after preprocessing the initial analysis dataset is the periodic analysis dataset.

[0038] In some embodiments, for example, assuming the start time of the effective sitting period is 09:00:00, the current analysis time is taken as 09:30:00 when the first sedentary reminder is triggered.

[0039] The first step is to extract the start time of the effective seating cycle, 09:00:00, and extract all the original multi-source seat surface sensing data from 09:00:00 to 09:30:00. This includes 18,000 sets of pressure array data with a 10Hz sampling rate and 180 sets of temperature data with a 1Hz sampling rate. These data are then combined to obtain the preliminary analysis dataset. The second step is to preprocess the preliminary analysis dataset: the pressure data is averaged at a 1-second granularity and matched with the temperature data at a time stamp, which is set once per second, to complete time alignment; then, by using the preset reasonable pressure range threshold, one set of abnormal pressure jump points of 820N caused by instantaneous external force is removed to complete outlier filtering. The third step is to generate a periodic analysis dataset containing time-series data of aligned pressure distribution and seat temperature.

[0040] Step 104: Perform time-series statistics and posture feature extraction on the periodic analysis dataset to generate posture quantification result data; The quantitative results of sitting posture refer to the set of structured results output by sitting posture analysis within a single effective sitting period, including cumulative sitting time, sitting posture type, and sitting posture health score.

[0041] In some embodiments, step 104 includes the following sub-steps: Perform time-series continuity statistics on the periodic analysis dataset to calculate the cumulative sitting time within the effective sitting period; Extract the cyclic pressure distribution characteristics from the cyclic analysis dataset; The periodic pressure distribution characteristics are input into a pre-built quantitative assessment model of sitting posture. After sitting posture classification and quantitative scoring, the target sitting posture type and sitting posture health score are output. By integrating cumulative sitting time, target sitting posture type, and sitting posture health score, quantitative results of sitting posture are generated.

[0042] A sitting posture quantitative assessment model refers to a pre-built set of multi-dimensional sitting posture assessment rules (such as a decision model that includes sitting posture classification rules and weighted scoring rules) used to simultaneously output sitting posture type and health score based on pressure distribution characteristics.

[0043] Periodic pressure distribution characteristics refer to quantifiable distribution parameters (such as the left-right center of gravity ratio, pressure standard deviation, etc.) extracted from the pressure array data of the periodic analysis dataset, and are the core input indicators for assessing sitting posture quality.

[0044] Specifically, step 104 uses the periodic analysis dataset as input to complete the statistical processing of sedentary time and the quantitative processing of sitting posture quality. The specific process is as follows: First, time-series continuity statistics are performed on the periodic analysis dataset. Based on the continuous and valid seating sensor data within the period, the cumulative sitting time within the current valid seating period is calculated. Secondly, the pressure array data in the periodic analysis dataset is partitioned and statistically analyzed to extract core parameters such as the pressure ratio of the left and right partitions, the pressure ratio of the front and rear rows, the standard deviation of the pressure distribution, and the highest pressure point, thus forming the periodic pressure distribution characteristics. Subsequently, the periodic pressure distribution characteristics are input into the pre-built sitting posture quantitative assessment model. The target sitting posture type is first identified by matching the preset classification rules, and then the sitting posture health score is calculated by multi-dimensional weighted scoring rules. The two types of results are output simultaneously. Finally, the cumulative sitting time, target sitting posture type, and sitting posture health score are structured and encapsulated to generate quantitative sitting posture results data for the corresponding period, which are then fed into the subsequent intervention matching process.

[0045] In some embodiments, exemplarily, this embodiment takes the periodic analysis dataset corresponding to the 09:00:00-09:30:00 interval generated in step 103 and performs posture quantification processing, specifically including: The first step is to statistically analyze the temporal continuity of the periodic analysis dataset to confirm that there is no interruption in leaving the seat within the interval, and calculate the cumulative sitting time of the effective sitting cycle as 30 minutes. The second step is to extract the periodic pressure distribution characteristics: the 64-point pressure array is divided into three regions, left, middle and right, and the pressure is summed to obtain the following results: left side pressure accounts for 25%, center pressure accounts for 33%, and right side pressure accounts for 42%; the front row pressure accounts for 40% and the back row pressure accounts for 60%; the standard deviation of the pressure at all points is calculated to be 11.2N, and the highest pressure point is confirmed to be point M7 in the central region, with a value of 52N.

[0046] The third step is to input the above features into the sitting posture quantitative assessment model: After classification rule matching, the proportion of the right side is higher than that of the left side and the deviation is in the range of 10%-20%, and the target sitting posture type is determined to be mild right tilt; after weighted scoring calculation, the center of gravity is centered 20 points, the pressure uniformity is 30 points, the ischial tuberosity force matching degree is 20 points, and the front and back balance is 10 points, with a total sitting posture health score of 80 points, corresponding to a good level; The fourth step involves structurally integrating the cumulative sitting time of 30 minutes, the type of mild right-leaning posture, and the sitting health score of 80 points to generate quantitative sitting posture data for this effective sitting period.

[0047] Step 105: Based on preset mapping rules and combined with the user's sedentary health behavior profile, generate target sitting posture correction guidance information that matches the quantitative results of sitting posture.

[0048] Preset mapping rules refer to a set of standardized matching logics pre-stored locally in the system or in the cloud (such as using cumulative sitting time and poor sitting posture type as matching dimensions, and binding corresponding basic stretching and special correction action materials), which are used to output standardized correction solutions based on objective quantitative data of sitting posture.

[0049] User sedentary health behavior profiles refer to a standardized set of attribute tags formed by collecting users' work habits, physiological posture characteristics, preferences for reminders and stretching content, and historical sedentary time patterns (such as preference for short stretches of 1-3 minutes, high incidence of neck pain, and aversion to complex multi-set movements), providing a basis for personalized selection of corrective content.

[0050] The target posture correction guidance information refers to the personalized correction plan obtained after standardized rule matching and secondary screening based on user profile preferences. It includes integrated action content that adapts to the user's current sitting posture and fits the individual's physical symptoms and usage habits.

[0051] In some embodiments, the construction of a user's sedentary health behavior profile includes: Obtain user health behavior dataset; Extract historical intervention behavior data, historical authorized device feedback data, and historical seating cycle record data from the user health behavior dataset; Multi-dimensional feature statistics were performed on historical intervention behavior data, historical authorized device feedback data, and historical seating cycle record data to determine user intervention preference characteristics and sitting posture behavior characteristics. Perform time-series statistical analysis on user health behavior datasets to determine the distribution characteristics of sedentary time periods; Based on user intervention preference features and sitting posture behavior features, combined with the distribution features of sedentary time periods, multi-dimensional feature integration and attribute label mapping processing are performed to construct a user sedentary health behavior profile.

[0052] User health behavior dataset refers to a structured storage collection that gathers a single user's full-cycle sitting sensing, posture quantification, graded reminders, intervention feedback, and external device linkage records (including effective sitting cycle sensing data, multimodal reminder execution records, and wristband / adjustable desk feedback records), providing a full range of original data sources for user profile construction.

[0053] Historical intervention behavior data refers to all interaction records generated by users in response to multimodal reminder commands (such as reminder response status, response delay duration, stretch completion records, user likes / dislikes feedback), which are used to extract user intervention acceptance preferences.

[0054] Historical authorized device feedback data refers to historical behavior records (such as walking / heart rate data from smart bracelets, standing office time at height-adjustable desks, and workstation lighting linkage trigger records) transmitted back by external smart devices that have been pre-authorized by the user through a preset communication link. This data serves as an objective basis for verifying the effectiveness of the intervention behavior.

[0055] Historical sitting cycle records refer to the start and end times of each valid sitting cycle, the average sitting posture quantitative score within the cycle, and the duration of each cycle, which are archived and stored to statistically analyze the user's daily sedentary patterns.

[0056] User intervention preference characteristics refer to a set of user tendencies regarding reminder modalities, reminder timing, and corrective stretching content obtained through multi-dimensional statistics (such as preference for voice reminders, aversion to forced screen lock, and adaptation to short-duration stretching of 1-3 minutes).

[0057] Sitting posture characteristics refer to a set of normalized sitting posture defects extracted from historical sensor data (such as long-term rightward tilting of the sitting posture, excessive high-frequency neck pressure, and high incidence of crossing legs).

[0058] The distribution characteristics of prolonged sitting periods refer to the time-series attributes obtained from the full-scale sitting cycle statistics, such as the high-incidence prolonged sitting intervals, the peak period of prolonged sitting on a single day, and the interval of extremely long sitting on a single occasion.

[0059] Multi-dimensional feature integration and attribute tag mapping refers to the unified classification and integration of three heterogeneous features: intervention, posture, and time period, and the processing logic of converting numerical statistical features into standardized user attribute tags (such as mapping multiple ignoring of the 90-minute mandatory reminder to the tag of rejecting the level 4 screen lock intervention, and mapping the highest proportion of prolonged sitting from 13:30 to 17:00 every day to the tag of the high-risk prolonged sitting period in the afternoon).

[0060] Preset natural time periods refer to standard time segments divided by fixed time intervals (such as 09:00–12:00 morning period, 13:30–17:00 afternoon period, 18:00–21:00 evening overtime period, and basic sub-time periods every 30 minutes). They are divided using a universal clock time and do not have custom offsets based on user schedules. They are used to aggregate and statistically analyze user seating behavior in different time periods.

[0061] Optionally, the above steps are preliminary data modeling steps for personalized adaptive adjustment, following the aforementioned user health behavior dataset archiving process. The specific execution process is as follows: The first step is to retrieve the complete user health behavior dataset of the corresponding user stored locally or in the cloud, and lock all the historical archived data required to build the current profile; The second step is to extract three types of core raw data from the user health behavior dataset: historical intervention behavior data, historical authorized device feedback data, and historical seating cycle record data. The third step is to simultaneously perform multi-dimensional feature statistical operations on the three types of raw data extracted: based on historical intervention behavior data and historical authorized device feedback data, statistical analysis is conducted on the user's acceptance and rejection patterns of various reminders and corrective actions, and long-term fixed bad sitting posture patterns are extracted simultaneously to output user intervention preference features and sitting posture behavior features. The fourth step is to conduct time-series statistical analysis on all historical sitting cycle records in the user health behavior dataset, aggregate and statistically analyze the sitting frequency and single sitting duration distribution according to preset natural time periods, and output the sitting time distribution characteristics. The fifth step involves taking three types of structured features—user intervention preference features, sitting posture behavior features, and sedentary time distribution features—as input, performing multi-dimensional feature integration and attribute label mapping processing, converting the numerical statistical results into a standardized set of user attribute labels, and integrating them to generate a user sedentary health behavior profile.

[0062] For example, using a dataset of user health behavior data of programmer user Zhang San who has used the target cushion for a total of 2 weeks as the execution sample, a user health behavior profile is constructed. The specific process is as follows: The first step is to retrieve the user health behavior dataset corresponding to Zhang San. The dataset contains complete archives of all cushion sensor records within 2 weeks, 126 multimodal reminder issuance records, daily linkage feedback from the smart bracelet, interaction records of the height adjustment table, and start and end time series data of all valid sitting cycles. The second step is to extract three types of raw data from the dataset: historical intervention behavior data, including all reminder response records, stretching completion feedback, and user active evaluation; historical authorized device feedback data, including daily heart rate of wristband, walking and standing records, and standing office time of height-adjustable desk; and historical sitting cycle record data, including the start and end time of each effective sitting period and the corresponding quantitative score of sitting posture. The third step is to perform multi-dimensional feature statistics on the three types of raw data: the statistics show that the user intervention preference features are high acceptance of voice reminders, multiple ignoring of forced screen lock level 4 reminders, and only a preference for short stretching of 1 to 3 minutes; the sitting posture behavior features are extracted simultaneously, such as long-term slight right-leaning sitting posture and high frequency of excessive neck pressure. The fourth step is to perform time-series statistical analysis on all historical sitting cycle records, aggregate the sitting time of each time period, and obtain the distribution characteristics of prolonged sitting time: 13:30 to 17:00 on a single day is the peak period of prolonged sitting, and a single continuous sitting session often reaches 90 to 120 minutes. The fifth step involves inputting the aforementioned intervention preference characteristics, sitting posture behavior characteristics, and sedentary time distribution characteristics into the feature integration and label mapping module to complete feature classification and generate standardized attribute labels: high work intensity, difficulty in interrupting coding focus periods, right-handedness leading to right pelvic tilt, high incidence of neck pain, rejection of forced screen-locking intervention, preference for short-term neck and shoulder stretching, and afternoon hours being a high-risk sedentary period. All standardized attribute tags are integrated and archived to generate a user profile of Zhang San's sedentary health behavior. This profile can then be used to achieve personalized adaptation when matching posture correction guidance and adjusting graded early warning strategies.

[0063] In some embodiments, step 105 includes the following sub-steps: Extract intervention preference features, physical characteristics, and sedentary time distribution features from the user's sedentary health behavior profile; Extract the cumulative sitting time and target sitting posture type from the sitting posture quantification results data; According to the preset mapping rules, match the posture adjustment scheme based on the cumulative sitting time and the targeted posture correction scheme based on the target sitting posture type; Combine posture adjustment plans and targeted posture correction plans to generate preliminary sitting posture correction guidance information; Based on intervention preference characteristics, body posture characteristics, and the distribution characteristics of sedentary time periods, the preliminary sitting posture correction guidance information is filtered and the arrangement of the plan is optimized to generate corresponding target sitting posture correction guidance information.

[0064] Body posture characteristics refer to the set of tags in the portrait that record the user's long-term fixed posture defects and frequently strained areas (such as unilateral pelvic force imbalance, long-term strain of neck and shoulder muscles, and continuous slight tilt of the trunk).

[0065] Posture adjustment programs refer to standardized relief and adjustment programs that are suitable for common sedentary fatigue and do not involve specific postural deviations, and are not limited to stretching movements.

[0066] A targeted posture correction program refers to a specific adjustment program designed for specific postural defects such as trunk tilt, muscle strain, and pelvic misalignment.

[0067] The preliminary posture correction guidance information refers to standardized correction content obtained solely based on the quantitative results of posture and the preset mapping rules, without taking into account individual user preferences and physiological characteristics for personalized screening and adjustment.

[0068] The scheme filtering process refers to using the three types of characteristics of the profile as screening conditions to eliminate adjustment schemes that do not match the user's tolerance and physical condition.

[0069] The scheme layout optimization process refers to adjusting the display priority, recommendation order, and single execution duration of the remaining adjustment schemes after filtering.

[0070] Optionally, step 105, based on the quantitative results of sitting posture and combined with the user's sedentary health behavior profile, generates preliminary correction guidance information through standardized mapping rules, and then completes two-layer personalized optimization based on the profile features. The specific execution process is as follows: The first step is feature extraction: retrieve the user's sedentary health behavior profile corresponding to the current user, and extract three types of screening features from it: intervention preference features, body posture features, and sedentary time distribution features; at the same time, extract two matching parameters, namely the cumulative sedentary time and the target sedentary posture type, from the current sitting posture quantification results data. The second step is standardized scheme matching: retrieve the pre-stored preset mapping rule library, use the cumulative sitting time as the matching benchmark to match the corresponding general posture adjustment scheme, and then use the target sitting posture type as the matching benchmark to match the corresponding targeted posture correction scheme. The third step is to synthesize basic information: integrate and summarize the matched general posture adjustment scheme and the targeted posture correction scheme to generate preliminary sitting posture correction guidance information. The fourth step is a two-layer optimization process: based on the extracted intervention preference features, body posture features, and sedentary time distribution features, the preliminary posture correction guidance information is sequentially subjected to scheme filtering and scheme arrangement optimization processes. Step 5, Information Output and Push: Integrate the preliminary posture correction guidance information after the completion of the two-layer optimization, supplement the key points of standardized action execution and the single training time reminder, generate the target posture correction guidance information and push it to the user's office terminal for display.

[0071] In some embodiments, for example, the sample of programmer user Zhang San mentioned above is used, and the user sedentary health behavior profile generated during his 2-week usage period is reused. This profile is then combined with the quantitative results of example sitting posture to generate target sitting posture correction guidance information. The specific process includes: The first step was to retrieve Zhang San's user profile of sedentary health behavior and extract three core characteristics: intervention preference characteristics, which indicated acceptance of only 1 to 3 minutes of short-term adjustment and rejection of complex multi-step adjustment processes; physical characteristics, which indicated high-frequency strain on the neck and shoulders and long-term imbalance of the right pelvis; and the distribution of sedentary time, which was 13:30-17:00, belonging to the high-risk sedentary period. At the same time, the current sitting posture quantitative results data were extracted: the cumulative sitting time was 30 minutes, and the target sitting posture type was mild right tilt. The second step is to call the preset mapping rule library to complete the standardized matching: based on the 30-minute short sitting duration, a general posture adjustment program is matched, which includes three adjustment processes: neck movement, shoulder relaxation, and wrist relaxation; based on the mild right-leaning sitting posture type, a targeted posture correction program is matched, which includes two adjustment processes: left-side muscle relaxation and center of gravity balance training. The original program material has a maximum execution time of 4 minutes. The third step is to combine and summarize the two types of adjustment schemes mentioned above to generate preliminary sitting posture correction guidance information; The fourth step is to perform two-layer optimization based on the user's sedentary health behavior profile: filter the execution plan, remove 4-minute long adjustment materials and reduce the cumbersome balance training sub-processes; optimize the execution plan layout, increase the display priority of adjustment plans for neck and shoulder strain, and replace them with a simplified process of 1 to 3 minutes. The fifth step is to integrate and optimize the existing preliminary posture correction guidance information, supplement the key points of force exertion and the total duration of a single training session, generate target posture correction guidance information that is adapted to Zhang San's physical defects, strain symptoms and short-term adjustment preferences, and push it to Zhang San's office computer terminal for display.

[0072] In some embodiments, the above-mentioned posture quantification and correction guidance method based on multi-source seat surface perception further includes: Obtain status data from external authorized devices and schedule data from the office system; Perform scene feature matching on external authorized device status data and office system schedule data to identify the current work scenario; Extract the cumulative sitting time from the sitting posture quantification results data and match the cumulative sitting time with the target duration threshold range; A target-level early warning strategy that matches target duration threshold ranges; Based on the current working scenario, threshold calibration and modal combination matching are performed on the target hierarchical early warning strategy to generate multimodal reminder execution instructions.

[0073] External authorized device status data refers to the real-time operation and perception status data of external smart devices that have been pre-authorized and bound by the user (such as smart bracelet motion perception data, smart height adjustment table height status data, and smart light working status data). This data is used to assist in verifying the user's real-time posture and activity status and to provide hardware data support for scene recognition.

[0074] Office system schedule data refers to the structured schedule time-series data synchronized by the terminal office system (such as meeting time slots and work task time slots), including schedule start and end times and schedule type tags, which are used to accurately identify differentiated work scenarios such as meetings and focused work.

[0075] Scene feature matching refers to the process of comparing and classifying features of external device status data, office schedule data, and terminal operation behavior data based on preset scene recognition rules, in order to distinguish the differentiated work scenarios in which users are currently located.

[0076] The current work scenario refers to the user's real-time work status label obtained through feature matching and identification (such as coding focus scenario, meeting and office scenario, regular office scenario, and high-risk scenario of working overtime and sitting for long periods of time), which provides a scenario basis for the dynamic adjustment of early warning strategies.

[0077] The duration threshold range refers to the pre-divided standardized sedentary duration range (such as short-term sedentary range, regular sedentary range, and high-risk excessive sedentary range), and different ranges correspond to different levels of basic graded early warning strategies.

[0078] The graded early warning strategy refers to a set of pre-set standardized strategies for intervening in sedentary behavior. The early warning levels are divided according to the duration of sedentary behavior, and include core strategy parameters such as reminder trigger frequency, reminder execution intensity, multimodal reminder combination methods, and external device linkage rules.

[0079] Threshold calibration and modal combination matching processing refers to the secondary optimization of the graded early warning strategy in combination with the current work scenario. Threshold calibration is to adjust the sedentary trigger duration and reminder interval duration, while modal combination matching is to adjust the combination form and execution intensity of reminder modalities such as voice, pop-up window, and device linkage to adapt to the scenario.

[0080] Multimodal reminder execution commands refer to executable control commands generated after strategy calibration and modality matching, which can simultaneously trigger multi-dimensional intervention actions such as terminal message push, voice broadcast, and external smart device linkage reminders.

[0081] Optionally, the above steps constitute a scenario-based adaptive optimization process for sedentary intervention strategies. Based on standardized hierarchical early warning strategies, personalized and scenario-based strategy calibration is performed in conjunction with the user's current work scenario, ultimately generating multimodal reminder instructions adapted to the current scenario. The specific execution process is as follows: The first step is real-time data collection: synchronously retrieve the current status data of authorized external smart devices and the latest schedule data of the office system, and at the same time collect terminal desktop operation behavior data, including keyboard input frequency, mouse operation trajectory, and the type of application running in the foreground, and collect them into the raw dataset for scene recognition. The second step is work scenario identification: Pre-defined scenario feature matching rules are retrieved, and multi-dimensional feature comparison and matching analysis is performed on the original scenario identification dataset to identify the user's current work scenario and output standardized scenario labels. For example, high-frequency keyboard input and code application running features are used to determine coding focus scenarios; office schedule and meeting time slots are used to determine meeting scenarios; and the frequency of prolonged sitting during fixed time periods is used to determine high-risk prolonged sitting scenarios. All other states are uniformly identified as regular office scenarios. The third step is time interval matching: extract the cumulative sitting time from the current sitting posture quantification result data, compare and match it with the preset various sitting time threshold intervals, and determine the target time threshold interval corresponding to the cumulative sitting time. The fourth step is basic strategy matching: Based on the target duration threshold range obtained by matching, the target hierarchical early warning strategy corresponding to the target duration threshold range is retrieved, and the reminder frequency, early warning level, modal combination and device linkage rules corresponding to the cumulative sedentary time are determined. The fifth step is scenario-based strategy calibration: Based on the identified current work scenario, threshold calibration and modality combination matching are performed on the target-level early warning strategy. Optionally, for scenarios that cannot be interrupted, such as focused coding and meetings, the reminder trigger threshold is extended and the intensity of immediate intervention is reduced; for high-risk sedentary office scenarios in the afternoon, such as fatigue-related office scenarios, the reminder interval is shortened, the early warning level is increased, and the intensity of device-linked intervention is strengthened; for reminder modalities that users have historically disliked, the modality combination is replaced and optimized. Step 6, Command Integration and Generation: Assemble the calibrated early warning strategy parameters into commands, integrate multi-dimensional execution parameters such as terminal reminders, voice broadcasts, and smart device linkage, generate multi-modal reminder execution commands, and execute intervention actions according to the multi-modal reminder execution commands.

[0082] For example, following the application scenario of programmer user Zhang San, and combining his work characteristics, sedentary habits, and device authorization status, the entire process of scenario recognition, policy calibration, and instruction generation is executed. The specific implementation process is as follows: The first step is to collect current scene data: retrieve the current status data of the smart bracelet, smart height-adjustable desk and smart lights that Zhang San has authorized, read the office system schedule data at the same time, monitor the terminal operation behavior, identify that the front desk is currently running a code compilation program, the keyboard is in a high-frequency input state, there is no meeting schedule, and the current time is 14:20, which belongs to the high-risk prolonged sitting period of 13:30-17:00. The second step is to perform scene feature matching: combining the characteristics of high-frequency coding operations on the terminal and the characteristics of high-risk sedentary periods, matching the preset scene rules, determining that the user is currently in a scene of concentrated coding + high-risk sedentary periods in the afternoon, and outputting the corresponding standardized scene label. The third step is to match the sedentary duration threshold range: extract the cumulative sedentary duration from the current sitting posture quantification result as 45 minutes, match the preset duration threshold range, which corresponds to the regular sedentary threshold range, and call the corresponding fifth-level warning strategy. The fifth-level warning strategy: the reminder interval is set to 50 minutes, and the reminder method is: first-level warning reminder, using a combination of pop-up window and lock screen reminder mode. Step 4, Contextualized Strategy Calibration Processing: This involves performing dual calibration by combining the identified overlay working scenarios. First, threshold calibration: for coding focus scenarios, the reminder trigger time is delayed by 5 minutes to avoid interrupting the user's coding and debugging process; for high-risk sedentary scenarios in the afternoon, the original 50-minute reminder interval is shortened to 45 minutes, and the warning level is upgraded to level two to increase the intensity of intervention. Second, modal combination matching. Based on the user profile characteristics of disliking forced screen lock and preferring voice reminders, the forced screen lock modality was canceled, the core voice broadcast modality was retained, and a new linkage reminder modality of smart bracelet vibration and orange light flashing at the workstation was added, with points incentive reminders superimposed simultaneously. The fifth step is to assemble and generate execution instructions: integrate the calibrated trigger thresholds (including reminder trigger duration and reminder interval), secondary warning reminders, and multimodal combination parameters of voice broadcast + wristband vibration + light warning to generate complete multimodal reminder execution instructions. These instructions can be triggered instantly when the user finishes compiling code or when there is an operation gap. At the same time, they actively push reminders to drink water and move around, taking into account the user's work focus and sedentary health intervention needs, and realizing scenario-based and personalized precise sedentary intervention.

[0083] In some embodiments, the above-mentioned posture quantification and correction guidance method based on multi-source seat surface perception further includes: Acquire user intervention behavior data in response to multimodal reminder execution commands; After binding the periodic analysis dataset, posture quantification results data, multimodal reminder execution instructions, and intervention behavior data, the sitting cycle record data corresponding to the effective sitting cycle is generated; Store the periodic record data into the user's health behavior dataset; At preset time intervals, a health analysis model is used to perform health analysis on the user's health behavior dataset and generate a user periodic health analysis report.

[0084] Intervention behavior data refers to the behavioral records generated by users responding to multimodal reminders (including whether they responded, the duration of the response delay, the type of stretching movements completed, etc.), which are used to evaluate the intervention effect and user compliance.

[0085] Periodic record data refers to a structured archive that stores complete information about a single effective sitting period (including the start and end time of sitting, cumulative sitting time, quantitative score of sitting posture, issuance of reminder instructions, intervention verification results, etc.), and completely retains the data of a single sitting process.

[0086] User lifecycle health analysis reports refer to summary documents (such as daily, weekly, and monthly reports) of users' sedentary behavior and posture health generated at fixed time intervals, which include statistical indicators, risk level assessments, and targeted improvement suggestions.

[0087] Optionally, the above steps constitute a closed-loop data archiving and analysis process, the specific process of which is as follows: First, after the multimodal reminder execution command is issued, the user's response to the multimodal reminder and subsequent actions are collected in real time to generate corresponding intervention behavior data, which is then linked to the corresponding multimodal reminder command and the effective seating period. Subsequently, the periodic analysis dataset, posture quantification result data, multimodal reminder execution instructions, and intervention behavior data corresponding to the same valid sitting cycle are bound and associated with the cycle number as a unique identifier, and uniformly archived and stored in the user health behavior dataset of the corresponding user. Finally, when the preset time interval node is reached, the health analysis model is invoked to aggregate, statistically analyze, compare trends, and classify the full range of periodic data in the user's health behavior dataset, generating a corresponding periodic user periodic health analysis report, which is then output to the user's terminal for display.

[0088] For example, taking user Zhang San's daily health analysis as an example, the preset time interval is 24 hours. The health analysis model includes four types of rules: duration statistics, sitting posture percentage statistics, response rate calculation, and risk classification. The specific process is as follows: The first step is to generate 4 valid sitting cycle records on the same day. Each record is bound to the corresponding cycle analysis dataset, sitting posture quantification results, graded reminder instructions and user intervention behaviors, and is uniformly archived into Zhang San's user health behavior dataset. Statistics show that a total of 12 tiered alerts were triggered on the day, with 8 valid responses and 4 non-responses collected. The corresponding intervention behavior data were synchronously linked to the periodic record data of the valid seating period. The second step involves triggering a cycle analysis at 24:00 on the same day. The health analysis model statistically analyzes all data within the user's health behavior dataset for the day: total sedentary time was 6 hours and 10 minutes, the longest single cycle was 120 minutes, the average posture health score was 76 points, the proportion of poor posture was 35%, and the reminder response rate was 66%. Based on the risk assessment, the patient was determined to be at the moderate level of prolonged sitting. Combined with the characteristic of frequent right-leaning sitting posture, corresponding improvement suggestions were matched, and finally, a daily health analysis report for Zhang San was generated and pushed to the user's terminal according to the preset push channel.

[0089] In some embodiments, the above-mentioned posture quantification and correction guidance method based on multi-source seat surface perception further includes: Extract user activity status data from intervention behavior data returned by external authorized devices, corresponding to multimodal reminder execution commands; According to the preset intervention effectiveness judgment rules, verify the effectiveness of the intervention response to the user activity status data, generate the intervention response verification result and bind it to the periodic record data; If the response status in the intervention response verification result is "valid standing up", then the cumulative sedentary time is reset.

[0090] User activity status data refers to objective human activity data (such as walking records from a wristband's accelerometer, heart rate values, and standing time when sitting down from a height-adjustable desk) collected and transmitted back by external authorized devices through built-in sensing units. This data is used to objectively verify whether a user has engaged in standing-up activity.

[0091] The preset intervention effectiveness judgment rules refer to standardized objective judgment logic (such as continuous walking time of 2 minutes or standing office time of no less than 3 minutes is judged as effective getting up; simply closing the pop-up window or having no device activity data is judged as an invalid response), which is used to distinguish between real activities and fake operations.

[0092] Intervention response verification results refer to the structured records (including response status, response duration, effective activity duration, corresponding reminder instruction number, etc.) generated after verification based on objective sensor data from the device. These records are used for archiving and iteratively updating the user's sedentary health behavior profile.

[0093] Optionally, the above steps constitute a closed-loop feedback loop for sedentary intervention. This is executed after the multimodal reminder instruction is issued, verifying the authenticity of the user's response using objective sensor data returned by external smart devices, completing data archiving, and updating the sedentary timing baseline. The specific execution process is as follows: The first step is targeted data extraction: traverse the current intervention behavior data, filter out user activity status data returned by external authorized devices that carry the unique identifier of the corresponding multimodal reminder execution command, and remove historical device sensor records that are not bound to this reminder; The second step is to verify the effectiveness of the intervention: retrieve the preset rules for determining the effectiveness of the intervention, compare the user activity status data with the judgment threshold, perform the intervention response effectiveness verification, and generate an intervention response verification result that includes the response status, response duration, and duration of effective activity. The third step is data archiving and binding: the intervention response verification results are bound and archived with the period record data corresponding to the current valid seating period, and synchronously written into the user health behavior dataset as the original historical intervention behavior data for subsequent profile iterations; The fourth step is to update the timing logic: read the response status field in the intervention response verification result. If it is determined to be a valid standing up, clear the currently continuously accumulated sitting time and restart a new round of sitting timer; if it is an invalid response or no response, keep the accumulated sitting time accumulating and wait for the next round of graded warning to be triggered.

[0094] For example, using the application scenario of programmer user Zhang San mentioned earlier, we reuse his hardware configuration of binding a smart bracelet, an electric height-adjustable desk, and workstation lights to fully execute the entire process of intervention feedback verification, data archiving, and timer reset: The first step is to issue a Level 2 multimodal reminder execution command. This command carries a unique number R_20241215_1430 and simultaneously triggers three reminder modes: voice broadcast, wristband vibration, and flashing orange light at the workstation. 35 seconds after the reminder is issued, Zhang San gets up and walks around to stretch. His smart wristband collects walking data through the accelerometer and simultaneously transmits user activity status data: continuous walking duration of 3 minutes, 250 steps, and a simultaneous increase in heart rate. The wristband activity status data with the binding number R_20241215_1430 is extracted from the data of this intervention behavior, and records from other irrelevant devices at other times are removed. The second step is to retrieve the preset intervention effectiveness judgment rules for verification. If the wristband walks continuously for 3 minutes to reach the effective activity time threshold, the intervention response status is determined to be effective standing up. At the same time, the response time of 35 seconds is calculated, and a complete intervention response verification result is generated, including the instruction number, effective response status, 35-second response time, and 3-minute effective activity time. The third step is to bind the intervention response verification result to the periodic record data of the current valid sitting cycle, and simultaneously archive it to Zhang San's exclusive user health behavior dataset as historical intervention behavior data for subsequent updates to the user's sedentary health behavior profile and to strengthen the 52-minute interval reminder strategy. The fourth step is to read the response status from the intervention response verification results. If the response status is "valid getting up", the timer reset logic is executed to clear the current cumulative 52 minutes of sitting time. The seat cushion sensor module starts counting the new round of sitting time from zero. If the user repeatedly turns off the level 4 forced screen lock reminder and the wristband has no activity sensor data, the response status is marked as invalid response, the cumulative sitting time continues to accumulate, and the forced screen lock mode is automatically avoided when generating reminder instructions in the future.

[0095] In some embodiments, the above-mentioned posture quantitative correction guidance method based on multi-source seat surface perception is also equipped with a gamified health incentive mechanism as a follow-up application to the multimodal reminder intervention. Specifically, health points are calculated based on the intervention response verification results, posture quantitative data, and daily sedentary statistics: points can be accumulated for effectively getting up, completing stretching, and achieving the posture score standard, while points are deducted for prolonged sitting or ignoring reminders multiple times; tiered achievement badges are set up simultaneously, which are unlocked and rewarded with points based on usage time, consecutive healthy days, and cumulative stretching count.

[0096] Simultaneously, it supports the formation of teams to conduct periodic health challenges, ranking teams based on their average daily sedentary time, reminder response rate, and stretching completion rate, and awarding them a batch of points. The earned points can be redeemed for corresponding rewards, thereby improving user cooperation through dual incentives of individual growth and team competition.

[0097] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0098] The following describes the posture quantification correction guidance device based on multi-source seat surface sensing provided in the embodiments of this application. The posture quantification correction guidance device based on multi-source seat surface sensing described below can be referred to in correspondence with the posture quantification correction guidance method based on multi-source seat surface sensing described above.

[0099] Reference Figure 2 , Figure 2 This is an optional structural schematic diagram of a posture quantification and correction guidance device based on multi-source seat surface sensing provided in an embodiment of the present invention. The device is used to execute the aforementioned posture quantification and correction guidance method based on multi-source seat surface sensing, including: Data acquisition module 201 is used to acquire multi-source seat surface sensing data of the target seat cushion; The seating determination module 202 is used to integrate multi-source seat surface perception data, perform user seat status recognition, and determine the effective seating period corresponding to the seating status after recognizing that the user is in a seated state. The periodic interception module 203 is used to intercept multi-source seat surface sensing data within the effective seating period to obtain a periodic analysis dataset. The sitting posture quantification module 204 is used to perform time-series statistics and sitting posture feature extraction on the periodic analysis dataset to generate sitting posture quantification result data. The correction matching module 205 is used to generate target sitting posture correction guidance information based on preset mapping rules and combined with the user's sedentary health behavior profile, and to generate quantitative results data matching of sitting posture.

[0100] In some embodiments, the seating determination module 202 is configured as follows: By fusing multi-source seat surface perception data, the user's seat status is identified based on the fusion result; If the user's seat status is "sitting down", then record the start time of the sitting down status. If a user's seat status changes from seated to unseated, the corresponding state change time is recorded, and the valid seated period is determined based on the state start time and state change time. The above-mentioned fusion of multi-source seat surface perception data, based on the fusion result, identifies the user's seat status, including: Extract the total seat pressure, seat temperature, and seat pressure distribution features from multi-source seat sensing data, and perform multi-dimensional joint fusion judgment; If the total pressure on the seat surface is greater than the first pressure threshold, and the seat surface temperature is higher than the human body temperature threshold, and the seat surface pressure distribution meets the characteristics of human body sitting pressure distribution, then the user's seat state is determined to be a sitting state. If the total pressure on the seat is less than the second pressure threshold and the duration exceeds the preset duration, or if the seat temperature is lower than the ambient temperature threshold, the user's seat status will be determined as an unseated state.

[0101] In some embodiments, the periodic interception module 203 is configured as follows: Extract the start time of the valid seating cycle; Multi-source seat-surface sensing data between the start time and the current analysis time are extracted to obtain a preliminary analysis dataset; The preliminary analysis dataset is time-aligned and outlier filtered to generate a periodic analysis dataset.

[0102] In some embodiments, the sitting posture quantification module 204 is configured as follows: Perform time-series continuity statistics on the periodic analysis dataset to calculate the cumulative sitting time within the effective sitting period; Extract the cyclic pressure distribution characteristics from the cyclic analysis dataset; The periodic pressure distribution characteristics are input into a pre-built quantitative assessment model of sitting posture. After sitting posture classification and quantitative scoring, the target sitting posture type and sitting posture health score are output. By integrating cumulative sitting time, target sitting posture type, and sitting posture health score, quantitative results of sitting posture are generated.

[0103] In some embodiments, the correction matching module 205 is configured as follows: Extract intervention preference features, physical characteristics, and sedentary time distribution features from the user's sedentary health behavior profile; Extract the cumulative sitting time and target sitting posture type from the sitting posture quantification results data; According to the preset mapping rules, match the posture adjustment scheme based on the cumulative sitting time and the targeted posture correction scheme based on the target sitting posture type; Combine posture adjustment plans and targeted posture correction plans to generate preliminary sitting posture correction guidance information; Based on intervention preference characteristics, body posture characteristics, and the distribution characteristics of sedentary time periods, the preliminary sitting posture correction guidance information is filtered and the arrangement of the plan is optimized to generate corresponding target sitting posture correction guidance information.

[0104] In some embodiments, the above-mentioned posture quantification and correction guidance device based on multi-source seat surface perception further includes a portrait construction module, which is configured as follows: Obtain user health behavior dataset; Extract historical intervention behavior data, historical authorized device feedback data, and historical seating cycle record data from the user health behavior dataset; Multi-dimensional feature statistics were performed on historical intervention behavior data, historical authorized device feedback data, and historical seating cycle record data to determine user intervention preference characteristics and sitting posture behavior characteristics. Perform time-series statistical analysis on user health behavior datasets to determine the distribution characteristics of sedentary time periods; Based on user intervention preference features and sitting posture behavior features, combined with the distribution features of sedentary time periods, multi-dimensional feature integration and attribute label mapping processing are performed to construct a user sedentary health behavior profile.

[0105] In some embodiments, the above-mentioned posture quantification and correction guidance device based on multi-source seat surface perception further includes a sedentary reminder module, which is configured as follows: Obtain status data from external authorized devices and schedule data from the office system; Perform scene feature matching on external authorized device status data and office system schedule data to identify the current work scenario; Extract the cumulative sitting time from the sitting posture quantification results data and match the cumulative sitting time with the target duration threshold range; A target-level early warning strategy that matches target duration threshold ranges; Based on the current working scenario, threshold calibration and modal combination matching are performed on the target hierarchical early warning strategy to generate multimodal reminder execution instructions.

[0106] In some embodiments, the above-described posture quantification and correction guidance device based on multi-source seat surface perception further includes an intervention verification module, which is configured as follows: Extract user activity status data from intervention behavior data returned by external authorized devices, corresponding to multimodal reminder execution commands; According to the preset intervention effectiveness judgment rules, verify the effectiveness of the intervention response to the user activity status data, generate the intervention response verification result and bind it to the periodic record data; If the response status in the intervention response verification result is "valid standing up", then the cumulative sedentary time is reset.

[0107] In some embodiments, the above-mentioned posture quantification and correction guidance device based on multi-source seat surface perception further includes a health analysis module, which is configured as follows: Acquire user intervention behavior data in response to multimodal reminder execution commands; After binding the periodic analysis dataset, posture quantification results data, multimodal reminder execution instructions, and intervention behavior data, the sitting cycle record data corresponding to the effective sitting cycle is generated; Store the periodic record data into the user's health behavior dataset; At preset time intervals, a health analysis model is used to perform health analysis on the user's health behavior dataset and generate a user periodic health analysis report.

[0108] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described device and module can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0109] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0110] The modules described as separate components may or may not be physically separate. Similarly, the components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0111] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0112] If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0113] This invention also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the aforementioned posture quantification and correction guidance method based on multi-source seat surface perception. This electronic device can be any smart terminal, including a tablet computer.

[0114] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0115] Please see Figure 3 , Figure 3 The hardware structure of an electronic device according to another embodiment is illustrated, including: The processor 901 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present invention. The memory 902 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 902 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and called by the processor 901 to execute the posture quantification correction guidance method based on multi-source seat surface sensing of the embodiments of the present invention. The input / output interface 903 is used to implement information input and output; The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 905 transmits information between various components of the device (e.g., processor 901, memory 902, input / output interface 903, and communication interface 904); The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.

[0116] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described posture quantification correction guidance method based on multi-source seat surface perception.

[0117] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0118] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0119] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for quantitative correction guidance of sitting posture based on multi-source seat surface perception, characterized in that, include: Acquire multi-source seat surface perception data of the target seat cushion; By integrating the multi-source seat surface perception data, user seat status recognition is performed. After recognizing that the user is in a seated state, the effective seating period corresponding to the seated state is determined. Multi-source seat surface sensing data within the effective seating cycle are extracted to obtain a cycle analysis dataset; The periodic analysis dataset is subjected to time-series statistics and sitting posture feature extraction processing to generate sitting posture quantification result data; Based on preset mapping rules and combined with the user's sedentary health behavior profile, target posture correction guidance information is generated by matching the quantitative results of the sitting posture.

2. The sitting posture quantitative correction guidance method based on multi-source seat surface perception according to claim 1, characterized in that, The process of fusing the multi-source seat surface perception data and performing user seat status recognition, after recognizing that the user is seated, determines the effective seating period corresponding to the seated state, including: By fusing the multi-source seat surface perception data, the user's seat status is identified based on the fusion result; If the user's seat status is "sitting down", then record the start time of the "sitting down" status. If the user's seat status changes from the seated state to the unseated state, the corresponding state change time is recorded, and the effective seated period is determined based on the state start time and the state change time.

3. The sitting posture quantitative correction guidance method based on multi-source seat surface perception according to claim 2, characterized in that, The process of fusing the multi-source seat surface perception data and identifying the user's seat status based on the fusion result includes: Extract the total seat pressure, seat temperature, and seat pressure distribution features from the multi-source seat sensing data, and perform multi-dimensional joint fusion judgment; If the total pressure on the seat surface is greater than the first pressure threshold, and the seat surface temperature is higher than the human body temperature threshold, and the seat surface pressure distribution meets the human body sitting pressure distribution characteristics, then the user's seat state is determined to be a sitting state. If the total pressure on the seat surface is less than the second pressure threshold and the duration exceeds the preset duration, or if the seat surface temperature is lower than the ambient temperature threshold, then the user's seat status is determined to be an unseat state.

4. The sitting posture quantitative correction guidance method based on multi-source seat surface perception according to claim 1, characterized in that, The process of performing time-series statistics and posture feature extraction on the periodic analysis dataset to generate posture quantification result data includes: Perform time-series continuity statistics on the periodic analysis dataset to calculate the cumulative sitting time within the effective sitting period; Extract the periodic pressure distribution characteristics from the periodic analysis dataset; The periodic pressure distribution characteristics are input into a pre-constructed sitting posture quantitative assessment model. After sitting posture classification and quantitative scoring, the target sitting posture type and sitting posture health score are output. The cumulative sitting time, the target sitting posture type, and the sitting posture health score are integrated to generate quantitative sitting posture result data.

5. The sitting posture quantitative correction guidance method based on multi-source seat surface perception according to claim 1, characterized in that, The construction of the user's sedentary health behavior profile includes: Obtain user health behavior dataset; Extract historical intervention behavior data, historical authorized device feedback data, and historical seating cycle record data from the user health behavior dataset; Multi-dimensional feature statistics are performed on the historical intervention behavior data, the historical authorized device feedback data, and the historical sitting cycle record data to determine user intervention preference characteristics and sitting posture behavior characteristics. Time-series statistical analysis was performed on the user health behavior dataset to determine the distribution characteristics of sedentary time periods; Based on the user intervention preference features and the sitting posture behavior features, combined with the sedentary time period distribution features, multi-dimensional feature integration and attribute label mapping processing are performed to construct a user sedentary health behavior profile.

6. The sitting posture quantitative correction guidance method based on multi-source seat surface perception according to claim 1, characterized in that, The process of generating target posture correction guidance information based on preset mapping rules and combined with a user's sedentary health behavior profile, matching the quantitative results of the sitting posture data, includes: Extract intervention preference features, physical characteristics, and sedentary time distribution features from the user's sedentary health behavior profile; Extract the cumulative sitting time and target sitting posture type from the sitting posture quantification results data; According to the preset mapping rules, the posture adjustment scheme based on the cumulative sitting time and the targeted posture correction scheme based on the target sitting posture type are matched; The posture adjustment scheme and the targeted posture correction scheme are combined to generate preliminary sitting posture correction guidance information; Based on the intervention preference characteristics, body posture characteristics, and sedentary time distribution characteristics, the preliminary posture correction guidance information is filtered and the layout of the plan is optimized to generate the corresponding target posture correction guidance information.

7. The sitting posture quantitative correction guidance method based on multi-source seat surface perception according to claim 4, characterized in that, Also includes: Obtain status data from external authorized devices and schedule data from the office system; The current work scenario is identified by matching the status data of the external authorized devices and the schedule data of the office system with scene features. Extract the cumulative sitting time from the sitting posture quantification result data, and match the cumulative sitting time with the target duration threshold range; A target-level early warning strategy that matches the target duration threshold range; Based on the current working scenario, the target hierarchical early warning strategy is subjected to threshold calibration and modal combination matching processing to generate multimodal reminder execution instructions.

8. The sitting posture quantitative correction guidance method based on multi-source seat surface perception according to claim 7, characterized in that, Also includes: Acquire user intervention behavior data in response to the multimodal reminder execution command; After binding the periodic analysis dataset, the posture quantification result data, the multimodal reminder execution command, and the intervention behavior data, the sitting period record data corresponding to the effective sitting period is generated. The periodic record data is stored in the user's health behavior dataset; At preset time intervals, a health analysis model is used to perform health analysis on the user's health behavior dataset, generating a user periodic health analysis report.

9. The method for quantitative correction guidance of sitting posture based on multi-source seat surface perception according to claim 8, characterized in that, Also includes: Extract the user activity status data returned by the external authorized device from the intervention behavior data, corresponding to the multimodal reminder execution command; According to the preset intervention effectiveness determination rules, the effectiveness of the intervention response to the user activity status data is verified, an intervention response verification result is generated and bound to the periodic record data; If the response status in the intervention response verification result is "valid standing up", then the cumulative sedentary time is reset.

10. A posture quantification and correction guidance device based on multi-source seat surface sensing, characterized in that, include: The data acquisition module is used to acquire multi-source seat surface sensing data of the target seat cushion; The seating determination module is used to integrate the multi-source seat surface perception data, perform user seat status recognition, and determine the effective seating period corresponding to the seating status after recognizing that the user is in a seated state. The periodic interception module is used to intercept the multi-source seat surface sensing data within the effective seating period to obtain a periodic analysis dataset. The sitting posture quantification module is used to perform time-series statistics and sitting posture feature extraction processing on the periodic analysis dataset to generate sitting posture quantification result data. The correction matching module is used to generate target sitting posture correction guidance information that matches the quantitative sitting posture result data based on preset mapping rules and combined with the user's sedentary health behavior profile.

Images