Patient gait abnormity evaluation auxiliary system based on human body posture detection

Abstract

The invention discloses a patient gait abnormity evaluation auxiliary system based on human body posture detection, and relates to the technical field of intelligent medical treatment, and the system comprises the steps: collecting patient gait video data, preprocessing the patient gait video data, and marking patient skeleton key points; based on the patient skeleton data set, calculating position coordinates of patient skeleton key points, and according to the position coordinates of the patient skeleton key points, identifying motion tracks of the patient skeleton key points; on the basis of the patient skeleton key point motion trail, patient gait features are extracted, and a comprehensive gait feature vector is formed; training a gait prediction model by using the patient skeleton data set, inputting the comprehensive gait feature vector into the gait prediction model, and predicting a gait anomaly index of the patient; and evaluating the prediction result, and generating an evaluation report. The gait abnormity index of the patient is predicted by recognizing the motion trail of the skeleton key points of the patient, comprehensive evaluation of the gait state of the patient is achieved, and clinical diagnosis accuracy and efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent medical technology, and in particular to a patient gait abnormality assessment auxiliary system based on human body posture detection. Background Art

[0002] With the development of medical technology, gait analysis, as a non-invasive diagnostic tool, has received increasing attention in clinical applications. Traditional gait analysis mainly relies on manual observation and manual recording, which is not only time-consuming and labor-intensive, but also highly subjective and difficult to achieve accurate quantification. In recent years, with the advancement of computer vision and machine learning technology, automated gait analysis systems have gradually become a research hotspot. These systems can capture and analyze patients' gait data in real time through video monitoring and posture detection technology, improving the efficiency and accuracy of gait analysis.

[0003] However, the existing automated gait analysis systems still have some shortcomings. Most of the existing gait analysis systems rely on fixed cameras and restricted experimental environments, and cannot adapt well to a variety of application scenarios. In addition, most systems only focus on the changes in a single joint or part when extracting features, and fail to fully consider the integrity and dynamics of the gait. This results in low recognition accuracy of the system when faced with complex gait abnormalities. In addition, in terms of model training, many systems use simple machine learning algorithms, fail to fully utilize the characteristics of time series data, and ignore the time dependency of gait data, which affects the prediction ability of the model. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a patient gait abnormality assessment auxiliary system based on human posture detection to solve the problem of low recognition accuracy caused by reliance on fixed equipment, focusing on single features and ignoring time dependency in the prior art.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: The embodiment of the present invention provides a patient gait abnormality assessment auxiliary system based on human posture detection, which includes a data acquisition and preprocessing module: acquiring patient gait video data, preprocessing the patient gait video data and marking the patient skeleton key points; Skeleton key point labeling module: Based on the preprocessed patient gait video data, label the key point data in the patient skeleton and generate the patient skeleton dataset; Skeleton key point trajectory recognition module: Based on the patient skeleton data set, the position coordinates of the patient's skeleton key points are calculated, and the movement trajectory of the patient's skeleton key points is identified according to the position coordinates of the skeleton key points; Gait feature extraction module: based on the motion trajectory of the patient's skeleton key points, extract the patient's gait features to form a comprehensive gait feature vector; Gait abnormality index prediction module: Use the patient skeleton data set to train a gait prediction model, input the comprehensive gait feature vector into the gait prediction model, and predict the patient's gait abnormality index; Evaluation and report generation module: evaluates the prediction results and generates an evaluation report.

[0007] As a preferred solution of the patient gait abnormality assessment auxiliary system based on human posture detection of the present invention, wherein: the motion trajectory of the patient's skeleton key points is identified according to the position coordinates of the skeleton key points, and the specific steps are as follows: Based on the key point coordinates of the patient's skeleton , generate the motion trajectory of the patient's skeleton key points in each gait cycle, the expression is: in, represents the motion trajectory of the i-th patient's skeleton key point in a gait cycle, represents the last time point of the gait cycle, It represents the number of data points of key point positions recorded in one gait cycle, where T is the gait cycle.

[0008] As a preferred solution of the patient gait abnormality assessment auxiliary system based on human posture detection of the present invention, wherein: based on the pre-processed patient gait video data, the key point data in the patient skeleton is annotated to generate the patient skeleton data set, and the specific steps are as follows: The pre-processed patient gait video data is segmented into a series of continuous frames using OpenCV, and the target detection algorithm SSD is used to identify the patient's outline in each frame. Using AlphaPose to identify the patient's outline, key points in the patient's skeleton are annotated, including the shoulder, elbow, hip, knee, and ankle. The location information of the key point data in each annotated patient skeleton and the corresponding timestamp are integrated to generate a patient skeleton dataset.

[0009] As a preferred solution of the patient gait abnormality assessment auxiliary system based on human posture detection of the present invention, wherein: the position coordinates of the key points of the patient skeleton are calculated based on the patient skeleton data set, and the specific steps are as follows: Extracting a position data series of the ankle joint from the patient skeleton data set, the data time series includes the vertical position of the ankle joint at each time point; By analyzing the vertical displacement changes of the ankle joint, the local minimum point in the vertical displacement of the ankle joint is identified and the gait cycle is defined; Based on the gait cycle, the position coordinates of the key points of the patient's skeleton are calculated, and the expression is: in, Indicates the skeleton key point of the i-th patient at time The location coordinates of Indicates the skeleton key point of the i-th patient at time The horizontal coordinate of Indicates the skeleton key point of the i-th patient at time The hammer position coordinates.

[0010] As a preferred solution of the patient gait abnormality assessment auxiliary system based on human posture detection of the present invention, wherein: the motion trajectory of the patient's skeleton key points is identified according to the position coordinates of the skeleton key points, and the specific steps are as follows: Based on the key point coordinates of the patient's skeleton , generate the motion trajectory of the patient's skeleton key points in each gait cycle, the expression is: in, represents the motion trajectory of the i-th patient's skeleton key point in a gait cycle, represents the last time point of the gait cycle, It represents the number of data points of key point positions recorded in one gait cycle, where T is the gait cycle.

[0011] As a preferred solution of the patient gait abnormality assessment auxiliary system based on human posture detection of the present invention, wherein: the patient gait features are extracted based on the motion trajectory of the patient's skeleton key points, and the specific steps are as follows: By calculating the distance from one foot landing to the next landing of the same foot in each gait cycle T, the stride length feature is extracted. By calculating the number of times the patient's footsteps move in each gait cycle T, the patient's step frequency characteristics are extracted; Through the position coordinate changes of the hip joint, elbow joint, knee joint and shoulder joint in the gait cycle T, the change angle features of the patient's hip joint, elbow joint, knee joint and shoulder joint are extracted; The patient's knee joint change angle characteristics are extracted through the position coordinates of the knee joint in the gait cycle T.

[0012] As a preferred solution of the patient gait abnormality assessment auxiliary system based on human posture detection of the present invention, wherein: the formation of the comprehensive gait feature vector comprises the following specific steps: Motion trajectory based on key points of the patient's skeleton , calculate the comprehensive gait feature F, the expression is: in, represents the stride length, Represents the step frequency characteristics, Indicates the change angle of the hip joint. Indicates the change angle of the knee joint Indicates the change angle of the elbow joint. Indicates the change angle of the shoulder joint.

[0013] As a preferred solution of the patient gait abnormality assessment auxiliary system based on human posture detection of the present invention, wherein: the patient skeleton data set is used to train a gait prediction model, and the specific steps are as follows: LSTM was used as the basic model and the patient skeleton dataset was used to train LSTM; Binary cross entropy is used as the loss function to quantify the difference between the model prediction value and the true label; The Adam optimizer is used to minimize the loss function and the model parameters are gradually adjusted until the model can accurately distinguish whether the gait is abnormal.

[0014] As a preferred solution of the patient gait abnormality assessment auxiliary system based on human posture detection of the present invention, wherein: the comprehensive gait feature vector is input into the gait prediction model to predict the patient's gait abnormality index, and the specific steps are as follows: The comprehensive gait feature vector F is input into the gait prediction model to predict the patient's gait abnormality index. The expression is: in, is the patient's gait abnormality index, is the weight matrix of the output layer of the gait prediction model, is the bias term of the output layer, is the weight matrix of the fully connected layer of the gait prediction model, is the bias term of the fully connected layer, and F is the comprehensive gait feature vector.

[0015] As a preferred solution of the patient gait abnormality assessment auxiliary system based on human posture detection of the present invention, wherein: the prediction results are evaluated and an evaluation report is generated, and the specific steps are as follows: Based on the patient skeleton data set, a gait abnormality standard threshold I is defined, and the gait abnormality index p is compared with the gait abnormality standard threshold I to evaluate the patient's gait; when When >I, the patient’s gait is judged to be abnormal; when When ≤I, the patient’s gait is judged to be normal; A patient gait assessment report is generated based on the patient's gait assessment results.

[0016] The beneficial effects of the present invention are as follows: through the data acquisition and preprocessing module, the quality of the original video data is improved, and the key points of the patient's skeleton are marked; through the skeleton key point marking module, the accurate marking of the patient's skeleton key points is achieved, and the patient skeleton data set is generated; through the skeleton key point trajectory recognition module, the accurate recognition of the motion trajectory of the patient's skeleton key points is achieved; through the gait feature extraction module, the comprehensive extraction of the patient's gait features is achieved to form a comprehensive gait feature vector; through the gait abnormality index prediction module, the gait prediction is achieved by using the characteristics of time series data, and the accuracy of gait abnormality assessment is improved; through the evaluation and report generation module, a comprehensive evaluation of the patient's gait status is achieved, and a detailed evaluation report is generated, which ultimately improves the accuracy and efficiency of clinical diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0018] Figure 1 This is a flow chart of the patient gait abnormality assessment auxiliary system based on human posture detection in Example 1.

[0019] Figure 2 This is a flow chart for evaluating a patient's gait and generating an evaluation report in Example 1. DETAILED DESCRIPTION

[0020] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0021] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0023] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a patient gait abnormality assessment auxiliary system based on human posture detection, comprising the following steps: S1. Collect patient gait video data, pre-process the patient gait video data and mark the patient skeleton key points.

[0024] Furthermore, a medical monitoring camera is used to record a video of the patient in a normal walking state. The video should contain data of more than one walking cycle of the patient to ensure that the sample size is rich enough. The data of multiple walking cycles refers to video data containing multiple complete gait cycles of the patient (from one foot landing to the same foot landing again); The collected patient gait video data includes video size standardization, denoising and contrast enhancement.

[0025] It should also be noted that size normalization refers to adjusting the acquired video frames to a uniform resolution to ensure that all video frames have the same size. This process is achieved by using the resampling function in an image processing library (such as OpenCV) to scale each video frame to a predetermined standard resolution (e.g. 640x480 pixels) to ensure consistency and efficiency in subsequent processing steps.

[0026] Denoising refers to the process of removing random noise from video frames through algorithms to improve image quality. This process specifically includes applying Gaussian blur and median filtering techniques to reduce high-frequency noise in video frames, making the image clearer and smoother, thereby improving the accuracy of subsequent feature detection.

[0027] Contrast enhancement refers to making the details in the image more prominent by adjusting the brightness and contrast of the image. This process is done by using histogram equalization or contrast-limited adaptive histogram equalization (CLAHE) technology to enhance the contrast of the image, thereby making the important features in the image more obvious and easier to process and analyze later.

[0028] S2: Based on the preprocessed patient gait video data, the key point data in the patient skeleton is annotated to generate a patient skeleton dataset.

[0029] Furthermore, the pre-processed patient gait video data is segmented into a series of continuous frames through OpenCV, and the target detection algorithm SSD is used to identify the patient's outline in each frame; for example, suppose there is a video with a resolution of 640x480 and 30 frames per second. Use OpenCV to read the video file and extract the image frame by frame. For each frame, use the SSD model to detect the patient's body outline. Assume that in a certain frame, the SSD model recognizes that the patient's outline is located in the upper left corner of the image and draws a rectangular box that contains the patient's entire body. Finally, the patient's outline information in each frame is obtained.

[0030] AlphaPose is used to mark the key points of the patient's skeleton on the identified patient's outline, including the shoulder, elbow, hip, knee and ankle; for example, in the above frame image, the AlphaPose model will mark five key points on the detected patient's outline: shoulder, hip, knee and ankle. Assume that the coordinates of the patient's left shoulder in the image are (100, 150), the right shoulder is (200, 150), the left elbow is (120, 250), the right elbow is 180, 250, and so on. AlphaPose will generate such a labeling data containing the coordinates of the key points for each frame of the image.

[0031] The location information of the key point data in each annotated patient skeleton and the corresponding timestamp are integrated to generate a patient skeleton dataset.

[0032] It should also be noted that the position information of the key point data in each annotated patient skeleton and the corresponding timestamp are integrated to generate a patient skeleton dataset. This can ensure that the dynamic characteristics of gait over time are captured, facilitate time series analysis and feature extraction, and thus improve the accuracy and reliability of gait assessment.

[0033] S3. Based on the patient skeleton data set, the position coordinates of the patient skeleton key points are calculated, and the motion trajectory of the patient skeleton key points is identified according to the position coordinates of the skeleton key points.

[0034] Furthermore, the ankle joint position data sequence is extracted from the patient skeleton data set, and the data time series contains the vertical position of the ankle joint at each time point; the specific process is: according to each frame timestamp and the corresponding ankle joint position coordinates recorded in the patient skeleton data set, the vertical position coordinates of the ankle joint at each time point are extracted to form a vertical position data sequence arranged in chronological order, so as to further analyze the vertical displacement changes of the ankle joint.

[0035] By analyzing the changes in the vertical displacement of the ankle joint, the local minimum points in the vertical displacement of the ankle joint are identified. These points represent the moment when the foot touches the ground and define the gait cycle. The specific process is: first, the vertical position data sequence of the ankle joint at each time point is obtained, and then the noise is removed by smoothing the sequence. Then, the peak detection algorithm is used to find the local minimum points in the vertical displacement curve. These minimum points correspond to the moment when the foot touches the ground, and the time period between two adjacent minimum points is defined as a gait cycle.

[0036] Based on the gait cycle, the position coordinates of the key points of the patient's skeleton are calculated, and the expression is: in, Indicates the skeleton key point of the i-th patient at time The location coordinates of Indicates the skeleton key point of the i-th patient at time The horizontal coordinate of Indicates the skeleton key point of the i-th patient at time The hammer position coordinates.

[0037] Based on the key point coordinates of the patient's skeleton , generate the motion trajectory of the patient's skeleton key points in each gait cycle, the expression is: in, represents the motion trajectory of the i-th patient's skeleton key point in a gait cycle, represents the last time point of the gait cycle, It represents the number of data points of key point positions recorded in one gait cycle, where T is the gait cycle.

[0038] It should also be noted that based on the coordinates of the patient's skeleton key points, the motion trajectory of the patient's skeleton key points in each gait cycle is generated. The specific process is: first determine the start and end time points of the gait cycle, then extract the position coordinates of each key point in these time points from the patient's skeleton data set, and connect these coordinates in chronological order to form a continuous position change path for each key point in the gait cycle, thereby generating the motion trajectory within the gait cycle, which helps to analyze gait characteristics and evaluate whether the gait is abnormal.

[0039] S4. Based on the motion trajectory of the patient's skeleton key points, the patient's gait features are extracted to form a comprehensive gait feature vector.

[0040] Furthermore, the stride length feature is extracted by calculating the distance of the patient's ankle joint from one foot landing to the same foot landing again in each gait cycle T. The expression is: ; in, represents the stride length, Indicates the start time of the gait cycle, represents the end time point of the gait cycle, t represents the time point within the gait cycle, Indicates the ankle joint at time The location coordinates of Indicates the ankle joint at time The location coordinates of By calculating the number of footsteps of the patient in each gait cycle T, the patient's gait frequency characteristics are extracted (gait frequency v = number of footsteps in gait cycle T / gait cycle time length); Through the position coordinate changes of the hip joint, elbow joint, knee joint and shoulder joint in the gait cycle T, the change angle features of the patient's hip joint, elbow joint, knee joint and shoulder joint are extracted; Specifically: Taking the hip joint as an example, the expression of the change angle of the hip joint in the gait cycle T is: in, It's in time The angle of the hip joint changes when In time The position coordinates of the hip joint at It's in time The position coordinates of the knee joint at It's in time The position coordinates of the ankle joint at time; The patient's knee joint change angle characteristics are extracted through the position coordinates of the knee joint in the gait cycle T.

[0041] Motion trajectory based on key points of the patient's skeleton , calculate the comprehensive gait feature F, the expression is: in, represents the stride length, Represents the step frequency characteristics, Indicates the change angle of the hip joint. Indicates the change angle of the knee joint Indicates the change angle of the elbow joint. Indicates the change angle of the shoulder joint.

[0042] It should also be noted that constructing a comprehensive gait feature vector can fully capture the patient's gait characteristics, thereby more accurately assessing whether the patient's gait is abnormal, which helps to improve the accuracy and reliability of gait analysis.

[0043] S5. Use the patient skeleton data set to train a gait prediction model, input the comprehensive gait feature vector into the gait prediction model, and predict the patient's gait abnormality index.

[0044] Furthermore, LSTM was adopted as the basic model and the patient skeleton dataset was used to train LSTM; It should be noted that LSTM (Long Short-Term Memory Network) is used as the basic model because LSTM can capture the temporal dependencies in gait data, thereby more accurately identifying gait patterns and potential abnormalities; Binary cross entropy is used as a loss function to quantify the difference between the model prediction value and the true label; the specific process is: calculate the difference between the probability value predicted by the model and the actual label (normal or abnormal), and measure this difference through the binary cross entropy loss function, so that the model can learn the correct classification boundary and optimize the prediction accuracy. For example, if the model predicts that the probability of a certain step being abnormal is 0.8, and the actual label is abnormal (label is 1), the binary cross entropy loss function will calculate the difference between this prediction and the actual label, and adjust the model parameters accordingly.

[0045] The Adam optimizer is used to minimize the loss function and gradually adjust the model parameters until the model can accurately distinguish whether the gait is abnormal. The specific process is: during the training process, the Adam optimizer adjusts the model parameters according to the size and direction of the gradient, accelerates convergence through the adaptive learning rate method, and gradually optimizes the model parameters until the model can accurately distinguish whether the gait is abnormal. For example, the Adam optimizer will adaptively adjust the learning rate according to the historical gradient information of each parameter, so as to find the minimum value of the loss function more quickly.

[0046] The comprehensive gait feature vector F is input into the gait prediction model to predict the patient's gait abnormality index. The expression is: in, is the patient's gait abnormality index, is the weight matrix of the output layer of the gait prediction model, is the bias term of the output layer of the gait prediction model, is the weight matrix of the fully connected layer of the gait prediction model, is the bias term of the fully connected layer of the gait prediction model, and F is the comprehensive gait feature vector; It should be noted that the process of calculating the patient's gait abnormality index through a multi-layer neural network, in which the input feature F is linearly transformed in the first layer And the activation function After processing, it undergoes the second layer of linear transformation , and finally the Sigmoid function outputs a value between 0 and 1, which represents the patient's gait abnormality index.

[0047] S6. Evaluate the prediction results and generate an evaluation report.

[0048] Furthermore, based on the patient skeleton data set, a gait abnormality standard threshold I is defined, and the gait abnormality index p is compared with the gait abnormality standard threshold I to evaluate the patient's gait; when When >I, the patient’s gait is judged to be abnormal; when When ≤I, the patient’s gait is judged to be normal; A patient gait assessment report is generated based on the patient's gait assessment results.

[0049] It should also be noted that the patient gait assessment report should include the patient ID, gait abnormality index, gait status ("normal" or "abnormal") and recommended measures, for example, when the patient's gait is abnormal, further medical examination and treatment are recommended, and when the patient's gait is normal, it is recommended to continue monitoring and maintain the current health status.

[0050] In summary, the present invention achieves quality improvement of original video data and marks patient skeleton key points through: data acquisition and preprocessing module; through skeleton key point marking module, accurate marking of patient skeleton key points is achieved and patient skeleton data set is generated; through skeleton key point trajectory recognition module, accurate recognition of patient skeleton key point motion trajectory is achieved; through gait feature extraction module, comprehensive extraction of patient gait features is achieved to form a comprehensive gait feature vector; through gait abnormality index prediction module, gait prediction is achieved using time series data characteristics to improve the accuracy of gait abnormality assessment; through evaluation and report generation module, comprehensive evaluation of patient gait status is achieved, detailed evaluation report is generated, and ultimately the accuracy and efficiency of clinical diagnosis are improved.

[0051] Example 2, referring to Table 1, is the second example of the present invention. To further verify the technical solution of the present invention, experimental simulation data of a patient gait abnormality assessment auxiliary system based on human posture detection is provided.

[0052] The experimental subjects were a group of healthy adults aged between 25 and 35 years old. This age group was chosen because people in this age group usually have a more stable gait pattern, which is conducive to the reliability and repeatability of the experimental results.

[0053] First, the experiment used a high-definition medical surveillance camera to record videos of participants in a normal walking state, and performed preprocessing to improve image quality; Secondly, we used image processing technology and machine learning algorithms to mark the key points of the participants’ skeletons and generate a skeleton dataset; Next, gait features were extracted based on the skeleton dataset, and the gait abnormality index was calculated; Finally, the comprehensive gait feature vector was input into the LSTM model to predict the participant’s gait status and compared with the set threshold to determine whether the gait was abnormal.

[0054] The details are shown in Table 1 below: Table 1 Patient gait data analysis table As can be seen from the table above, the stride length of all test subjects was between 73.8 cm and 77.0 cm, and the step frequency was between 1.23 Hz and 1.32 Hz. The hip joint change angle was distributed between 29.7 degrees and 31.4 degrees, and the knee joint change angle was between 54.5 degrees and 56.1 degrees. Based on these gait characteristics, the gait abnormality index of each participant was calculated by the LSTM model and compared with the standard threshold of gait abnormality. The results showed that the gait abnormality index of some participants exceeded the set standard threshold of gait abnormality (assuming the threshold is 0.33), so the gait of these participants was judged to be abnormal.

[0055] The patient gait abnormality assessment auxiliary system based on human posture detection of the present invention can not only accurately capture and quantify the patient's gait characteristics, but also use machine learning technology to identify subtle gait abnormalities, which reflects the significant advantages of the present invention in improving the accuracy and reliability of gait assessment.

[0056] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A patient gait abnormality assessment auxiliary system based on human posture detection, characterized in that: include, Data acquisition and preprocessing module: collect patient gait video data, preprocess the patient gait video data and mark the key points of the patient skeleton; Skeleton key point labeling module: Based on the preprocessed patient gait video data, label the key point data in the patient skeleton and generate the patient skeleton dataset; Skeleton key point trajectory recognition module: Based on the patient skeleton data set, the position coordinates of the patient's skeleton key points are calculated, and the movement trajectory of the patient's skeleton key points is identified according to the position coordinates of the skeleton key points; Gait feature extraction module: based on the motion trajectory of the patient's skeleton key points, extract the patient's gait features to form a comprehensive gait feature vector; Gait abnormality index prediction module: Use the patient skeleton data set to train a gait prediction model, input the comprehensive gait feature vector into the gait prediction model, and predict the patient's gait abnormality index; Evaluation and report generation module: evaluates the prediction results and generates an evaluation report.

2. The patient gait abnormality assessment auxiliary system based on human posture detection as claimed in claim 1, characterized in that: The specific steps of collecting patient gait video data, preprocessing the patient gait video data and marking the patient skeleton key points are as follows: Use medical surveillance cameras to record videos of patients walking normally; The collected patient gait video data includes video size standardization, denoising and contrast enhancement.

3. The patient gait abnormality assessment auxiliary system based on human posture detection as claimed in claim 2, characterized in that: Based on the pre-processed patient gait video data, the key point data in the patient skeleton is annotated to generate the patient skeleton data set. The specific steps are as follows: The pre-processed patient gait video data is segmented into a series of continuous frames using OpenCV, and the target detection algorithm SSD is used to identify the patient's outline in each frame. Using AlphaPose to identify the patient's outline, key points in the patient's skeleton are annotated, including the shoulder, elbow, hip, knee, and ankle. The location information of the key point data in each annotated patient skeleton and the corresponding timestamp are integrated to generate a patient skeleton dataset.

4. The patient gait abnormality assessment auxiliary system based on human posture detection as claimed in claim 3, characterized in that: The specific steps of calculating the position coordinates of the key points of the patient skeleton based on the patient skeleton data set are as follows: Extracting a position data series of the ankle joint from the patient skeleton data set, the data time series includes the vertical position of the ankle joint at each time point; By analyzing the vertical displacement changes of the ankle joint, the local minimum point in the vertical displacement of the ankle joint is identified and the gait cycle is defined; Based on the gait cycle, the position coordinates of the key points of the patient's skeleton are calculated, and the expression is: in, Indicates the skeleton key point of the i-th patient at time The location coordinates of Indicates the skeleton key point of the i-th patient at time The horizontal coordinate of Indicates the skeleton key point of the i-th patient at time The coordinates of the hammer position.

5. The patient gait abnormality assessment auxiliary system based on human posture detection as claimed in claim 4, characterized in that: The specific steps of identifying the motion trajectory of the patient's skeleton key points according to the position coordinates of the skeleton key points are as follows: Based on the key point coordinates of the patient's skeleton , generate the motion trajectory of the patient's skeleton key points in each gait cycle, the expression is: in, represents the motion trajectory of the i-th patient's skeleton key point in a gait cycle, represents the last time point of the gait cycle, It represents the number of data points of key point positions recorded in one gait cycle, where T is the gait cycle.

6. The patient gait abnormality assessment auxiliary system based on human posture detection as claimed in claim 5, characterized in that: The specific steps of extracting the patient's gait features based on the patient's skeleton key point motion trajectory are as follows: By calculating the distance from one foot landing to the next landing of the same foot in each gait cycle T, the stride length feature is extracted. By calculating the number of times the patient's footsteps move in each gait cycle T, the patient's step frequency characteristics are extracted; Through the position coordinate changes of the hip joint, elbow joint, knee joint and shoulder joint in the gait cycle T, the change angle features of the patient's hip joint, elbow joint, knee joint and shoulder joint are extracted; The patient's knee joint change angle characteristics are extracted through the position coordinates of the knee joint in the gait cycle T.

7. The patient gait abnormality assessment auxiliary system based on human posture detection according to claim 6, characterized in that: The steps of forming a comprehensive gait feature vector are as follows: Motion trajectory based on key points of the patient's skeleton , calculate the comprehensive gait feature F, the expression is: in, represents the stride length, Represents the step frequency characteristics, Indicates the change angle of the hip joint. Indicates the change angle of the knee joint Indicates the change angle of the elbow joint. Indicates the change angle of the shoulder joint.

8. The patient gait abnormality assessment auxiliary system based on human posture detection as claimed in claim 7, characterized in that: The patient skeleton dataset is used to train a gait prediction model. The specific steps are as follows: LSTM was used as the basic model and the patient skeleton dataset was used to train LSTM; Binary cross entropy is used as the loss function to quantify the difference between the model prediction value and the true label; The Adam optimizer is used to minimize the loss function and the model parameters are gradually adjusted until the model can accurately distinguish whether the gait is abnormal.

9. The patient gait abnormality assessment auxiliary system based on human posture detection as claimed in claim 8, characterized in that: The comprehensive gait feature vector is input into the gait prediction model to predict the patient's gait abnormality index. The specific steps are as follows: The comprehensive gait feature vector F is input into the gait prediction model to predict the patient's gait abnormality index. The expression is: in, is the patient's gait abnormality index, is the weight matrix of the output layer of the gait prediction model, is the bias term of the output layer, is the weight matrix of the fully connected layer of the gait prediction model, is the bias term of the fully connected layer, and F is the comprehensive gait feature vector.

10. The patient gait abnormality assessment auxiliary system based on human posture detection according to claim 9, characterized in that: The prediction results are evaluated and an evaluation report is generated. The specific steps are as follows: Based on the patient skeleton data set, a gait abnormality standard threshold I is defined, and the gait abnormality index p is compared with the gait abnormality standard threshold I to evaluate the patient's gait; when When >I, the patient’s gait is judged to be abnormal; when When ≤I, the patient’s gait is judged to be normal; A patient gait assessment report is generated based on the patient's gait assessment results.

Images