Parkinson patient rehabilitation training method based on video gait analysis

Abstract

The invention discloses a Parkinson's disease patient rehabilitation training method based on video gait analysis, relates to the technical field of image analysis, solves the problem of dependence on subjective evaluation in the prior art, and realizes objective quantitative analysis of a rehabilitation effect. An athletic ability index Xpzs, a gait stability index Yb and a lower limb rehabilitation progress coefficient Ws of a patient are calculated and evaluated to ensure that an evaluation result is more reliable and accurate; meanwhile, an established real-time data feedback mechanism is combined with historical data, the problem that rehabilitation effect monitoring is insufficient is solved, and through regular tracking and effect evaluation, a training scheme and a rehabilitation target are dynamically adjusted, and it is ensured that the rehabilitation progress of a patient is tracked in a whole cycle; besides, the method is high in technical adaptability, personalized analysis can be provided according to illness states and rehabilitation progresses of different patients, the training intensity and the rehabilitation target can be dynamically adjusted, the flexibility and adaptability of a rehabilitation scheme are improved, the method is particularly suitable for patient groups with different illness states, and the rehabilitation effect is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image analysis, and specifically to a rehabilitation training method for Parkinson's patients based on video gait analysis. Background Art

[0002] Parkinson's Disease (PD) is a common neurodegenerative disease, and patients often exhibit symptoms such as movement disorders, tremors, and stiffness. With the continuous in-depth medical research, rehabilitation training methods for Parkinson's patients have gradually developed. In the early stage, drug treatment was mainly used, and in recent years, more attention has been paid to exercise therapy and physical therapy. Modern rehabilitation methods usually include exercise training, speech training, cognitive training, etc., to help patients improve their motor function and quality of life. The existing technologies mainly include traditional exercise programs based on physical therapy, virtual reality (VR) rehabilitation training, and robot-assisted therapy. Through these technical means, patients can carry out rehabilitation training in a safe and effective environment, improving the effect and speed of functional recovery.

[0003] However, there are still some deficiencies in the existing rehabilitation training methods for Parkinson's patients in terms of rehabilitation analysis, which are mainly reflected in the following aspects:

[0004] 1. More subjective evaluations than objective quantification: In the existing technologies, the progress of rehabilitation often depends on subjective evaluations, such as doctor or patient self-reports, lacking accurate, data-based objective quantitative analysis.

[0005] 2. Insufficient monitoring of rehabilitation effects: Most technologies lack long-term and full-cycle tracking and monitoring of rehabilitation effects, resulting in the inability to accurately evaluate the long-term recovery of patients, thereby affecting the adjustment of training programs.

[0006] 3. Poor adaptability of technologies: Some existing rehabilitation methods have poor adaptability in different patient groups. Especially for patients with mild or severe conditions, the rehabilitation effect may be greatly reduced, lacking a flexible analysis and adaptation mechanism. Summary of the Invention

[0007] In view of the deficiencies of the existing technologies, the present invention provides a rehabilitation training method for Parkinson's patients based on video gait analysis, solving the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention is realized through the following technical solutions: A rehabilitation training method for Parkinson's patients based on video gait analysis, including the following steps:

[0009] S1. Real-time collect gait video-related data of Parkinson's patients in different rehabilitation environments;

[0010] S2. Preprocess the gait video-related data, integrate the gait video-related data from different video sources to form a multi-source gait information set, and at the same time perform quality control and compliance checks on the multi-source gait information set;

[0011] S3. Quantitatively evaluate the patient's movement state based on the multi-source gait information set; by constructing a motion analysis model, combined with the diagnostic criteria for movement disorders, calculate the movement ability index Xpzs of the patient's lower limbs and evaluate it; then calculate and evaluate the patient's gait stability index Yb;

[0012] S4. Based on the evaluation content of the gait stability index Yb, collect the patient's historical rehabilitation data in real time, and once again use the motion analysis model to calculate the lower limb rehabilitation progress coefficient Ws of the patient, and compare it with a preset threshold to dynamically adjust the training plan and rehabilitation goals;

[0013] S5. Establish a real-time data feedback mechanism and monitor the patient's rehabilitation progress, including long-term tracking and effect evaluation based on the change trend of the movement ability index Xpzs.

[0014] Preferably, S1 specifically includes:

[0015] Through multiple video acquisition devices arranged in the rehabilitation environment, including high-definition cameras, infrared sensors, and depth cameras, which are connected to the data acquisition system through a wireless connection method to collect the gait video-related data of Parkinson's patients in real time; during the collection process of the gait video-related data, the video acquisition devices capture all aspects of the patient's walking in different directions, speeds, and postures through set acquisition parameters, including acquisition frequency, resolution, and angle; the collected gait video-related data is uploaded through the front-end interface and then preprocessed.

[0016] Preferably, S2 specifically includes:

[0017] Unify the formats of gait video-related data from different sources, including the conversion of video frame rate, resolution, and encoding format; then, use timestamp and motion synchronization algorithms to perform time alignment and spatial registration on gait video-related data from multiple video sources, precisely match and integrate the gait video data from different video sources to generate a unified multi-source gait information set; the multi-source gait information set contains multi-angle gait information data of the patient in different environments; during the integration process, automatically identify and exclude low-quality video data, including blurred, distorted, and unclear motion video segments; subsequently, apply a data quality control algorithm to check the quality of the integrated multi-source gait information set, detect and correct noise, duplicate frames, and missing frames in the data, and verify the integrity and continuity of the data; finally, perform compliance checks on the multi-source gait information set, including relevant privacy protection compliance checks and data processing regulation requirement checks.

[0018] Preferably, S3 specifically includes:

[0019] Adopt image processing and deep learning algorithms to extract the patient's gait cycle and key motion features from the multi-source gait information set; then establish a motion analysis model through a convolutional neural network, and combine the clinical diagnostic criteria for motor disorders to train and analyze the extracted key motion features through the motion analysis model, identify the patient's current motion pattern, and determine whether there are signs of lower limb motor disorders when the patient is undergoing rehabilitation.

[0020] When there are no signs of lower limb motor disorders when the patient is undergoing rehabilitation, extract key point coordinates including foot lift-off points and stride boundaries from the multi-source gait information set through video analysis technology and image recognition algorithms, and combine timestamps and the motion analysis model to obtain gait feature-related data including gait symmetry coefficient Xa, gait cycle change rate Xb, single-step duration Xc, and stride rhythm inconsistency Xd in real time.

[0021] After extracting the gait symmetry coefficient Xa, gait cycle change rate Xb, single-step duration Xc, and gait rhythm inconsistency Xd and performing dimensionless processing, calculate and obtain the motor ability index Xpzs through the following formula.

[0022] Preferably, S3 specifically further includes:

[0023] Preset a motor ability threshold X and compare and evaluate it with the motor ability index Xpzs. The specific evaluation content is as follows:

[0024] When the motor ability index Xpzs ≥ the motor ability threshold X, it indicates that the patient's current motor ability is qualified; at this time, evaluate the patient's gait during rehabilitation.

[0025] When the exercise ability index Xpzs < the exercise ability threshold X, it indicates that the patient's current exercise ability is unqualified, and at this time, the rehabilitation training plan is adjusted.

[0026] Preferably, S3 specifically further includes:

[0027] The gait feature-related data further includes the stride coefficient of variation Ya, the support time ratio Yb, the gait symmetry difference Yc, and the dynamic balance index Yd;

[0028] When the patient's current exercise ability is qualified, the stride coefficient of variation Ya, the support time ratio Yb, the gait symmetry difference Yc, and the dynamic balance index Yd are extracted and dimensionless processed, and then the patient's current gait stability index Ybzs is calculated using the following formula.

[0029] Preferably, S3 specifically further includes:

[0030] The preset gait stability threshold Y is compared and evaluated with the gait stability index Ybzs. The specific evaluation content is as follows:

[0031] When the gait stability index Ybzs ≥ the gait stability threshold Y, it indicates that the patient's current gait stability is qualified, and the gait control ability at this time meets the rehabilitation expectation;

[0032] When the gait stability index Ybzs < the gait stability threshold Y, it indicates that the patient's current gait stability is unqualified, and the gait control ability at this time does not meet the rehabilitation expectation.

[0033] Preferably, S4 specifically includes:

[0034] When the patient's current gait stability is qualified, the historical rehabilitation data of the patient is collected in real time. Using the motion analysis model, the pace coordination index Wa, the gait symmetry difference index Wb, and the pace lag duration Wc are obtained in real time, and after dimensionless processing, the lower limb rehabilitation progress coefficient Ws is calculated through the following formula;

[0035] In the formula, n represents the number of measurements, and i represents the measurement point index;

[0036] Wa i represents the i-th measurement value of the pace coordination index, Wb i represents the i-th measurement value of the gait symmetry difference index, Wc i represents the i-th measurement value of the pace lag duration;

[0037] δ is a constant index, and its value in the formula is 2, indicating the weighted adjustment of the rehabilitation progress coefficient.

[0038] Preferably, S4 specifically further includes:

[0039] Preset the first rehabilitation progress threshold W1 and the second rehabilitation progress threshold W2, and compare and evaluate them with the lower limb rehabilitation progress coefficient Ws to dynamically adjust the training plan and rehabilitation goals; among them, the first rehabilitation progress threshold W1 is greater than the second rehabilitation progress threshold W2, and the specific evaluation content is as follows:

[0040] When the lower limb rehabilitation progress coefficient Ws ≥ the first rehabilitation progress threshold W1, it means that the lower limb rehabilitation of the patient has exceeded the expected rehabilitation threshold; it means that the current training plan and rehabilitation goals are not applicable to the patient, and then re-customize the training plan and rehabilitation goals, where the new training intensity is 20% higher than the existing training intensity;

[0041] When the first rehabilitation progress threshold W1 > the lower limb rehabilitation progress coefficient Ws ≥ the second rehabilitation progress threshold W2, it means that the lower limb rehabilitation of the patient has reached the expected rehabilitation threshold; it means that the current training plan is effective, continue to use the current training plan and rehabilitation goals, and at the same time maintain the existing training intensity;

[0042] When the second rehabilitation progress threshold W2 > the lower limb rehabilitation progress coefficient Ws, it means that the lower limb rehabilitation of the patient has not reached the expected rehabilitation threshold; it means that the current training plan and rehabilitation goals are not applicable to the patient, and the training plan and rehabilitation goals need to be adjusted, including adjusting the new training intensity to be 10% lower than the existing training intensity.

[0043] Preferably, S5 specifically includes:

[0044] During the preset time period in the patient's rehabilitation process, the motor ability index Xpzs is collected and calculated in real time within a fixed cycle; then after three time periods, analyze the change trend of the patient's motor ability index Xpzs; when the change trend of the motor ability index Xpzs shows that the patient's gait control ability is on the rise, promote the patient to enter a higher level of rehabilitation stage; otherwise, adjust the existing rehabilitation plan to provide a more personalized rehabilitation strategy;

[0045] In addition, after analyzing the change trend of the patient's motor ability index Xpzs, the gait stability index Yb and the rehabilitation progress coefficient Ws will also be recalculated and evaluated; finally, according to the results of the re-evaluation, dynamically adjust the patient's training plan and rehabilitation goals.

[0046] The present invention provides a rehabilitation training method for Parkinson's patients based on video gait analysis. It has the following

[0047] Beneficial effects:

[0048] (1) The rehabilitation training method for Parkinson's patients based on video gait analysis solves the problem that subjective evaluation is more than objective quantification in the prior art: through accurate data collection based on video gait analysis and the integration of multi-source gait information, this method realizes the accurate and objective quantification analysis of indicators including the motor ability index Xpzs, gait stability index Yb, and rehabilitation progress coefficient Ws of Parkinson's patients during the rehabilitation training process; it is obtained through the calculation of the motion analysis model, and the differences in data under different environments are eliminated through dimensionless processing, making the evaluation of the rehabilitation effect more dependent on real-time data rather than subjective reports from doctors or patients themselves, thereby improving the reliability and accuracy of the evaluation results.

[0049] (2) The rehabilitation training method for Parkinson's patients based on video gait analysis solves the problem of insufficient monitoring of rehabilitation effects in the prior art. By establishing a real-time data feedback mechanism and combining the patient's historical rehabilitation data, it regularly tracks and evaluates the motor ability index Xpzs to ensure that the progress of each stage during the rehabilitation process can be timely reflected in the adjustment of the rehabilitation goals; multiple parameters such as the patient's gait stability index Yb, step coordination index Wa, gait symmetry difference Wb, and step lag duration Wc can be collected and calculated in real time, and a long-term monitoring mechanism is used to achieve full-cycle tracking of the rehabilitation process, ensuring that a timely and accurate evaluation of the patient's rehabilitation situation can be made and the training plan and rehabilitation goals can be dynamically adjusted.

[0050] (3) The rehabilitation training method for Parkinson's patients based on video gait analysis solves the problem of poor technical adaptability in the prior art: the multi-source gait information set and its integrated processing method adopted by this method have high adaptability and can provide personalized analysis and evaluation for the conditions and rehabilitation progress of different patients. Through the comprehensive evaluation of the gait stability index Yb and the rehabilitation progress coefficient Ws, combined with the changing trend of the patient's motor ability at different stages, the training intensity and rehabilitation goals are dynamically adjusted, especially suitable for patients with different degrees of illness, ensuring the flexibility and adaptability of the rehabilitation plan and greatly improving the rehabilitation effects of different patient groups. Brief Description of the Drawings

[0051] Figure 1 It is a schematic diagram of the step flow of the rehabilitation training method for Parkinson's patients based on video gait analysis of the present invention. Detailed Embodiments

[0052] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0053] Example 1

[0054] Please refer to Figure 1 , the present invention provides a rehabilitation training method for Parkinson's patients based on video gait analysis, which is characterized in that it includes the following steps:

[0055] S1. Real-time collect the gait video-related data of Parkinson's patients in different rehabilitation environments;

[0056] S2. Preprocess the gait video-related data, integrate the gait video-related data from different video sources to form a multi-source gait information set, and at the same time perform quality control and compliance checks on the multi-source gait information set;

[0057] S3. Quantitatively evaluate the patient's motion state based on the multi-source gait information set; by constructing a motion analysis model, combining the diagnostic criteria for movement disorders, calculate the motion ability index Xpzs of the patient's lower limbs and evaluate it; then calculate and evaluate the gait stability index Yb of the patient;

[0058] S4. Based on the evaluation content of the gait stability index Yb, real-time collect the patient's historical rehabilitation data, and use the motion analysis model again, calculate the lower limb rehabilitation progress coefficient Ws of the patient, and compare it with a preset threshold to dynamically adjust the training plan and rehabilitation goal;

[0059] S5. Establish a real-time data feedback mechanism and monitor the patient's rehabilitation progress, including long-term tracking and effect evaluation based on the change trend of the motion ability index Xpzs.

[0060] In this embodiment, step S1 can realize real-time acquisition of the gait video data of Parkinson's patients in different rehabilitation environments, providing basic data support for subsequent analysis; step S2 ensures the high quality and consistency of the data through preprocessing of the gait video data and multi-source data integration, reducing the risk of affecting the evaluation results due to data inconsistency or noise; step S3 quantifies and evaluates the motion ability index Xpzs and gait stability index Yb of the patient by constructing a motion analysis model and combining the diagnostic criteria for movement disorders, which can accurately evaluate the patient's current motion state and facilitate the formulation of personalized rehabilitation plans; step S4 calculates the rehabilitation progress coefficient Ws of the patient based on the gait stability index Yb and historical rehabilitation data, and compares it with a preset threshold, which helps to dynamically adjust the rehabilitation goal and training plan to ensure the maximization of the treatment effect; step S5 monitors the patient's rehabilitation progress through a real-time data feedback mechanism, combines the change trend of the motion ability index Xpzs for long-term tracking and effect evaluation, so that the rehabilitation plan is more flexible and efficient, and promotes the patient's rehabilitation process.

[0061] Example 2

[0062] S1 specifically includes:

[0063] Multiple video acquisition devices arranged in the rehabilitation environment, including high-definition cameras, infrared sensors, and depth cameras, are connected to the data acquisition system through a wireless connection method to collect gait video-related data of Parkinson's patients in real time; during the collection of gait video-related data, the video acquisition devices capture comprehensively according to the set acquisition parameters, including acquisition frequency, resolution, and angle, for different directions, speeds, and postures of the patient's walking; the collected gait video-related data is uploaded through the front-end interface and then preprocessed.

[0064] S2 specifically includes:

[0065] Unify the formats of gait video-related data from different sources, including the conversion of video frame rate, resolution, and encoding format; then, use the timestamp and motion synchronization algorithm to perform time alignment and spatial registration on the gait video-related data from multiple video sources, accurately match and integrate the gait video data from different video sources to generate a unified multi-source gait information set; the multi-source gait information set contains multi-angle gait information data of the patient in different environments; during the integration process, automatically identify and exclude low-quality video data, including blurred, distorted, and unclear motion video segments; subsequently, apply the data quality control algorithm to check the quality of the integrated multi-source gait information set, detect and correct the noise, duplicate frames, and lost frames in the data, and verify the integrity and continuity of the data; finally, perform compliance checks on the multi-source gait information set, including relevant privacy protection compliance checks and data processing regulation requirement checks.

[0066] In this embodiment, in step S1, by arranging a variety of video acquisition devices such as high-definition cameras, infrared sensors, and depth cameras, the gait information of the patient can be comprehensively captured, and by setting the acquisition frequency, resolution, and angle, the gait data in different directions, speeds, and postures can be accurately collected, uploaded in real time and preprocessed to ensure the high quality and real-time nature of the data;

[0067] In step S2, by unifying the formats of video data from different sources and using the timestamp and motion synchronization algorithm for time alignment and spatial registration, the consistency and accuracy of the multi-source gait information set are ensured. At the same time, low-quality video data is excluded and quality control is performed, optimizing the data integrity, continuity, and compliance checks, ensuring the high precision and legality of the data, making the subsequent analysis results more reliable and accurate; the advantage of this process is that by comprehensively capturing and optimizing video data, it ensures that the system can provide high-quality and accurate gait analysis, provides a scientific basis for the rehabilitation training of Parkinson's patients, enhances the personalization and pertinence of the training plan, and improves the ability of rehabilitation effect monitoring and dynamic adjustment.

[0068] Among them, the unified multi-source gait information set includes temporal consistency, spatial consistency, data format consistency, and feature consistency.

[0069] Embodiment 3

[0070] S3 specifically includes:

[0071] Using image processing and deep learning algorithms, extract the gait cycle and key motion features of the patient from the multi-source gait information set; then establish a motion analysis model through a convolutional neural network, and combine the clinical diagnostic criteria for motor disorders. Through the motion analysis model, train and analyze the extracted key motion features, identify the patient's current motion pattern, and determine whether there are signs of lower limb motor disorders when the patient is undergoing rehabilitation;

[0072] When there are no signs of lower limb motor disorders during the patient's rehabilitation, extract the key point coordinates including the foot lift-off point and stride boundary from the multi-source gait information set through video analysis technology and image recognition algorithms, and combine the time stamp and the motion analysis model to obtain gait feature-related data including the gait symmetry coefficient Xa, the gait cycle change rate Xb, the single-step duration Xc, and the stride rhythm inconsistency degree Xd in real time;

[0073] After extracting the gait symmetry coefficient Xa, the gait cycle change rate Xb, the single-step duration Xc, and the gait rhythm inconsistency degree Xd and performing dimensionless processing, calculate and obtain the motor ability index Xpzs through the following formula:

[0074]

[0075] S3 specifically further includes:

[0076] Preset a motor ability threshold X, and compare and evaluate it with the motor ability index Xpzs. The specific evaluation content is as follows:

[0077] When the motor ability index Xpzs ≥ the motor ability threshold X, it means that the patient's current motor ability is qualified; at this time, evaluate the patient's gait during rehabilitation;

[0078] When the motor ability index Xpzs < the motor ability threshold X, it means that the patient's current motor ability is unqualified, and at this time, adjust the rehabilitation training plan.

[0079] S3 specifically further includes:

[0080] The gait feature-related data also includes the stride variation coefficient Ya, the support time ratio Yb, the stride symmetry difference Yc, and the dynamic balance index Yd;

[0081] When the patient's current motor ability is qualified, the coefficient of stride variation Ya, the support time ratio Yb, the difference in gait symmetry Yc, and the dynamic balance index Yd are extracted and dimensionless processed, and then the following formula is used to calculate the patient's current gait stability index Ybzs:

[0082]

[0083] S3 specifically further includes:

[0084] The preset gait stability threshold Y is compared and evaluated with the gait stability index Ybzs. The specific evaluation content is as follows:

[0085] When the gait stability index Ybzs ≥ the gait stability threshold Y, it indicates that the patient's current gait stability is qualified, and the gait control ability at this time meets the rehabilitation expectation;

[0086] When the gait stability index Ybzs < the gait stability threshold Y, it indicates that the patient's current gait stability is unqualified, and the gait control ability at this time does not meet the rehabilitation expectation.

[0087] S4 specifically includes:

[0088] When the patient's current gait stability is qualified, the historical rehabilitation data of the patient is collected in real time. Using the motion analysis model, the step coordination index Wa, the gait symmetry difference index Wb, and the step lag duration Wc are obtained in real time, and after dimensionless processing, the following formula is used to calculate the lower limb rehabilitation progress coefficient Ws:

[0089]

[0090] In the formula, n represents the number of measurements, and i represents the measurement point index;

[0091] Wa i represents the i-th measurement value of the step coordination index, Wb i represents the i-th measurement value of the gait symmetry difference index, Wc i represents the i-th measurement value of the step lag duration;

[0092] δ is a constant exponent, and its value in the formula is 2, indicating the weighted adjustment of the rehabilitation progress coefficient.

[0093] S4 specifically further includes:

[0094] The preset first rehabilitation progress threshold W1 and the second rehabilitation progress threshold W2 are compared and evaluated with the lower limb rehabilitation progress coefficient Ws, and the training plan and rehabilitation goal are dynamically adjusted; among them, the first rehabilitation progress threshold W1 is greater than the second rehabilitation progress threshold W2. The specific evaluation content is as follows:

[0095] When the lower limb rehabilitation progress coefficient Ws ≥ the first rehabilitation progress threshold W1, it indicates that the patient's lower limb rehabilitation has exceeded the expected rehabilitation threshold; it means that the current training plan and rehabilitation goal are not applicable to the patient, and then a new training plan and rehabilitation goal need to be customized, where the new training intensity is 20% higher than the existing training intensity;

[0096] When the first rehabilitation progress threshold W1 > the lower limb rehabilitation progress coefficient Ws ≥ the second rehabilitation progress threshold W2, it indicates that the patient's lower limb rehabilitation has reached the expected rehabilitation threshold; it means that the current training plan is effective, and the current training plan and rehabilitation goal are continuously used while maintaining the existing training intensity;

[0097] When the second rehabilitation progress threshold W2 > the lower limb rehabilitation progress coefficient Ws, it indicates that the patient's lower limb rehabilitation has not reached the expected rehabilitation threshold; it means that the current training plan and rehabilitation goal are not applicable to the patient, and the training plan and rehabilitation goal need to be adjusted, including adjusting the new training intensity to be 10% lower than the existing training intensity.

[0098] S5 specifically includes:

[0099] During a preset time period in the patient's rehabilitation process, the evaluation of the motor ability index Xpzs is collected and calculated in real time within a fixed cycle; then, after three time periods, the change trend of the patient's motor ability index Xpzs is analyzed; when the change trend of the motor ability index Xpzs shows that the patient's gait control ability is on the rise, the patient is promoted to a higher level of rehabilitation stage; otherwise, the existing rehabilitation plan is adjusted to provide a more personalized rehabilitation strategy;

[0100] In addition, after analyzing the change trend of the patient's motor ability index Xpzs, the gait stability index Yb and the rehabilitation progress coefficient Ws will also be recalculated and evaluated; finally, according to the results of the re-evaluation, the patient's training plan and rehabilitation goal are dynamically adjusted.

[0101] In this embodiment, in step S3, the gait cycle and key motion features of the patient are extracted from the multi-source gait information set through image processing and deep learning algorithms. Combining a convolutional neural network with the clinical diagnostic criteria for movement disorders for motion analysis can identify the patient's motion pattern in real time and determine whether there are signs of lower limb movement disorders. When there is no disorder, the key point coordinates are extracted through video analysis technology, and gait feature-related data including the gait symmetry coefficient Xa, the gait cycle change rate Xb, the single-step duration Xc, and the stride rhythm inconsistency degree Xd are calculated. The motor ability index Xpzs is calculated through a formula to evaluate the patient's motor ability. Combining with a preset motor ability threshold X for evaluation, when the motor ability index Xpzs is greater than the preset threshold, it indicates that the motor ability is qualified; otherwise, the training plan is adjusted. Step S3 also includes calculating the gait stability index Ybzs and comparing it with the gait stability threshold Y for evaluation. If the gait stability index Ybzs exceeds the preset threshold, it indicates that the gait stability is qualified and meets the rehabilitation expectation; otherwise, the gait control ability needs to be adjusted. Through this step, it can help accurately monitor the patient's rehabilitation progress;

[0102] In step S4, the patient's historical rehabilitation data is collected in real time, and the pace coordination index Wa, the gait symmetry difference index Wb, and the pace lag duration Wc are obtained through the motion analysis model. The lower limb rehabilitation progress coefficient Ws is calculated and compared with the preset rehabilitation progress thresholds W1 and W2 to dynamically adjust the training plan and rehabilitation goal;

[0103] In step S5, the motor ability index Xpzs is collected in real time over a preset time period and a fixed cycle, and its change trend is analyzed. It can dynamically evaluate the change of the patient's gait control ability, so as to promote the patient to enter a higher level of rehabilitation stage or adjust the existing rehabilitation plan according to the trend, providing a personalized rehabilitation strategy. This step not only focuses on the short-term changes in motor ability, but also involves the recalculation and evaluation of the gait stability index Yb and the rehabilitation progress coefficient Ws. Through real-time data feedback and effect evaluation, the rehabilitation plan is made more flexible and accurate, meeting the individualized rehabilitation needs of the patient. Compared with the prior art, this step realizes a personalized rehabilitation process through a data-driven dynamic adjustment mechanism, avoids the blindness of the traditional fixed training plan, improves the rehabilitation effect, enhances the patient's sense of participation and enthusiasm, and provides a more scientific and effective rehabilitation management method.

[0104] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A rehabilitation training method for Parkinson's disease patients based on video gait analysis, characterized in that: The following steps are involved: S1. Real-time collection of gait video data of Parkinson's patients in different rehabilitation environments; S2. Preprocessing gait video related data, and integrating gait video related data from different video sources to form a multi-source gait information set, and performing quality control and compliance inspection on the multi-source gait information set; S3. Quantitatively evaluate the patient's movement state based on a multi-source gait information set; calculate and evaluate the patient's lower limb movement ability index Xpzs by constructing a movement analysis model and combining it with the diagnostic criteria for movement disorders; then calculate and evaluate the patient's gait stability index Yb; S4. Based on the evaluation content of the gait stability index Yb, the patient's historical rehabilitation data is collected in real time, and the motion analysis model is used again to calculate the patient's lower limb rehabilitation progress coefficient Ws, and then compared with the preset threshold value to dynamically adjust the training plan and rehabilitation goals; S5. Establish a real-time data feedback mechanism and monitor the patient's rehabilitation progress, including long-term tracking and effect evaluation based on the changing trend of the exercise capacity index Xpzs.

2. The method for rehabilitation training of Parkinson's disease patients based on video gait analysis according to claim 1, characterized in that: S1 specifically includes: Multiple video acquisition devices, including high-definition cameras, infrared sensors and depth cameras, are arranged in the rehabilitation environment and connected to the data acquisition system through wireless connection to collect gait video data of Parkinson's patients in real time. During the gait video data collection process, the video acquisition device captures the different directions, speeds and postures of the patient's walking in all directions through the set acquisition parameters, including acquisition frequency, resolution and angle. The collected gait video data is uploaded through the front-end interface and then pre-processed.

3. The Parkinson's disease patient rehabilitation training method based on video gait analysis according to claim 1, characterized in that: S2 specifically includes: The formats of gait video-related data from different sources are unified, including conversion of video frame rate, resolution and encoding format; then, the timestamp and motion synchronization algorithms are used to time align and spatially register the gait video-related data from multiple video sources, accurately match and integrate the gait video data from different video sources, and generate a unified multi-source gait information set; the multi-source gait information set contains multi-angle gait information data of patients in different environments; during the integration process, low-quality video data is automatically identified and excluded, including blurred, distorted and unclear motion video clips; then, the data quality control algorithm is used to perform a quality check on the integrated multi-source gait information set, detect and correct noise, duplicate frames and lost frames in the data, and verify the integrity and continuity of the data; finally, the multi-source gait information set is checked for compliance, including relevant privacy protection compliance checks and data processing regulatory requirements checks.

4. The method for rehabilitation training of Parkinson's disease patients based on video gait analysis according to claim 1, characterized in that: S3 specifically includes: Image processing and deep learning algorithms are used to extract the patient's gait cycle and key motion features from a multi-source gait information set. Then, a motion analysis model is established through a convolutional neural network. Combined with the clinical diagnostic criteria for movement disorders, the extracted key motion features are trained and analyzed through the motion analysis model to identify the patient's current movement pattern and determine whether the patient has signs of lower limb movement disorders during rehabilitation. When the patient has no signs of lower limb movement disorder during rehabilitation, the coordinates of key points including the starting and landing points and stride boundaries are extracted from the multi-source gait information set through video analysis technology and image recognition algorithms. Combined with the timestamp and motion analysis model, the gait feature-related data including the gait symmetry coefficient Xa, gait cycle change rate Xb, single step duration Xc, and stride rhythm inconsistency Xd are obtained in real time; The gait symmetry coefficient Xa, gait cycle change rate Xb, single step duration Xc, and gait rhythm inconsistency Xd were extracted and dimensionlessly processed to obtain the exercise capacity index Xpzs.

5. The Parkinson's disease rehabilitation training method based on video gait analysis according to claim 1, characterized in that: S3 specifically includes: The preset athletic ability threshold X is compared with the athletic ability index Xpzs for evaluation. The specific evaluation contents are as follows: When the exercise capacity index Xpzs ≥ the exercise capacity threshold X, it means that the patient's current exercise capacity is qualified; at this time, the patient's gait during rehabilitation is evaluated; When the exercise capacity index Xpzs is less than the exercise capacity threshold X, it means that the patient's current exercise capacity is not up to standard, and the rehabilitation training plan should be adjusted at this time.

6. The Parkinson's disease patient rehabilitation training method based on video gait analysis according to claim 1, characterized in that: S3 specifically includes: The gait characteristic related data also include stride variation coefficient Ya, support time ratio Yb, step symmetry difference Yc, and dynamic balance index Yd; When the patient's current motor ability is qualified, the stride variation coefficient Ya, support time ratio Yb, step symmetry difference Yc, and dynamic balance index Yd are extracted and dimensionlessly processed to calculate the patient's current gait stability index Ybzs.

7. The method for rehabilitation training of Parkinson's disease patients based on video gait analysis according to claim 1, characterized in that: S3 specifically includes: The preset gait stability threshold Y is compared with the gait stability index Ybzs for evaluation. The specific evaluation contents are as follows: When the gait stability index Ybzs ≥ gait stability threshold Y, it means that the patient's current gait stability is qualified, and the gait control ability at this time meets the rehabilitation expectations; When the gait stability index Ybzs is less than the gait stability threshold Y, it means that the patient's current gait stability is unqualified and the gait control ability at this time does not meet the rehabilitation expectations.

8. The Parkinson's disease patient rehabilitation training method based on video gait analysis according to claim 1, characterized in that: S4 specifically includes: When the patient's current gait stability is qualified, the patient's historical rehabilitation data is collected in real time, and the motion analysis model is used to obtain the step coordination index Wa, gait symmetry difference index Wb and step delay duration Wc in real time. After dimensionless processing, the lower limb rehabilitation progress coefficient Ws is calculated using the following formula; Wa i represents the i-th measurement value of the step coordination index, Wb i represents the ith measurement value of the gait symmetry difference index, Wc i represents the i-th measurement value of the step delay duration; δ is a constant exponent, and its value is 2, which indicates the weighted adjustment of the rehabilitation progress coefficient.

9. The Parkinson's disease patient rehabilitation training method based on video gait analysis according to claim 1, characterized in that: S4 specifically includes: The first rehabilitation progress threshold W1 and the second rehabilitation progress threshold W2 are preset, and compared with the lower limb rehabilitation progress coefficient Ws for evaluation, so as to dynamically adjust the training program and rehabilitation goals; wherein, the first rehabilitation progress threshold W1 is greater than the second rehabilitation progress threshold W2, and the specific evaluation contents are as follows: When the lower limb rehabilitation progress coefficient Ws ≥ the first rehabilitation progress threshold W1, it means that the patient's lower limb rehabilitation has exceeded the expected rehabilitation threshold; it means that the training program and rehabilitation goals at this time are not suitable for the patient, and the training program and rehabilitation goals are re-customized next, in which the new training intensity is 20% higher than the current training intensity; When the first rehabilitation progress threshold W1>lower limb rehabilitation progress coefficient Ws≥second rehabilitation progress threshold W2, it means that the patient's lower limb rehabilitation has reached the expected rehabilitation threshold; it means that the training program at this time is effective, and the current training program and rehabilitation goals should be continued, while maintaining the current training intensity; When the second rehabilitation progress threshold W2>the lower limb rehabilitation progress coefficient Ws, it means that the patient's lower limb rehabilitation has not reached the expected rehabilitation threshold; it means that the training plan and rehabilitation goals at this time are not suitable for the patient, and the training plan and rehabilitation goals need to be adjusted, including adjusting the new training intensity to 10% lower than the current training intensity.

10. The Parkinson's disease patient rehabilitation training method based on video gait analysis according to claim 1, characterized in that: S5 specifically includes: During the patient's rehabilitation process, the motor ability index Xpzs is collected and calculated in real time within a fixed period during a preset time period; then, after three time periods, the change trend of the patient's motor ability index Xpzs is analyzed; when the change trend of the motor ability index Xpzs shows that the patient's gait control ability is on the rise, the patient is pushed to a higher level of rehabilitation; otherwise, the existing rehabilitation plan is adjusted to provide a more personalized rehabilitation strategy; In addition, after analyzing the changing trend of the patient's motor ability index Xpzs, the gait stability index Yb and rehabilitation progress coefficient Ws will also be recalculated and evaluated; finally, based on the results of the re-evaluation, the patient's training plan and rehabilitation goals will be dynamically adjusted.

Images