Human body balance ability evaluation method and system based on wearable inertial sensor

Abstract

The invention provides a human body balance ability evaluation method and system based on a wearable inertial sensor, and the method comprises the steps: building a data collection environment, carrying out the data collection, including experiment preparation, sensor wearing and experiment testing, and guaranteeing the obtaining of comprehensive and reliable three-dimensional human body motion data; preprocessing the collected data, extracting characteristic parameters, optimizing the data by adopting adaptive filtering, coordinate system conversion, data fusion and error correction methods, and extracting balance capability related characteristic parameters; combining the extracted characteristic parameters with a standard balance evaluation result, constructing a balance capability evaluation model, and modeling by adopting multivariable regression analysis and a machine learning algorithm; and monitoring and evaluating the balance ability of the human body based on the constructed evaluation model. According to the invention, high-precision and portable balance capability evaluation can be realized in a daily environment.

Description

Technical Field

[0001] The present invention belongs to the field of human body detection, and in particular relates to a method and system for evaluating human balance ability based on a wearable inertial sensor. Background Art

[0002] With the increasing number of people with balance disorders, it is increasingly important to research portable, economical, easy-to-use, and home-use human balance assessment technologies. Existing balance abilities mainly rely on the use of balance testers. Under the guidance of professionals, the tester operates the balance tester according to preset instructions to complete specified actions such as standing and gait testing. The balance tester uses built-in sensors to monitor the tester's balance-related indicators (such as center of gravity changes, pressure distribution, etc.) in real time during the movement, and generates quantitative balance assessment results through a data processing module. This technical solution is widely used in medical institutions and professional rehabilitation scenarios, and can more accurately reflect the tester's balance state.

[0003] Moreover, existing equipment for assessing human balance ability has many limitations in practical applications. The currently widely used balance testers usually rely on pressure sensors to measure the trajectory of the human body's center of gravity and evaluate the body's balance ability by calculating the changes in the center of pressure. However, such equipment is often expensive and only suitable for medical institutions or professional research sites, making it difficult for ordinary users to use in daily life. In addition, these devices are large and inconvenient to carry, which limits their use in home environments. On the other hand, current detection methods are mainly based on the analysis of the two-dimensional center of pressure trajectory and lack the capture of the overall movement posture of the human body, resulting in certain errors in the evaluation results, especially when it comes to dynamic balance analysis in three-dimensional space.

[0004] Currently, mainstream balance assessment methods fall into the following categories. The first category involves traditional balance testers, such as the Tecnobody system, which uses pressure sensors to detect the projected position of the human center of gravity and analyzes its trajectory to assess balance. While this method offers high accuracy, it is expensive and lacks portability. The second category involves wearable device testing, such as the smart bracelet and foot sensor combination proposed by Huawei Technologies Co., Ltd. (CN 202310362009.6). This method primarily assesses a user's static balance and dynamic gait stability, but struggles to accurately measure overall center of mass motion. The third category involves multi-sensor fusion detection systems, such as the wearable posture-sensing belt combined with a load-bearing pressure platform proposed by the Institute of Computing Technology, Chinese Academy of Sciences (CN 202210569569.4). This method simultaneously collects waist posture information and plantar pressure data to improve detection accuracy. However, its reliance on the pressure platform limits its portability. The fourth category is mathematical modeling methods. For example, CN 202410394481.2 and Anhui Aili Intelligent Technology Co., Ltd. use data such as human height and weight to calculate the center of gravity position, and combine pressure center data to evaluate balance ability. However, this method still does not solve the problem of accurately calculating the three-dimensional motion trajectory of the center of mass.

[0005] Therefore, existing balance assessment technologies have the following shortcomings: First, the assessment method relies on a pressure sensing platform, which is expensive and lacks portability, making it unsuitable for daily life. Second, the two-dimensional center of gravity trajectory is calculated based solely on plantar pressure, lacking accurate assessment of three-dimensional center of mass motion, resulting in large errors in dynamic testing. Finally, existing assessment models fail to fully utilize inertial data, resulting in a lack of accuracy and individualized adaptability in the assessment results. Summary of the Invention

[0006] In order to solve the problems of insufficient versatility, high cost and poor portability in the existing balance testers for evaluating human balance ability, the present invention provides a method and system for evaluating human balance ability based on wearable inertial sensors, aiming to achieve more accurate balance ability evaluation by introducing three-dimensional inertial sensing data and machine learning technology. First, in order to improve the portability of the equipment, the present invention adopts wearable inertial sensors, so that users can monitor their balance ability anytime and anywhere in their daily lives without relying on large equipment. Secondly, in order to solve the problem that traditional methods are only based on two-dimensional data analysis, the present invention not only calculates the two-dimensional center of gravity trajectory of the human body, but also combines three-dimensional posture data to fully reflect the dynamic balance state of the human body. In addition, the present invention uses machine learning algorithms to extract features and model the collected data to improve the accuracy and robustness of the evaluation. Finally, this method can be applied to multiple scenarios such as home health monitoring, rehabilitation training, and sports evaluation for the elderly to achieve efficient and accurate balance ability evaluation. This system uses a wearable inertial measurement unit (IMU) to assess human balance in three dimensions. The inertial measurement unit, worn around the waist, collects motion data in real time and, combined with body parameters such as height and leg length, calculates the three-dimensional trajectory of the body's center of mass, enabling more accurate assessment of static and dynamic balance. Furthermore, the system utilizes a machine learning algorithm to optimize the assessment model, eliminating the need for a plantar pressure sensing platform and enabling highly accurate, portable balance assessment in everyday settings.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] A method for evaluating human balance ability based on a wearable inertial sensor comprises the following steps:

[0009] Step 1: Set up the data collection environment and conduct data collection, including experimental preparation, sensor wearing and experimental testing, to ensure the acquisition of comprehensive and reliable 3D human motion data;

[0010] Step 2: Preprocess the collected data and extract characteristic parameters. Adaptive filtering, coordinate system transformation, data fusion and error correction methods are used to optimize the data and extract characteristic parameters related to the balance ability, including swing area, center of mass offset, tilt angle, swing speed, and center of mass swing frequency.

[0011] Step 3: Combine the extracted characteristic parameters with the standard balance assessment results to construct a balance ability assessment model, using multivariate regression analysis and machine learning algorithms for modeling;

[0012] Step 4: Based on the constructed evaluation model, the human body's balance ability is monitored and evaluated.

[0013] Furthermore, in step 1, the experimental preparation stage includes screening subjects according to the inclusion and exclusion criteria, recording their basic information, and introducing the experimental precautions and procedures to the subjects; the sensor wearing stage includes calculating the subject's center of mass position and wearing the inertial sensor, and checking the equipment status; the experimental testing stage includes the subject completing a variety of standard balance test movements, the inertial sensor collecting data in real time, and the researchers monitoring and handling illegal movements.

[0014] Furthermore, in step 2, the adaptive filtering and data fusion dynamically adjusts the filtering parameters according to the motion patterns of different subjects, and fuses the data of the three-axis accelerometer and the three-axis gyroscope to calculate the fused acceleration; the coordinate system conversion and trajectory calculation uses the quaternion coordinate system conversion method to convert the data to the global coordinate system, and calculates the motion trajectory of the human body center of mass through the double integration method; the error correction uses the zero-speed update algorithm for error correction.

[0015] Furthermore, in step 2, the extracted balance ability-related characteristic parameters include:

[0016] sway area, reflecting stability when standing;

[0017] a shift in the center of mass, indicating the degree of impairment of balance;

[0018] Lean angle, which describes the ability to adjust posture while maintaining balance;

[0019] Swing speed, a measure of the sensitivity and coordination of balance adjustments;

[0020] The center of mass oscillation frequency reflects the control stability when maintaining balance.

[0021] Furthermore, in step 3, a five-fold cross-validation method is used to comprehensively evaluate the balance ability assessment model, and the evaluation indicators include mean square error and determination coefficient; if the model evaluation result does not meet the expected accuracy requirements, optimization is performed, including feature optimization and fitting method optimization.

[0022] Furthermore, in step 4, mobile phone software is developed to realize functions such as user information entry, balance data collection, balance feature display and evaluation result description; at the same time, a management website is developed, the back end is built, and a database is selected for data storage, and the long-term operation and maintenance of the system are realized with the help of Tencent Cloud Server.

[0023] Furthermore, the wearable inertial sensor includes a three-axis accelerometer, a three-axis gyroscope and a three-axis magnetometer, which is worn on the human waist and back and is used to collect real-time motion trajectory data of the human body's center of mass in three-dimensional space.

[0024] The present invention also provides a human balance ability assessment system based on a wearable inertial sensor, comprising the following modules:

[0025] The data acquisition module builds the data acquisition environment and performs data acquisition, including experimental preparation, sensor wearing, and experimental testing, to ensure the acquisition of comprehensive and reliable three-dimensional human motion data;

[0026] The parameter extraction module pre-processes the collected data and extracts characteristic parameters. It uses adaptive filtering, coordinate system transformation, data fusion and error correction methods to optimize the data and extract characteristic parameters related to the balance ability, including swing area, center of mass offset, tilt angle, swing speed, and center of mass swing frequency.

[0027] The modeling module combines the extracted characteristic parameters with the standard balance assessment results to construct a balance ability assessment model, using multivariate regression analysis and machine learning algorithms for modeling;

[0028] The evaluation module monitors and evaluates the human body's balance ability based on the constructed evaluation model.

[0029] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the above-mentioned method for evaluating human balance ability based on wearable inertial sensors are implemented.

[0030] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-mentioned method for evaluating human balance ability based on wearable inertial sensors.

[0031] Beneficial effects:

[0032] 1. High portability. This invention uses a wearable inertial sensor, which is compact and easy to wear and use. To more accurately assess a person's balance ability, the sensor can be worn on the user's back, allowing them to use the sensor at home without having to visit a specialized medical institution and obtain assessment results at any time.

[0033] 2. Affordable. This device is less expensive than balance testers, and a single sensor can complete the assessment task. This makes balance assessment more affordable and particularly suitable for those with balance impairments who require long-term monitoring, thereby reducing medical costs and increasing accessibility.

[0034] 3. Simple operation. The use of the present invention is very simple. Users only need to wear the sensor on their waist and back and achieve the test effect according to normal daily movements without the need for complicated testing procedures. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1This is a flow chart of a method for evaluating human balance ability based on a wearable inertial sensor according to the present invention;

[0036] Figure 2 This is the data collection flow chart;

[0037] Figure 3 Data preprocessing is the three-dimensional trajectory calculation graph. DETAILED DESCRIPTION

[0038] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only intended to illustrate the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0039] like Figure 1 As shown, a method for evaluating human balance ability based on a wearable inertial sensor of the present invention comprises the following steps:

[0040] Step 1: Build a data collection environment and collect data;

[0041] Step 2: preprocess the data and extract feature parameters;

[0042] Step 3: Combine the extracted characteristic parameters with the Tecnobody balance assessment results to construct a balance ability assessment model;

[0043] Step 4: Monitor and evaluate balance ability.

[0044] Specifically, if Figure 2 As shown, the step 1 includes:

[0045] The data collection process includes three stages: experimental preparation, sensor wearing, and experimental testing to ensure the acquisition of comprehensive and reliable three-dimensional human motion data.

[0046] During the experimental preparation phase, subjects were screened based on the inclusion and exclusion criteria (inclusion and exclusion criteria) developed by the research team, ultimately identifying those who met the experimental requirements. All subjects were required to sign an informed consent form, and researchers then recorded their basic information, including name, age, gender, height, weight, leg length, contact information, and medical history. Before the actual test, researchers provided subjects with detailed instructions and the complete experimental procedures to ensure a smooth progress.

[0047] During the sensor fitting phase, researchers instructed subjects to stand on the balance test system and ensured the stability of the testing environment. Based on this, the subject's center of mass was calculated based on key parameters such as height and leg length, and an inertial sensor was placed at this location. The inertial sensor utilizes a three-axis accelerometer, a three-axis gyroscope, and a three-axis magnetometer to accurately capture the motion trajectory of the subject's center of mass in three-dimensional space. After fitting, researchers further checked the status of all devices to ensure the accuracy and consistency of data collection.

[0048] During the experimental testing phase, the subjects completed the following experiments in sequence according to the preset standard balance test process. Each experiment was repeated three times, and the average value was taken to improve data reliability: standing on both feet with eyes open for 30 seconds, standing on both feet with eyes closed for 30 seconds, standing on one leg or each leg with eyes closed for 10 seconds, standing on one leg or each leg with eyes open for 10 seconds, stretching both hands out horizontally for 10 seconds, and walking 5 meters back and forth. During the test, the inertial sensor collected three-axis acceleration, three-axis angular velocity, and three-axis magnetic field intensity data in real time to ensure that the movement changes of the subject's center of mass in three-dimensional space could be captured, and their balance ability could be comprehensively analyzed. The researchers monitored the subjects' movements throughout the process to ensure the safety of the experiment. On the one hand, if the subjects made any illegal movements (such as excessive shaking or ending the test early), the test data would be invalidated and not included in the analysis.

[0049] Specifically, the data preprocessing in step 2 is a key step in ensuring the accuracy of human balance assessment. The present invention uses methods such as adaptive filtering, coordinate system transformation, data fusion, and error correction to optimize the raw data collected by the inertial sensor, ensuring more accurate calculation of the motion trajectory of the human center of mass in three-dimensional space.

[0050] like Figure 3 As shown in the figure, the raw sensor data is first filtered using a Butterworth low-pass filter to effectively remove high-frequency noise. Subsequently, a rotation matrix is calculated based on the quaternion data output by the sensor. This rotation matrix is used to transform the filtered data, converting the sensor coordinate system to the global coordinate system. Gravity compensation is then performed, eliminating the gravity component to obtain the true motion acceleration signal.

[0051] Next, a double integration process is performed on the acceleration data after gravity compensation: first, the offset of the z-axis acceleration is eliminated and the velocity integration is performed, and the zero-bias calibration method is applied to reduce the impact of the cumulative error; then, the displacement integration is performed based on the velocity integration result to obtain the original three-dimensional motion trajectory.

[0052] Finally, the error generated in the integration process is further corrected by the Zero Velocity Update (ZUPT) technique to obtain the calibrated three-dimensional trajectory.

[0053] Specifically, data preprocessing includes:

[0054] (1) Adaptive filtering and data fusion: Due to the large individual differences among subjects, the data collected by the inertial sensor may contain noise and measurement errors. Therefore, the system uses an adaptive filter to dynamically adjust the filter parameters according to the movement patterns of different subjects, achieving personalized signal processing and improving data stability and accuracy.

[0055] To more accurately reflect the motion of the human body's center of mass in three-dimensional space, the present invention employs an acceleration and angular velocity fusion algorithm. Specifically, by fusing data from a three-axis accelerometer and a three-axis gyroscope, a fused acceleration is calculated, thereby reducing the potential bias that could arise from single-sensor data.

[0056] (2) Coordinate system conversion and trajectory calculation: Since the data measured by the inertial sensor is based on the sensor coordinate system, it needs to be converted to the global coordinate system to uniformly analyze the human body's motion state. This invention adopts the quaternion coordinate system conversion method to avoid the problem of universal lock and ensure the stability and accuracy of the three-dimensional rotation transformation.

[0057] ;

[0058] in, is the three-axis acceleration measured by the inertial sensor; is the rotation matrix obtained from the inertial sensor quaternion; is the acceleration after fusion; is the acceleration due to gravity.

[0059] After the coordinate system transformation is complete, the double integration method is used to calculate the trajectory of the human center of mass. By integrating the fused acceleration twice, the real-time trajectory of the center of mass in three-dimensional space is obtained, providing data support for subsequent balance analysis.

[0060] ;

[0061] in, is the three-dimensional position of the center of mass in the global coordinate system; is the center of mass at time The initial three-dimensional position of is the center of mass at time The three-dimensional velocity vector of is the three-dimensional acceleration of the center of mass in the global coordinate system; is the initial time; is the current time; The time parameter that serves as the integration variable.

[0062] (3) Error Correction: Because the double integration method may lead to cumulative errors (drift errors), this paper introduces a Zero Velocity Update (ZUPT) algorithm for error correction. When the subject is stationary, the ZUPT algorithm automatically identifies the zero velocity interval and adjusts the calculated three-dimensional motion trajectory to reduce the impact of cumulative errors on the evaluation results and ensure the accuracy of the human center of mass trajectory.

[0063] Specifically, the extraction of characteristic parameters related to balance ability in step 2 involves extracting balance assessment indicators that effectively characterize balance ability from the three-dimensional human center of mass motion trajectory. These features include sway area, center of mass offset, tilt angle, sway speed, and center of mass swing frequency, which are directly related to various aspects of human balance ability. The present invention utilizes a Long Short-Term Memory (LSTM) network for motion feature extraction. Leveraging its strong memory capacity when processing time series data, the LSTM network can effectively model the event dependencies of the human body in different equilibrium states and extract key features that effectively reflect balance ability.

[0064] This paper analyzes the three-dimensional trajectory of the human center of mass to extract a series of key evaluation indicators that accurately characterize the human body's balance ability. These characteristics can comprehensively reflect the individual's stability, control ability, and adjustment strategy in different balance states, and provide a scientific basis for the quantitative assessment of balance ability. The following key features are extracted from the three-dimensional center of mass trajectory:

[0065] Sway Area: This refers to the range of oscillation of the body's center of mass in three-dimensional space, reflecting an individual's stability when standing. A larger sway area generally indicates poorer balance.

[0066] ;

[0067] Where A is the sway area of the center of mass in the horizontal plane; , is the horizontal coordinate at the i-th moment; N is the total number of sampling points.

[0068] Center of Mass Displacement: This refers to the displacement of the center of mass of the human body relative to the static reference position. Large displacements can indicate that the individual's balance ability is impaired.

[0069] ;

[0070] Where d is the maximum offset distance of the center of mass; , are the maximum and minimum values of the center of mass trajectory in the x direction; , are the maximum and minimum values of the center of mass trajectory in the y direction; , are the maximum and minimum values of the center of mass trajectory in the z direction.

[0071] Inclination Angle: Describes the degree of inclination of the center of mass in different directions, reflecting the body's ability to adjust posture while maintaining balance.

[0072] ;

[0073] Where θ is the maximum tilt angle of the center of mass.

[0074] Sway Velocity: Calculates the displacement speed of the center of mass per unit time, which can measure the sensitivity and coordination of individual balance adjustments.

[0075] ;

[0076] in, is the average sway velocity of the center of mass; , , is the position coordinate of the center of mass at the i-th moment; 、 、 is the velocity component; N is the total number of sampling points.

[0077] Sway Frequency: Indicates the oscillation frequency of the center of mass movement. A higher sway frequency may mean that the individual's control instability when maintaining balance.

[0078] ;

[0079] in, is the maximum oscillation frequency of the center of mass; To use Fast Fourier Transform (FFT), the time domain signal is converted into a frequency domain signal X(f).

[0080] Specifically, the feature extraction method in step 2 includes:

[0081] To ensure efficient, discriminative, and universally applicable extracted center-of-mass motion features, this paper employs a feature screening strategy that integrates multidimensional feature modeling with a random forest algorithm. First, multiple features, including those from the time domain, frequency domain, and nonlinearity, are extracted from the three-dimensional center-of-mass trajectory. These include, but are not limited to, time-domain metrics such as mean, standard deviation, root mean square (RMS), center-of-mass offset path length, and sway area; frequency-domain features such as dominant frequency, spectral energy, and spectral entropy; and nonlinear dynamic features such as approximate entropy, sample entropy, and fractal dimension. These multidimensional features comprehensively reflect the subject's stability and motion control ability under static or dynamic conditions.

[0082] The present invention adopts random forest algorithm to perform regression modeling on the above-mentioned feature set, takes the balance ability score output by TecnobodyPK254 standard equipment as the target variable, and calculates the importance score of each feature in the prediction task during the training process. According to the feature importance ranking results, several key features most relevant to the balance ability are screened out, redundant features and noise interference are eliminated, the feature dimension is effectively compressed, and the modeling efficiency is improved. TecnobodyPK254 is a well-known device in the field. The device has been publicly sold through commercial channels and has been widely used in medical institutions. The technical parameters, usage methods and application of the device in balance function assessment have been clearly recorded in existing technical documents and research papers, such as "Study on the efficacy of balance testing and training system for proprioception rehabilitation after arthroscopic meniscus repair" and "Nursing study on the effect of balance intervention on strengthening joint control and preventing falls in the elderly".

[0083] After identifying key features, the present invention further introduces standardization (such as Z-score normalization) to reduce the impact of inter-individual signal distribution differences on model performance and enhance the model's generalization across diverse populations. Compared to traditional rule-of-thumbnail feature construction methods, the random forest-based importance assessment method employed in this invention is more objective and interpretable, clarifying the causal relationship between various motor features and balance ability, thereby providing a reliable quantitative basis for personalized intervention.

[0084] In summary, the present invention introduces a machine learning algorithm as a selection mechanism in the feature extraction stage. Through structured feature screening and normalization processing, it achieves efficient extraction and optimization of the characterization features of human balance ability, providing a solid data foundation for subsequent modeling and scoring.

[0085] Specifically, the construction of the balance ability assessment model in step 3 includes:

[0086] This invention constructs a human balance assessment model by fitting the extracted three-dimensional center of mass motion features with the Tecnobody balance assessment results (obtained by the subject wearing the sensor as required while standing on the test platform of the Tecnobody PK254 standard device and completing the specified movements. The device's built-in computer monitor displays the balance assessment results). This creates an accurate mathematical model and enables individualized balance assessment prediction. The balance assessment model is constructed using multivariate regression analysis and machine learning algorithms (such as support vector machines, random forests, or neural networks). The three-dimensional center of mass motion feature set (such as sway area, center of mass offset, tilt angle, sway speed, and sway frequency) is used as input variables. The Tecnobody balance assessment results serve as labeled data for supervised learning, ensuring that the model learns the true balance assessment criteria.

[0087] (1) Balance ability assessment model evaluation:

[0088] To ensure the stability and prediction accuracy of the human balance ability assessment model, this paper uses a K-fold cross-validation method to comprehensively evaluate the model. K-fold cross-validation divides the dataset into five subsets, using four of these subsets for model training and the remaining subset for testing. This is repeated five times and the average value is taken to ensure the model's generalization and robustness. The evaluation metrics used are the mean squared error (MSE) and the coefficient of determination (R² score). The mean squared error measures the average error between the model's predicted and actual values; smaller MSE values indicate higher prediction accuracy. The coefficient of determination assesses the model's goodness of fit. An R² value close to 1 indicates a good fit to the data and a high degree of accuracy in reflecting human balance characteristics. Using these evaluation methods and metrics, we determine whether the model meets the expected accuracy requirements and provide a quantitative basis for subsequent model optimization, ensuring that the final model accurately predicts an individual's balance ability and improving the system's practicality and reliability.

[0089] (2) Optimization of balance ability assessment model:

[0090] If the model's evaluation results do not meet the expected accuracy requirements, the present invention will optimize the model based on the evaluation data of the Tecnobody PK254 standard device to improve the model's predictive ability and stability. The optimization process mainly includes feature optimization and fitting optimization, aiming to improve the model's generalization ability and make it suitable for balance assessment in different individuals.

[0091] In terms of feature optimization, we first conduct an importance analysis on the extracted three-dimensional center of mass motion features, screening for features that contribute significantly to balance ability prediction and removing redundant or less relevant features to reduce data noise. To improve feature representation, we employ methods such as feature scaling and dimensionality reduction (such as principal component analysis (PCA)) to enable the model to more effectively learn the characteristic patterns of human balance ability.

[0092] In terms of fitting optimization, hyperparameters (such as learning rate and regularization coefficient) are adjusted based on the model evaluation results. Parameter optimization is performed for different learning algorithms (such as support vector machines, random forests, or neural networks) to improve the model's fitting ability. To further enhance prediction accuracy, appropriate regression methods (such as linear regression, nonlinear regression, or deep learning models) can be selected based on the data characteristics to ensure that the model can more accurately map the relationship between three-dimensional motion data and human balance ability. If the performance of a single model is still unsatisfactory, ensemble learning methods (such as gradient boosting trees and random forests) can be introduced to improve the robustness and generalization ability of the overall model by combining the prediction results of multiple models.

[0093] Through the above optimization measures, the present invention can ensure that the evaluation model has higher prediction accuracy and stability, so that it can provide reliable balance ability evaluation results in different application scenarios.

[0094] Specifically, Step 4 involves developing mobile software that will provide a simple user interface, supporting functions such as user information entry, balance data collection, balance feature display, and assessment result interpretation. Furthermore, a management website will be developed, a backend will be constructed, and a database will be selected for data storage, with Tencent Cloud servers providing long-term system operation and maintenance.

[0095] The present invention also provides a human balance ability assessment system based on a wearable inertial sensor, comprising the following modules:

[0096] The data acquisition module builds the data acquisition environment and performs data acquisition, including experimental preparation, sensor wearing, and experimental testing, to ensure the acquisition of comprehensive and reliable three-dimensional human motion data;

[0097] The parameter extraction module pre-processes the collected data and extracts characteristic parameters. It uses adaptive filtering, coordinate system transformation, data fusion and error correction methods to optimize the data and extract characteristic parameters related to the balance ability, including swing area, center of mass offset, tilt angle, swing speed, and center of mass swing frequency.

[0098] The modeling module combines the extracted characteristic parameters with the standard balance assessment results to construct a balance ability assessment model, using multivariate regression analysis and machine learning algorithms for modeling;

[0099] The evaluation module monitors and evaluates the human body's balance ability based on the constructed evaluation model.

[0100] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the above-mentioned method for evaluating human balance ability based on wearable inertial sensors are implemented.

[0101] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-mentioned method for evaluating human balance ability based on wearable inertial sensors.

[0102] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk drives, CD-ROMs, optical storage devices, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention may be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0103] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0104] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0105] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0106] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0107] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A method for evaluating human balance ability based on wearable inertial sensors, characterized in that: The following steps are involved: Step 1: Set up the data collection environment and conduct data collection, including experimental preparation, sensor wearing and experimental testing, to ensure the acquisition of comprehensive and reliable 3D human motion data; Step 2: Preprocess the collected data and extract characteristic parameters. Adaptive filtering, coordinate system transformation, data fusion and error correction methods are used to optimize the data and extract characteristic parameters related to the balance ability, including swing area, center of mass offset, tilt angle, swing speed, and center of mass swing frequency. Step 3: Combine the extracted characteristic parameters with the standard balance assessment results to construct a balance ability assessment model, using multivariate regression analysis and machine learning algorithms for modeling; Step 4: Based on the constructed evaluation model, the human body's balance ability is monitored and evaluated.

2. The method for evaluating human balance ability based on wearable inertial sensors according to claim 1, wherein: In step 1, the experimental preparation phase includes screening subjects according to the inclusion criteria, recording their basic information, and introducing the experimental precautions and procedures to the subjects; the sensor wearing phase includes calculating the subject's center of mass position and wearing the inertial sensor, and checking the equipment status; the experimental testing phase includes the subject completing a variety of standard balance test movements, the inertial sensor collecting data in real time, and the researchers monitoring and handling illegal movements.

3. The method for evaluating human balance ability based on wearable inertial sensors according to claim 1, wherein: In step 2, adaptive filtering and data fusion dynamically adjust the filtering parameters according to the motion patterns of different subjects, and fuse the data of the three-axis accelerometer and the three-axis gyroscope to calculate the fused acceleration; the coordinate system conversion and trajectory calculation use the quaternion coordinate system conversion method to convert the data to the global coordinate system, and calculate the human body center of mass motion trajectory through the double integration method; error correction uses the zero-speed update algorithm for error correction.

4. The method for evaluating human balance ability based on wearable inertial sensors according to claim 1, wherein: In step 2, the extracted balance ability-related characteristic parameters include: sway area, reflecting stability when standing; a shift in the center of mass, indicating the degree of impairment of balance; Lean angle, which describes the ability to adjust posture while maintaining balance; Swing speed, a measure of the sensitivity and coordination of balance adjustments; The center of mass oscillation frequency reflects the control stability when maintaining balance.

5. The method for evaluating human balance ability based on wearable inertial sensors according to claim 1, wherein: In step 3, a five-fold cross-validation method is used to comprehensively evaluate the balance ability assessment model, and the evaluation indicators include mean square error and determination coefficient; if the model evaluation result does not meet the expected accuracy requirements, optimization is performed, including feature optimization and fitting method optimization.

6. The method for evaluating human balance ability based on wearable inertial sensors according to claim 1, wherein: In step 4, mobile phone software is developed to realize functions such as user information input, balance data collection, balance feature display and evaluation result description; at the same time, a management website is developed, a backend is built, a database is selected for data storage, and the long-term operation and maintenance of the system are realized with the help of Tencent Cloud Server.

7. The method for evaluating human balance ability based on wearable inertial sensors according to claim 1, wherein: The wearable inertial sensor includes a three-axis accelerometer, a three-axis gyroscope and a three-axis magnetometer, which is worn on the human waist and back and is used to collect real-time motion trajectory data of the human body's center of mass in three-dimensional space.

8. A human balance ability assessment system based on wearable inertial sensors, characterized in that: Includes the following modules: The data acquisition module builds the data acquisition environment and performs data acquisition, including experimental preparation, sensor wearing, and experimental testing, to ensure the acquisition of comprehensive and reliable three-dimensional human motion data; The parameter extraction module pre-processes the collected data and extracts characteristic parameters. It uses adaptive filtering, coordinate system transformation, data fusion and error correction methods to optimize the data and extract characteristic parameters related to the balance ability, including swing area, center of mass offset, tilt angle, swing speed, and center of mass swing frequency. The modeling module combines the extracted characteristic parameters with the standard balance assessment results to construct a balance ability assessment model, using multivariate regression analysis and machine learning algorithms for modeling; The evaluation module monitors and evaluates the human body's balance ability based on the constructed evaluation model.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the method for evaluating human balance ability based on a wearable inertial sensor are implemented as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the computer program implements the steps of the method for evaluating human balance ability based on a wearable inertial sensor as claimed in any one of claims 1 to 7.

Images