An adaptive balance rehabilitation method based on multi-dimensional physiological characteristic weighted matching

Abstract

The application provides a kind of adaptive balance rehabilitation method based on multi-dimensional physiological characteristic weighted matching, comprising: obtaining the initial evaluation score of the balance function physiological dimension of the trainee;According to the preset basic weight coefficient of each physiological dimension and the initial evaluation score, the individual physiological dimension weight coefficient of the trainee is calculated;Based on the individual physiological dimension weight coefficient, the weighted matching algorithm is used to perform feature matching screening in the pre-established parameterized rehabilitation training task library, and an individual rehabilitation training scheme is generated;During the execution of the individual rehabilitation training scheme by the trainee, the difficulty level of the corresponding target training task is adjusted according to the trainee's training behavior data, and the scheme adjustment prompt is output synchronously.The method realizes the precise matching and adjustment of balance function rehabilitation training requirements, is suitable for rehabilitation scenes without professional guidance such as family and community, and can improve rehabilitation efficiency and safety.

Description

Technical Field

[0001] This invention relates to the field of rehabilitation medicine, and in particular to an adaptive balance rehabilitation method based on weighted matching of multidimensional physiological characteristics. Background Technology

[0002] Balance function refers to the basic physiological function of maintaining postural stability under static or dynamic conditions. Balance dysfunction can be caused by a variety of factors, including stroke, Parkinson's disease, sports injuries, muscle damage after orthopedic surgery, and natural functional decline in old age. Affected individuals often exhibit symptoms such as unsteadiness while standing, gait imbalance, and poor motor coordination. This not only seriously affects patients' daily activities but is also a leading risk factor for falls and injuries, placing a heavy burden on patients' families and the social healthcare system.

[0003] Currently, in clinical and rehabilitation settings, rehabilitation training for balance disorders primarily employs two types of programs: standardized rehabilitation training programs and personalized programs subjectively developed and adjusted by rehabilitation physicians. The former fails to adequately consider the differences in physiological function across multiple dimensions, including vision, vestibular sensation, proprioception, and posture control systems, making it difficult to accurately match patients' core functional weaknesses and achieve targeted reinforcement training for these weaker dimensions. This often leads to a mismatch between training content and patient needs. The latter heavily relies on the individual clinical experience of rehabilitation physicians, lacking unified, objective quantitative matching rules and data support, resulting in significant subjectivity and uncertainty in program development and adjustment. Furthermore, in scenarios such as home rehabilitation and community rehabilitation where professional rehabilitation therapists provide full guidance, patients and their families cannot independently develop and dynamically adjust personalized training programs, easily leading to inappropriate training and poor rehabilitation outcomes.

[0004] Therefore, there is an urgent need to develop a balance rehabilitation method that can automatically match targeted training tasks based on the patient's multidimensional physiological functional defects, generate personalized balance function rehabilitation training programs, and dynamically adjust the training difficulty according to the patient's training performance to achieve adaptive optimization of the training program. Summary of the Invention

[0005] The purpose of this invention is to provide a balance rehabilitation method that can automatically match targeted training tasks based on the patient's multi-dimensional physiological functional defects, generate personalized balance function rehabilitation training programs, and dynamically adjust the training difficulty according to the patient's training performance to achieve adaptive optimization of the training program.

[0006] To achieve this goal, the technical solution adopted in this experiment is an adaptive balance rehabilitation method based on multidimensional physiological feature weighted matching, which includes the following steps: S1: Obtain the initial assessment score of the balance function physiological dimension of the trainee. The initial assessment score of the physiological dimension is obtained by collecting objective kinematic data through external sensors or by input by the assessor. S2: Calculate the personalized physiological dimension weight coefficients of the trainee based on the preset basic weight coefficients of each physiological dimension and the initial evaluation score; S3: Based on the personalized physiological dimension weight coefficient, a weighted matching algorithm is used to perform feature matching and screening in a pre-established parametric rehabilitation training task library to generate a personalized rehabilitation training plan containing at least one target training task; wherein, the parametric rehabilitation training task library stores multiple training tasks, and each training task is configured with a corresponding quantitative value of training effect and difficulty level. S4: During the process of the trainee executing the personalized rehabilitation training program, continuously acquire the trainee's training behavior data, and adaptively adjust the difficulty level of the corresponding target training task according to the training behavior data, and output program adjustment prompts simultaneously.

[0007] The physiological dimensions include visual dimension A, vestibular sensory dimension B, proprioceptive dimension C, and posture control strategy dimension D; the corresponding basic weight coefficients are WA, WB, WC, and WD, respectively, and satisfy WA+WB+WC+WD=1.

[0008] The calculation process for the personalized physiological dimension weighting coefficients of the trainees is as follows: Obtain the initial assessment scores SA, SB, SC, and SD for each of the aforementioned physiological dimensions, with a maximum score of 100. Calculate the weighted defect score S' for each physiological dimension: SA′=(100−SA)×WA SB′=(100−SB)×WB SC′=(100−SC)×WC SD′=(100−SD)×WD; The weighted defect scores are linearly normalized to obtain the personalized physiological dimension weight coefficients KA, KB, KC, and KD, respectively. The calculation methods are as follows: KA= KB= KC= KD= .

[0009] The rehabilitation training task library includes single-dimensional training tasks targeting visual dimension A, vestibular sensory dimension B, proprioceptive dimension C, and postural control strategy dimension D, as well as composite training tasks that have training effects on multiple physiological dimensions; each training task has corresponding training effect evaluation values ​​RA, RB, RC, and RD for the four physiological dimensions.

[0010] Based on the aforementioned personalized physiological dimension weight coefficients, a weighted matching algorithm is further used for feature matching and screening to generate a personalized rehabilitation training plan, specifically including: The matching degree M between each training task in the parameterized rehabilitation training task library and the trainee is calculated using the formula: M = RAKA + RBKB + RCKC + RDKD; Sort all the training tasks in descending order of matching degree M; Select training tasks that are ranked first by a preset proportion or whose matching degree is greater than a preset threshold to form a candidate task set; In the candidate task set, a preset number of single-dimensional training tasks and / or compound training tasks are selected in descending order of matching degree to form the final personalized rehabilitation training program.

[0011] The training task is configured with N difficulty levels, where N is a natural number greater than 1.

[0012] The different difficulty levels of the training tasks are achieved by adjusting the corresponding training parameters. All tasks have a preset difficulty-task execution parameter correspondence table. The training parameters include, but are not limited to, physical environment parameters, kinematic instruction parameters, and multimedia stimulus intensity parameters. The difficulty adjustment methods for single-dimensional training tasks are as follows: visual training tasks adjust the difficulty by adjusting the intensity of hardware visual stimulation; vestibular sensory training tasks adjust the difficulty by adjusting the amplitude, speed, and frequency of head movements; proprioceptive tasks adjust the difficulty by adjusting the physical parameters of the support surface; posture control strategy training tasks adjust the difficulty by adjusting the training duration, the amplitude of posture changes, and the intensity of disturbances; and composite training tasks involve multiple physiological dimensions, and their difficulty is achieved by adjusting the corresponding training parameters proportionally according to the training weight of each dimension in the composite training task.

[0013] The training behavior data adaptively adjusts the difficulty level of the corresponding target training task, specifically including: Obtain the preset scoring rules for each target training task and the corresponding score thresholds for each difficulty level; Calculate the average score of the trainee during a preset time period or a preset number of consecutive training sessions; If the average score is higher than the upper limit threshold of the current difficulty level, the difficulty level of the target training task will be increased by one level. If the average score is lower than the lower limit threshold of the current difficulty level, the difficulty level of the target training task will be lowered by one level; the difficulty level adjustment will be carried out step by step within the range of the N levels.

[0014] The present invention has the following beneficial effects: This invention can achieve targeted training of patients' vision, vestibular sensation, proprioception, and posture control systems based on the personalized defect characteristics of patients' multidimensional physiological functions. It can accurately match the individual rehabilitation needs of different patients and significantly improve rehabilitation efficiency. This invention can dynamically adjust the training difficulty based on the patient's training performance, achieving adaptive optimization of the training content. It can scientifically advance the training progress while ensuring the patient's training confidence and safety. This invention can provide standardized and practical rehabilitation training guidance for scenarios such as families and grassroots communities that lack professional guidance, without the need for full-time intervention and guidance from rehabilitation physicians. Attached Figure Description

[0015] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0016] The invention will now be described in further detail with reference to experimental examples.

[0017] To address the problem that existing balance rehabilitation training programs lack personalization and rely on manual customization by rehabilitation physicians, this invention provides an adaptive balance rehabilitation method based on weighted matching of multidimensional physiological characteristics. This method includes the following steps: S1: Obtain the initial assessment score of the balance function physiological dimension of the trainee. The initial assessment score of the physiological dimension is obtained by collecting objective kinematic data through external sensors or by input by the assessor. S2: Calculate the personalized physiological dimension weight coefficients of the trainee based on the preset basic weight coefficients of each physiological dimension and the initial evaluation score; S3: Based on the personalized physiological dimension weight coefficient, a weighted matching algorithm is used to perform feature matching and screening in a pre-established parametric rehabilitation training task library to generate a personalized rehabilitation training plan containing at least one target training task; wherein, the parametric rehabilitation training task library stores multiple training tasks, and each training task is configured with a corresponding quantitative value of training effect and difficulty level. S4: During the process of the trainee executing the personalized rehabilitation training program, continuously acquire the trainee's training behavior data, and adaptively adjust the difficulty level of the corresponding target training task according to the training behavior data, and output program adjustment prompts simultaneously.

[0018] S1 aims to obtain initial balance function scores for trainees across various physiological dimensions, providing foundational data for subsequent personalized weighting coefficient calculations. In some experimental cases, the initial assessment scores for these physiological dimensions are obtained by collecting objective kinematic data from external sensors and normalizing it to a score range of 0-100; in other experimental cases, the initial assessment scores for these physiological dimensions are manually entered by a professional assessor.

[0019] The physiological dimensions include visual dimension A, vestibular sensory dimension B, proprioceptive dimension C, and posture control strategy dimension D. The scores for each dimension are in the range of 0-100. The initial evaluation scores of the four dimensions A, B, C, and D are recorded as SA, SB, SC, and SD, respectively.

[0020] S2 aims to calculate the personalized physiological dimension weight coefficients of trainees based on initial assessment scores and basic weight coefficients, providing a core matching basis for weighted matching of training tasks. The specific calculation process is as follows: First, calculate the weighted defect score S' for each physiological dimension: SA′=(100−SA)×WA SB′=(100−SB)×WB SC′=(100−SC)×WC SD′=(100−SD)×WD; Secondly, the weighted defect scores are linearly normalized to obtain the personalized physiological dimension weight coefficients KA, KB, KC, and KD, respectively. The calculation methods are as follows: KA= KB= KC= KD= .

[0021] Wherein, WA, WB, WC, and WD are the basic weight coefficients of visual dimension A, vestibular sensory dimension B, proprioceptive dimension C, and posture control strategy dimension D in maintaining balance function, respectively. The basic weight coefficients are determined by clinical balance function-related research data and satisfy WA+WB+WC+WD=1.

[0022] S3 aims to generate a personalized rehabilitation training program containing at least one target training task by using a weighted matching algorithm to perform feature matching and screening in a pre-established parametric rehabilitation training task library based on the patient's personalized physiological dimension weight coefficients.

[0023] The parameterized rehabilitation training task library stores multiple training tasks, each configured with a corresponding quantitative value for training effect and a difficulty level. The quantitative values ​​for the training effect of each task in the visual dimension A, vestibular sensory dimension B, proprioceptive dimension C, and posture control system dimension D are denoted as RA, RB, RC, and RD, respectively. The quantitative value R reflects the training effect of the task on the balance function of each dimension.

[0024] The training task is configured with N difficulty levels, where N is a natural number greater than 1.

[0025] The process of feature matching and selection in the parameterized rehabilitation training task library using a weighted matching algorithm is as follows: The first step is to calculate the matching degree M between each training task in the parameterized rehabilitation training task library and the trainee. The calculation formula is: M = RAKA + RBKB + RCKC + RDKD; The second step is to sort all the training tasks from high to low according to the matching degree M; The third step is to select training tasks that are ranked first by a preset proportion or whose matching degree is greater than a preset threshold to form a candidate task set. The fourth step involves selecting a preset number of single-dimensional training tasks and / or composite training tasks from the candidate task set in descending order of matching degree to form the final personalized rehabilitation training program.

[0026] S4 aims to adaptively adjust the difficulty level of the corresponding target training task based on the trainee's real-time training behavior data, achieving dynamic adaptation of training difficulty and ensuring the effectiveness and scientific nature of rehabilitation training. During the trainee's execution of the personalized rehabilitation training plan, the system continuously acquires the trainee's training behavior data and adaptively adjusts the difficulty level of the corresponding target training task based on this data, simultaneously outputting plan adjustment prompts.

[0027] The process of adaptively adjusting the difficulty level of the corresponding target training task is as follows: The first step is to obtain the preset scoring rules for each target training task and the corresponding score thresholds for each difficulty level; The second step is to calculate the average score of the trainee during a preset time period or a preset number of consecutive training sessions. Third, if the average score is higher than the upper limit threshold of the current difficulty level, the difficulty level of the target training task is increased by one level; if the average score is lower than the lower limit threshold of the current difficulty level, the difficulty level of the target training task is decreased by one level.

[0028] It should be noted that the difficulty level adjustments are made progressively within the N-level range.

[0029] It should be noted that the different difficulty levels of the training tasks are achieved by adjusting the corresponding training parameters. All tasks have a preset difficulty-task execution parameter correspondence table. The training parameters include, but are not limited to, physical environment parameters, kinematic instruction parameters, and multimedia stimulus intensity parameters.

[0030] The difficulty adjustment methods for single-dimensional training tasks are as follows: visual training tasks adjust the difficulty by adjusting the intensity of hardware visual stimulation; vestibular sensory training tasks adjust the difficulty by adjusting the amplitude, speed, and frequency of head movements; proprioceptive tasks adjust the difficulty by adjusting the physical parameters of the support surface; posture control strategy training tasks adjust the difficulty by adjusting the training duration, the amplitude of posture changes, and the intensity of disturbances; and composite training tasks involve multiple physiological dimensions, and their difficulty is achieved by adjusting the corresponding training parameters proportionally according to the training weight of each dimension in the composite training task.

[0031] When the trainee's scores for all training tasks in the current personalized rehabilitation training program are consistently higher than the upper limit threshold of the highest level of difficulty for the corresponding task, it indicates that the training goal of the personalized rehabilitation training program has been achieved. The current training program can be ended or step S1 can be repeated for balance function reassessment to generate a new personalized rehabilitation training program.

Claims

1. An adaptive balance rehabilitation method based on multidimensional physiological feature weighted matching, characterized in that, Includes the following steps: S1: Obtain the initial assessment score of the balance function physiological dimension of the trainee. The initial assessment score of the physiological dimension is obtained by collecting objective kinematic data through external sensors or by input by the assessor. S2: Calculate the personalized physiological dimension weight coefficients of the trainee based on the preset basic weight coefficients of each physiological dimension and the initial evaluation score; S3: Based on the personalized physiological dimension weight coefficient, a weighted matching algorithm is used to perform feature matching and screening in a pre-established parametric rehabilitation training task library to generate a personalized rehabilitation training plan containing at least one target training task; wherein, the parametric rehabilitation training task library stores multiple training tasks, and each training task is configured with a corresponding quantitative value of training effect and difficulty level. S4: During the process of the trainee executing the personalized rehabilitation training program, continuously acquire the trainee's training behavior data, and adaptively adjust the difficulty level of the corresponding target training task according to the training behavior data, and output program adjustment prompts simultaneously.

2. The adaptive balance rehabilitation method based on multidimensional physiological feature weighted matching according to claim 1, characterized in that, The physiological dimensions include visual dimension A, vestibular sensory dimension B, proprioceptive sensory dimension C, and posture control strategy dimension D; the corresponding basic weight coefficients are W, ... A W B W C W D And satisfy W A +W B +W C +W D =1.

3. The adaptive balance rehabilitation method based on multidimensional physiological feature weighted matching according to claim 1, characterized in that, The weighting coefficients for the individualized physiological dimensions of the trainees are: Obtain the initial assessment score S for each of the aforementioned physiological dimensions. A S B S C S D The maximum score is 100 points; Calculate the weighted defect score S' for each physiological dimension: S A ′=(100−S A )×W A S B ′=(100−S B )×W B S C ′=(100−S C )×W C S D ′=(100−S D )×W D ; The weighted defect scores are linearly normalized to obtain the personalized physiological dimension weight coefficient K. A K B K C K D The calculation methods are as follows: K A = K B = K C = K D = 。 4. The adaptive balance rehabilitation method based on multidimensional physiological feature weighted matching according to claim 1, characterized in that, The rehabilitation training task library includes single-dimensional training tasks targeting visual dimension A, vestibular sensory dimension B, proprioceptive dimension C, and postural control strategy dimension D, as well as composite training tasks that have training effects on multiple physiological dimensions; each training task has a corresponding quantitative value R for the training effect in the four physiological dimensions. A R B R C R D .

5. The adaptive balance rehabilitation method based on multidimensional physiological feature weighted matching according to claim 1, characterized in that, Based on the aforementioned personalized physiological dimension weight coefficients, a weighted matching algorithm is further used for feature matching and screening to generate a personalized rehabilitation training plan, specifically including: The matching degree M between each training task in the parameterized rehabilitation training task library and the trainee is calculated using the formula: M = R A K A + R B K B + R C K C + R D K D ; Sort all the training tasks in descending order of their matching degree M; Select training tasks that are ranked first by a preset proportion or whose matching degree is greater than a preset threshold to form a candidate task set; In the candidate task set, a preset number of single-dimensional training tasks and / or compound training tasks are selected in descending order of matching degree to form the final personalized rehabilitation training program.

6. The adaptive balance rehabilitation method based on multidimensional physiological feature weighted matching according to claim 1, characterized in that, The training task is configured with N difficulty levels, where N is a natural number greater than 1.

7. The adaptive balance rehabilitation method based on multidimensional physiological feature weighted matching according to claim 1, characterized in that, The different difficulty levels of the training tasks are achieved by adjusting the corresponding training parameters. All tasks have a preset difficulty-task execution parameter correspondence table. The training parameters include, but are not limited to, physical environment parameters, kinematic instruction parameters, and multimedia stimulus intensity parameters. The difficulty adjustment methods for single-dimensional training tasks are as follows: visual training tasks adjust the difficulty by adjusting the intensity of hardware visual stimulation; vestibular sensory training tasks adjust the difficulty by adjusting the amplitude, speed, and frequency of head movements; proprioceptive tasks adjust the difficulty by adjusting the physical parameters of the support surface; posture control strategy training tasks adjust the difficulty by adjusting the training duration, the amplitude of posture changes, and the intensity of disturbances; and composite training tasks involve multiple physiological dimensions, and their difficulty is achieved by adjusting the corresponding training parameters proportionally according to the training weight of each dimension in the composite training task.

8. The adaptive balance rehabilitation method based on multidimensional physiological feature weighted matching according to claim 1, characterized in that, The difficulty level of the corresponding target training task is adaptively adjusted based on the training behavior data, specifically including: Obtain the preset scoring rules for each target training task and the corresponding score thresholds for each difficulty level; calculate the average score of the trainee during a preset time period or a preset number of consecutive training sessions; if the average score is higher than the upper limit threshold of the current difficulty level, then the difficulty level of the target training task is increased by one level; if the average score is lower than the lower limit threshold of the current difficulty level, then the difficulty level of the target training task is decreased by one level; the adjustment of the difficulty level is carried out step by step within the N levels.

Images