Gait analysis based walking stability assessment and warning system for disabled elderly

Abstract

This invention provides a gait analysis-based walking stability assessment and early warning system for disabled elderly individuals, belonging to the field of geriatric health monitoring technology. It includes a data acquisition unit for collecting plantar pressure data, limb posture data, and global motion data during the walking process of disabled elderly individuals. This invention identifies trends of declining stability before falls occur through real-time assessment of gait characteristics, achieving proactive early warning rather than passively responding to fall events. This invention is based on an individual gait baseline for assessment, adapting to individual differences among disabled elderly individuals, avoiding assessment errors caused by universal standards, and reducing false alarms and false negatives. This invention combines the advantages of wearable devices and millimeter-wave radar, ensuring both detailed local gait acquisition and coverage of global motion states, reducing environmental interference and blind spots. The combination of millimeter-wave radar and flexible insoles avoids the privacy risks of video surveillance and is suitable for everyday home use.

Description

Technical Field

[0001] This invention relates to the field of elderly health monitoring technology, and in particular to a walking stability assessment and early warning system for disabled elderly people based on gait analysis. Background Technology

[0002] With the increasing aging of the population, the number of disabled elderly people continues to grow. Due to the decline in limb motor ability and weakened balance perception system, disabled elderly people are very prone to falls while walking. Falls often lead to serious consequences such as fractures and traumatic brain injuries, which not only seriously affect the quality of life of the elderly, but also bring a heavy care burden to their families and society.

[0003] Existing fall warning methods for the elderly mostly rely on wearable devices' accelerometer sensors, triggering alarms only after a fall occurs or at the moment of the fall, lacking early assessment and warning of gait stability. Some video-based gait analysis systems pose privacy risks and are severely affected by environmental factors such as lighting and occlusion, making them difficult to apply stably in everyday home settings. Furthermore, most existing technologies use universal gait assessment standards without personalized adaptation to the physical characteristics and gait patterns of disabled elderly individuals, resulting in insufficient assessment accuracy and high rates of false alarms and missed alarms. Therefore, this paper proposes a gait analysis-based gait stability assessment and early warning system for disabled elderly individuals. Summary of the Invention

[0004] The purpose of this invention is to address the problems raised in the existing background technology. To achieve the above-mentioned objective, this invention provides the following technical solution: a gait analysis-based walking stability assessment and early warning system for disabled elderly, comprising a data acquisition unit for collecting plantar pressure data, limb posture data, and global motion data during the walking process of disabled elderly; The gait feature extraction unit preprocesses the data collected by the data acquisition unit and extracts multi-dimensional gait features, including spatiotemporal features, dynamic features, and posture features. The personalized assessment unit compares real-time gait characteristics with the baseline of an elderly person with disabilities, and outputs a walking stability assessment score. The risk warning unit sets multi-level warning thresholds based on stability assessment scores, triggers corresponding level warning prompts, and pushes them to associated terminals.

[0005] Preferably, the data acquisition unit includes a flexible pressure-sensing insole, an inertial measurement module, and an environmental auxiliary acquisition module; The flexible pressure-sensing insole has a built-in distributed pressure sensor array to collect data on foot pressure distribution, pressure center trajectory, and gait support time. The inertial measurement module integrates an accelerometer, gyroscope, and magnetometer, and is worn on the limbs to collect limb movement posture data; The environmental auxiliary acquisition module is a low-power millimeter-wave radar used for non-contact acquisition of the elderly's stride length, walking speed, and global trajectory data.

[0006] Preferably, the flexible pressure sensing insole has a sampling frequency of ≥100Hz and adopts a flexible wearable design to adapt to different foot types; the inertial measurement module is worn at least on the thigh, calf and waist to assist in accurately dividing the gait cycle.

[0007] Preferably, the gait feature extraction unit includes a temporal feature model, which is constructed based on a long short-term memory network and is used to capture the trend features of gait data changing over time. The gait feature extraction unit calculates the pressure center offset rate during the gait cycle using the following formula. Used to quantify the degree of balance fluctuation during walking:

[0008] in: for Coordinates of the pressure center at any given time, in cm; The average coordinates of the pressure center over one gait cycle are given in cm. The average length of the sole of an elderly person's foot is in cm. The pressure center offset rate is dimensionless and ranges from 0 to 1.

[0009] Preferred personalized assessment units include: The baseline construction module continuously collects gait data from disabled elderly people in the initial stage and constructs a personalized gait health baseline based on their disability level and underlying disease type. The deviation calculation module calculates the characteristic deviation between real-time gait features and the baseline, including the frequency of pressure center offset, single-leg support time deviation rate, and limb swing symmetry deviation rate.

[0010] Preferably, the personalized evaluation unit uses an improved random forest algorithm to train a stability evaluation model, and the feature weights are optimized using the Gini index. The optimization formula is as follows:

[0011] in: The training dataset is dimensionless. For the first in the dataset The proportion of samples of each class, dimensionless; The total number of sample categories is dimensionless, corresponding to the number of risk levels in the stability assessment.

[0012] Preferably, the risk warning unit sets three warning thresholds: A Level 1 warning corresponds to a stability score of 60-80, suggesting that the elderly slow down their walking speed. A Level 2 warning corresponds to a stability score of 40-60, triggering a local voice prompt on the device and pushing a warning message to the nursing staff. A Level 3 warning corresponds to a stability score of <40, triggering a local audible and visual alarm and sending the elderly person's location and real-time gait data to the associated monitoring platform.

[0013] Preferably, the risk warning unit includes an adaptive threshold adjustment module, which dynamically adjusts the trigger sensitivity of the warning threshold based on environmental parameters such as ground friction and slope in the area where the elderly walk.

[0014] Preferably, the data management and feedback unit calculates the gait health trend change rate using the following formula. Used to assess changes in long-term walking stability in older adults:

[0015] in: The average stability score for the current cycle is dimensionless. The average stability score of the initial baseline period is dimensionless. This is the rate of change of the trend, dimensionless. A positive sign indicates increased stability, and a negative sign indicates decreased stability.

[0016] Preferably, the system adopts an edge computing and cloud collaborative architecture, where data preprocessing and real-time evaluation are completed on local edge devices, and non-real-time health data is uploaded to the cloud for storage and analysis, reducing data transmission latency and privacy leakage risks.

[0017] The beneficial effects of this invention are as follows: By real-time assessment of gait characteristics, it identifies trends of declining stability before a fall occurs, enabling proactive early warning rather than passive response to fall events. Assessment based on an individual gait baseline adapts to individual differences among disabled elderly individuals, avoiding assessment errors introduced by universal standards and reducing false alarms and false negatives. Combining the advantages of wearable devices and millimeter-wave radar ensures both detailed local gait acquisition and coverage of global movement states, reducing environmental interference and blind spots. The combination of millimeter-wave radar and flexible insoles avoids the privacy risks associated with video surveillance and is suitable for everyday home use. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings: Figure 1 A schematic diagram of the main functional units of the system provided by this invention; Figure 2 This is a schematic diagram of data acquisition provided by the present invention; Figure 3 A schematic diagram of the evaluation indicators provided by this invention. Detailed Implementation

[0019] Example 1: A walking stability assessment and early warning system for disabled elderly based on gait analysis. The system includes a data acquisition unit for collecting plantar pressure data, limb posture data and global motion data of disabled elderly during walking. The gait feature extraction unit preprocesses the data collected by the data acquisition unit and extracts multi-dimensional gait features, including spatiotemporal features, dynamic features, and posture features. The personalized assessment unit compares real-time gait characteristics with the baseline of an elderly person with disabilities, and outputs a walking stability assessment score. The risk warning unit sets multi-level warning thresholds based on stability assessment scores, triggers corresponding level warning prompts, and pushes them to associated terminals. The personalized assessment unit calculates the overall walking stability score S using the following formula:

[0020] in: For real-time data collection The gait dynamic characteristic values ​​are consistent with the corresponding baseline characteristics, including stride length (m), single-leg support time (s), and limb swing angle (°). This is the individual baseline value for this characteristic, in units of... Consistent; The total number of gait features involved in the calculation is dimensionless. The duration of single-leg support is measured in seconds (s). The duration of support with both feet, measured in seconds; The area of ​​the real-time pressure center trajectory, in units of ; The individual baseline value for the area of ​​the center of pressure trajectory, in units of ; Let be the weighting coefficient, satisfying , dimensionless, is determined by training with historical data on the health level and fall risk of disabled elderly.

[0021] The data acquisition unit includes a flexible pressure-sensing insole, an inertial measurement module, and an environmental auxiliary acquisition module; The flexible pressure-sensing insole has a built-in distributed pressure sensor array to collect data on foot pressure distribution, pressure center trajectory, and gait support time. The inertial measurement module integrates an accelerometer, gyroscope, and magnetometer, and is worn on the limbs to collect limb movement posture data; The environmental auxiliary acquisition module is a low-power millimeter-wave radar used for non-contact acquisition of the elderly's stride length, walking speed, and global trajectory data.

[0022] The flexible pressure-sensing insole has a sampling frequency of ≥100Hz and adopts a flexible wearable design to adapt to different foot types; the inertial measurement module is worn on at least the thigh, calf and waist to help accurately divide the gait cycle.

[0023] The gait feature extraction unit includes a temporal feature model, which is built based on a long short-term memory network and is used to capture the trend features of gait data over time. The gait feature extraction unit calculates the pressure center offset rate during the gait cycle using the following formula. Used to quantify the degree of balance fluctuation during walking:

[0024] in: for Coordinates of the pressure center at any given time, in cm; The average coordinates of the pressure center over one gait cycle are given in cm. The average length of the sole of an elderly person's foot is in cm. The pressure center offset rate is dimensionless and ranges from 0 to 1.

[0025] Personalized assessment units include: The baseline construction module continuously collects gait data from disabled elderly people in the initial stage and constructs a personalized gait health baseline based on their disability level and underlying disease type. The deviation calculation module calculates the characteristic deviation between real-time gait features and the baseline, including the frequency of pressure center offset, single-leg support time deviation rate, and limb swing symmetry deviation rate.

[0026] The personalized evaluation unit uses an improved random forest algorithm to train a stability evaluation model, and the feature weights are optimized using the Gini index. The optimization formula is as follows:

[0027] in: The training dataset is dimensionless. For the first in the dataset The proportion of samples of each class, dimensionless; The total number of sample categories is dimensionless, corresponding to the number of risk levels in the stability assessment.

[0028] The risk warning unit sets three warning thresholds: A Level 1 warning corresponds to a stability score of 60-80, suggesting that the elderly slow down their walking speed. A Level 2 warning corresponds to a stability score of 40-60, triggering a local voice prompt on the device and pushing a warning message to the nursing staff. A Level 3 warning corresponds to a stability score of <40, triggering a local audible and visual alarm and sending the elderly person's location and real-time gait data to the associated monitoring platform.

[0029] The risk warning unit includes an adaptive threshold adjustment module, which dynamically adjusts the trigger sensitivity of the warning threshold based on environmental parameters such as ground friction and slope in the area where the elderly walk.

[0030] It also includes a data management and feedback unit, which calculates the rate of change in gait health trends using the following formula. Used to assess changes in long-term walking stability in older adults:

[0031] in: The average stability score for the current cycle is dimensionless. The average stability score of the initial baseline period is dimensionless. This is the rate of change of the trend, dimensionless. A positive sign indicates increased stability, and a negative sign indicates decreased stability.

[0032] The system adopts an edge computing and cloud-based collaborative architecture. Data preprocessing and real-time evaluation are completed on local edge devices, while non-real-time health data is uploaded to the cloud for storage and analysis, reducing data transmission latency and the risk of privacy leaks. This system is based on multi-source data acquisition, with gait feature analysis as its core and personalized assessment as its basis, to achieve a closed-loop operation from data acquisition to risk warning. The specific workflow is divided into the following stages: Preliminary adaptation and baseline establishment phase: Customize flexible pressure-sensing insoles for disabled elderly people, adjust the fit of the sensor array according to the foot shape to ensure no blind spots in the acquisition of foot pressure; bind the inertial measurement module to the front of the elderly person's thigh, the outer side of the tibia and the waist to complete the initial zero-position calibration of the posture sensor and avoid data errors caused by initial angle deviation.

[0033] Daily walking data of the elderly was collected continuously for 3 to 7 days, covering typical home movements such as slow walking, turning around, and getting up, while simultaneously recording the elderly's underlying medical history and disability level information.

[0034] By combining collected data with clinical assessment standards, a personalized gait health baseline is established for the elderly, including normal ranges for characteristics such as stride length, support time, and center of gravity fluctuations, which serve as a reference benchmark for subsequent stability assessments.

[0035] Multi-source real-time data acquisition stage: Flexible pressure-sensing insoles capture the pressure distribution changes in areas such as the heel, forefoot, and arch during walking, record the pressure changes throughout the entire process of single-foot landing, support, and lift-off, and simultaneously output the dynamic changes in the center of gravity trajectory.

[0036] The inertial measurement module monitors the changes in the motion angles of the thigh, calf, and waist in real time, captures posture characteristics such as limb swing amplitude and flexion-extension frequency, assists in dividing the gait cycle, and accurately identifies key gait stages such as single-leg support and double-leg support.

[0037] Millimeter-wave radar transmits low-power millimeter-wave signals, which penetrate clothing to identify the micro-motion characteristics of the elderly's limbs. It records global motion information such as step length, walking speed, and walking trajectory without contact, supplementing the blind spots of wearable devices and avoiding data loss caused by clothing obstruction or limb movement.

[0038] Gait feature screening and extraction stage: The collected multi-source data is denoised to filter out environmental electromagnetic interference and limb micro-movement noise, and the data time synchronization of different acquisition devices is completed to ensure that the plantar pressure, limb posture and global trajectory data are aligned and matched in the time dimension.

[0039] Core features highly correlated with walking stability were selected from the synchronized data, including: Spatiotemporal characteristics: duration of single-leg support, duration of double-leg support, stride length uniformity, and stride speed stability. Balance characteristics: magnitude of center of gravity shift, symmetry of center of gravity movement Postural characteristics: matching degree of limb swing amplitude, changes in trunk tilt angle Abnormal Feature Marking: Features that deviate from the healthy baseline are automatically marked, such as single-leg support time being 20% ​​shorter than the baseline or center of gravity shifting beyond the normal range, generating a preliminary list of abnormal features.

[0040] Personalized stability assessment phase: The gait features extracted in real time are matched with the elderly’s exclusive health baseline item by item, and the degree of deviation of each feature is calculated. High-risk deviation types such as shortened single-leg support time, increased center of gravity shift, and decreased gait symmetry are given special attention.

[0041] The system adjusts feature weights based on the elderly person's disability level and underlying disease type. For example, it focuses on assessing limb swing symmetry for elderly people with stroke sequelae and on monitoring gait stability for elderly people with Parkinson's disease, ultimately outputting a comprehensive stability score. Based on the changing trends of the elderly person's daily walking data, the system automatically fine-tunes the health baseline range weekly to adapt to changes in the elderly person's rehabilitation training or physical condition, avoiding assessment biases caused by a fixed baseline.

[0042] Tiered risk warning stage: Based on the stability score and the preset warning level threshold, the walking risk is divided into four levels: stable, mild risk, moderate risk, and high risk.

[0043] Hierarchical response execution: Stable state: Only walking data is recorded, and health records are updated; Mild risk: The insole emits a gentle prompt through a built-in speaker, advising the elderly to slow down their walking speed; Moderate risk: Triggers the device's audio and visual alarm, pushing the elderly person's location and abnormal characteristics information to the caregiver's mobile device; High risk: Activate the emergency early warning mechanism, simultaneously push alerts to the community health platform and children's guardian terminal, and link the family emergency call device to ensure that intervention is received as soon as possible.

[0044] Scene-adaptive adjustment: Combining the walking scene identified by millimeter-wave radar, the warning threshold is automatically adjusted. For example, the warning sensitivity is increased in wet and slippery areas such as bathrooms, and the warning threshold is appropriately reduced in non-slip areas such as carpets to reduce false alarms.

[0045] Data feedback and rehabilitation assistance phase: Walking data, assessment results and early warning records are automatically summarized daily to generate a visual health report that shows the trend of gait characteristics.

[0046] Based on the type of gait abnormality, targeted rehabilitation training suggestions are provided. For example, single-leg standing balance training is recommended for abnormal center of gravity shift, and gait correction exercises are recommended for uneven stride length. Medical staff can view the elderly's gait data through a cloud-based backend and remotely adjust the rehabilitation training plan, achieving refined management of home-based rehabilitation.

[0047] Although the invention has been described in conjunction with specific features and embodiments, it is apparent that various modifications and combinations can be made therein without departing from the spirit and scope of the invention. The invention is also intended to include such modifications and variations if they fall within the scope of the claims and their equivalents.

Claims

1. A gait analysis-based system for assessing and warning the walking stability of disabled elderly individuals, characterized in that, include: The data acquisition unit is used to collect plantar pressure data, limb posture data, and global motion data of disabled elderly people during their walking process; The gait feature extraction unit preprocesses the data collected by the data acquisition unit and extracts multi-dimensional gait features, including spatiotemporal features, dynamic features, and posture features. The personalized assessment unit compares real-time gait characteristics with the baseline of an elderly person with disabilities, and outputs a walking stability assessment score. The risk warning unit sets multi-level warning thresholds based on the stability assessment score, triggers corresponding level warning prompts, and pushes them to associated terminals.

2. The walking stability assessment and early warning system for disabled elderly based on gait analysis according to claim 1, characterized in that, The data acquisition unit includes a flexible pressure-sensing insole, an inertial measurement module, and an environmental auxiliary acquisition module. The flexible pressure-sensing insole has a built-in distributed pressure sensor array for collecting data on foot pressure distribution, pressure center trajectory, and gait support time. The inertial measurement module integrates an accelerometer, a gyroscope, and a magnetometer, and is worn on the limbs to collect limb movement posture data; The environmental auxiliary acquisition module is a low-power millimeter-wave radar used for contactless acquisition of the elderly's stride length, walking speed, and global trajectory data.

3. The walking stability assessment and early warning system for disabled elderly based on gait analysis according to claim 2, characterized in that, The flexible pressure-sensing insole has a sampling frequency of ≥100Hz and adopts a flexible wearable design to adapt to different foot types; the inertial measurement module is worn at least on the thigh, calf and waist to assist in accurately dividing the gait cycle.

4. The walking stability assessment and early warning system for disabled elderly based on gait analysis according to claim 1, characterized in that, The gait feature extraction unit includes a temporal feature model, which is constructed based on a long short-term memory network and is used to capture the trend features of gait data changing over time. The gait feature extraction unit calculates the pressure center offset rate within the gait cycle using the following formula. Used to quantify the degree of balance fluctuation during walking: ； in: for Coordinates of the pressure center at any given time, in cm; The average coordinates of the pressure center over one gait cycle are given in cm. The average length of the sole of an elderly person's foot is in cm. The pressure center offset rate is dimensionless and ranges from 0 to 1.

5. The walking stability assessment and early warning system for disabled elderly based on gait analysis according to claim 1, characterized in that, The personalized assessment unit includes: The baseline construction module continuously collects gait data from disabled elderly people in the initial stage and constructs a personalized gait health baseline based on their disability level and underlying disease type. The deviation calculation module calculates the characteristic deviation between real-time gait features and the baseline, including the pressure center offset frequency, single-leg support time deviation rate, and limb swing symmetry deviation rate.

6. The walking stability assessment and early warning system for disabled elderly based on gait analysis according to claim 5, characterized in that, The personalized evaluation unit uses an improved random forest algorithm to train a stability evaluation model, and the feature weights are optimized using the Gini index.

7. The walking stability assessment and early warning system for disabled elderly based on gait analysis according to claim 1, characterized in that, The risk warning unit sets three warning thresholds: A Level 1 warning corresponds to a stability score of 60-80, suggesting that the elderly slow down their walking speed. A Level 2 warning corresponds to a stability score of 40-60, triggering a local voice prompt on the device and pushing a warning message to the nursing staff. A Level 3 warning corresponds to a stability score of <40, triggering a local audible and visual alarm and sending the elderly person's location and real-time gait data to the associated monitoring platform.

8. The walking stability assessment and early warning system for disabled elderly based on gait analysis according to claim 7, characterized in that, The risk warning unit includes an adaptive threshold adjustment module, which dynamically adjusts the trigger sensitivity of the warning threshold based on environmental parameters such as ground friction and slope in the area where the elderly walk.

9. The walking stability assessment and early warning system for disabled elderly based on gait analysis according to claim 1, characterized in that, It also includes a data management and feedback unit, which calculates the gait health trend change rate using the following formula. Used to assess changes in long-term walking stability in older adults: ； in: The average stability score for the current cycle is dimensionless. The average stability score for the initial baseline period is dimensionless. This is the rate of change of the trend, dimensionless. A positive sign indicates increased stability, and a negative sign indicates decreased stability.

10. The walking stability assessment and early warning system for disabled elderly based on gait analysis according to any one of claims 1-9, characterized in that, The system adopts an edge computing and cloud collaborative architecture. Data preprocessing and real-time evaluation are completed on local edge devices, while non-real-time health data is uploaded to the cloud for storage and analysis, reducing data transmission latency and privacy leakage risks.

Images