ADL ability training data management system for patients with neurological diseases

Abstract

The present invention relates to the field of medical informatics, and more specifically, to a data management system for ADL ability training for patients with neurological diseases. The system comprises a memory and a processor, wherein the processor executes a computer program stored in the memory to implement the following steps: obtaining posture images and scales of normal individuals and patients with neurological diseases during ADL ability training; evaluating the likelihood that the image to be evaluated and the reference image represent the same action based on the positional, depth, and temporal differences between the joint points in the image to be evaluated and those in the reference image, and selecting reference images; obtaining a movement completion index by combining the likelihood of the image to be evaluated and all reference images representing the same action and the time at which the image to be evaluated was captured; and determining the effective evaluation value of the training based on the difference in the changing characteristics of each indicator category with other indicators at each moment in the neurological disease patient's training, as well as each indicator category. The present invention can obtain accurate comprehensive results of ADL ability training data.

Description

Technical Field

[0001] The present invention relates to the technical field of medical informatics, and in particular to an ADL ability training data management system based on patients with neurological diseases. Background Art

[0002] Due to factors such as decreased muscle control and impaired balance, patients with neurological diseases suffer from certain motor dysfunctions and are unable to complete daily living activities stably and smoothly. Therefore, ADL ability training is usually needed to improve patients' fine motor skills and restore or improve their daily living abilities. ADL ability training usually includes training in muscle and endurance building, dressing, and eating. By managing and analyzing the data of each patient's ADL training (such as the data corresponding to each functional scale), dynamic monitoring of the patient's ADL ability can be achieved, allowing medical staff to intervene in the patient's rehabilitation training plan in a timely manner according to changes in ADL ability to ensure that the plan meets the patient's actual needs.

[0003] Existing methods for analyzing the ADL ability of patients with neurological diseases usually evaluate the patient's recovery based on a scale method (for example, the simplified upper limb FMA scoring scale is used to evaluate the patient's upper limb movements). The patient is asked to complete the specified training movements and the degree of completion of the patient's movements is observed to evaluate the patient's current ADL ability recovery. However, this method cannot accurately obtain subtle changes in the patient's limb control, making it difficult to comprehensively evaluate the ADL ability training data of patients with neurological diseases. Summary of the Invention

[0004] In order to solve the problem that existing methods are difficult to comprehensively evaluate the ADL ability training data of patients with neurological diseases, the purpose of the present invention is to provide an ADL ability training data management system for patients with neurological diseases. The technical solutions adopted are as follows:

[0005] The present invention provides an ADL ability training data management system for patients with neurological diseases, comprising a memory and a processor, wherein the processor executes a computer program stored in the memory to implement the following steps:

[0006] Obtain posture images and different types of scales during ADL ability training of normal people and patients with neurological diseases;

[0007] evaluating, based on differences in positional distribution, depth, and corresponding relative occurrence times between joint points in the image to be evaluated and the reference image, a likelihood value that the image to be evaluated and the reference image represent the same action, wherein the image to be evaluated is a posture image of a patient with a neurological disease and the reference image is a posture image of a normal person; and using the likelihood value to select a reference image for the image to be evaluated;

[0008] The probability that the image to be evaluated and all control images represent the same action, as well as the time at which the image to be evaluated was captured, is used to determine the movement completion index for each moment in the neurological disease patient. Based on the differences in the changing characteristics of each indicator type and other indicators at each moment in the neurological disease patient, as well as each indicator type, the coordination coefficient for each moment is determined. The indicators include the movement completion index and the score values ​​of different scales.

[0009] The effective evaluation value of the training was determined based on the changes in the coordination coefficient during the ADL ability training of patients with neurological diseases.

[0010] Preferably, the evaluating the likelihood value of the image to be evaluated and the reference image representing the same action based on the difference in position distribution, depth difference, and corresponding relative occurrence time between the joint points in the image to be evaluated and the reference image includes:

[0011] For any joint point: the ratio of the Euclidean distance between the joint point and the coordinate center point of the human body region where the joint point is located to the Euclidean distance between the two joint points with the greatest distance in the human body region is determined as the distance index of the joint point; the direction vector from the joint point to its adjacent joint point is used as the direction vector of the joint point; the ratio of the depth value of the joint point to the mean depth value of all relevant nodes in the image where the joint point is located is determined as the depth index of the joint point;

[0012] Obtaining a matching value between the candidate joint point and each joint point in the reference image based on a cosine value between a direction vector of the candidate joint point and a direction vector of each joint point in the reference image, a difference in a distance index between the candidate joint point and each joint point in the reference image, and a difference in a depth index, wherein the cosine value is positively correlated with the matching value, and the difference in the distance index and the difference in the depth index are both negatively correlated with the matching value; determining the joint point in the reference image with the largest matching value with the candidate joint point as the matching joint point of the candidate joint point; the candidate joint point is any joint point in the image to be evaluated;

[0013] Based on the difference in relative occurrence time between the image to be evaluated and the reference image where the candidate joint point is located, and the matching value of all matching joint point pairs in the image to be evaluated and the reference image where the candidate joint point is located, the possibility value that the image to be evaluated and the reference image are the same action is obtained; wherein the matching joint point pairs are composed of joint points in the image to be evaluated and joint points in the reference image that match the joint points in the image to be evaluated.

[0014] Preferably, obtaining the probability value that the image to be evaluated and the reference image are the same action according to the difference in relative occurrence time between the image to be evaluated where the candidate joint point is located and the reference image, and the matching value of all matching joint point pairs in the image to be evaluated where the candidate joint point is located and the reference image, comprises:

[0015] The ratio of the time between the initial moment and the acquisition moment of the image where the joint point is located to the total time to complete the current ADL ability training is recorded as the duration ratio of the corresponding image;

[0016] Based on the difference in the duration ratio between the image to be evaluated and the reference image where the candidate joint point is located, and the average of the matching values ​​of all matching joint point pairs formed by the joint points in the image to be evaluated and the reference image where the candidate joint point is located, the possibility value of the image to be evaluated and the reference image being the same action is obtained. The average of the matching values ​​is positively correlated with the possibility value, and the difference in the duration ratio is negatively correlated with the possibility value.

[0017] Preferably, the method of screening a reference image of the image to be evaluated by using the likelihood value includes:

[0018] For any normal person: the reference image corresponding to the maximum likelihood value is determined as the control image of the image to be evaluated.

[0019] Preferably, the probability value of the image to be evaluated and all the control images being the same action and the acquisition time of the image to be evaluated are integrated to obtain the movement completion index of the neurological disease patient at each moment, including:

[0020] For any image to be evaluated:

[0021] Calculating a first average value of the likelihood values ​​of any image to be evaluated and all its reference images representing the same action;

[0022] Calculate the product of the first average value and the number of frames of the control image of any image to be evaluated of the neurological disease patient; and determine the ratio of the product to the detection time corresponding to the posture image of the ADL ability training process of the neurological disease patient as the movement completion index of the neurological disease patient at the time of acquisition of any image to be evaluated.

[0023] Preferably, determining the coordination coefficient at each moment based on the difference in the change characteristics of each type of indicator and other indicators of the neurological disease patient at each moment and each type of indicator includes:

[0024] For any type of indicator: the difference between the indicator of the neurological disease patient at the next moment after the time to be analyzed and the indicator of the any type at the time to be analyzed is used as the change value of the indicator of the any type at the time to be analyzed; if the change value is a negative number, the time to be analyzed is determined to be the time of abnormal fluctuation of the indicator of the any type;

[0025] Mark any one type of index as a first type of index, and mark any other type of index except the first type of index as a second type of index; record the number of moments before and before the time to be analyzed when both the first type of index and the second type of index are at abnormal fluctuation moments as a first number, and record the number of moments before and after the time to be analyzed when the first type of index is at abnormal fluctuation moments as a second number; determine the ratio of the first number to the second number as the abnormal consistency coefficient between the first type of index and the second type of index at the time to be analyzed;

[0026] The correlation coefficient between the first category indicator and all other indicators at the time to be analyzed is obtained based on the abnormal consistency coefficient between the first category indicator and each other category of indicators at the time to be analyzed and before, and the difference in change values ​​between the first category indicator and the corresponding other indicators at the time of abnormal fluctuation.

[0027] According to the correlation coefficient of each type of indicator at the time to be analyzed and all other indicators, as well as each type of indicator at the time to be analyzed, the coordination coefficient of the time to be analyzed is obtained;

[0028] The time to be analyzed is any time during the ADL ability training process of the neurological disease patient.

[0029] Preferably, obtaining the coordination coefficient of the time to be analyzed based on the correlation coefficient of each type of indicator at the time to be analyzed and all other indicators, as well as each type of indicator at the time to be analyzed, includes:

[0030] The product of each type of indicator at the time to be analyzed and the correlation coefficient between each type of indicator and all other indicators is recorded as the first eigenvalue of each type of indicator at the time to be analyzed;

[0031] The cumulative sum of the first characteristic values ​​of all indicators at the time to be analyzed is determined as the coordination coefficient at the time to be analyzed.

[0032] Preferably, determining the effective evaluation value of the training according to the change of the coordination coefficient during the ADL ability training of the neurological disease patient includes:

[0033] The ratio of the coordination coefficient at the time to be analyzed to the coordination coefficient at the moment before the time to be analyzed is determined as the coordination ability growth index at the time to be analyzed; if the coordination ability growth index is greater than 1, the time to be analyzed is regarded as an increasing time; if the coordination ability growth index is less than or equal to 1, the time to be analyzed is regarded as a decreasing time;

[0034] A recovery index is obtained based on the time distribution of the growth period and the relative size relationship between the range of the growth index of the growth period and the range of the growth index of the decrease period; the growth period is composed of continuous growth moments; and the decrease period is composed of continuous decrease moments;

[0035] The effective evaluation value of the training is evaluated by combining the relative magnitude relationship between the coordination coefficient at the initial moment of the ADL ability training and the coordination coefficient at the current moment and the recovery index.

[0036] Preferably, the recovery index is obtained according to the time distribution of the growth period and the relative size relationship between the range of the growth index of the growth period and the range of the growth index of the reduction period, including:

[0037] Calculating a first ratio between an average value of the range of growth indices of all growth periods and an average value of the range of growth indices of all reduction periods, and a second ratio between a total duration of all growth periods and a total duration of the neurological disease patient detection process;

[0038] The recovery index is determined according to the maximum time interval between adjacent growth periods, the first ratio and the second ratio, wherein both the first ratio and the second ratio are positively correlated with the recovery index, and the maximum time interval is negatively correlated with the recovery index.

[0039] Preferably, the evaluation of the effective assessment value of the training based on the relative magnitude relationship between the coordination coefficient at the initial moment of the ADL ability training and the coordination coefficient at the current moment and the recovery index includes:

[0040] calculating a third ratio between the coordination coefficient at the initial moment of the ADL ability training and the coordination coefficient at the current moment;

[0041] The product of the third ratio and the recovery index is determined as an effective evaluation value of the training.

[0042] The present invention has at least the following beneficial effects:

[0043] The present invention first evaluates the possibility that the image to be evaluated and the reference image are the same action based on the differences in position distribution, depth difference, and corresponding relative occurrence time between the joint points in the posture images of the neurological disease patients during ADL ability training and the posture images of normal people. That is, the temporal features and the features presented by the joint points in the images are comprehensively considered to judge the possibility that the joint points in the posture images of the neurological disease patients during ADL ability training and the posture images of normal people belong to the same action. The possibility value is used to screen the control images that belong to the same action as the posture images of the neurological disease patients during ADL ability training. Then, the movement completion status of the neurological disease patients is evaluated, and the movement completion index at each moment is obtained. The movement completion index and the score values ​​of different types of scales are used as different indicators. According to the difference in the change characteristics of each type of indicator and other indicators at each moment of the neurological disease patients and the numerical distribution of each type of indicator, the effective evaluation value of the neurological disease patient training is determined, thereby realizing a comprehensive evaluation of the ADL ability training data of the neurological disease patients, improving the accuracy of the evaluation of the patient's ADL ability training data, and providing a reference for medical staff. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0045] Figure 1 This is a flowchart of a method executed by a data management system for ADL ability training of neurological disease patients provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the ADL ability training data management system for patients with neurological diseases proposed in accordance with the present invention is described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0047] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0048] The specific solution of the ADL ability training data management system for neurological disease patients provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0049] Example of ADL ability training data management system based on neurological disease patients:

[0050] The specific scenario targeted by this embodiment is: in the process of training the ADL ability of patients with neurological diseases, the patient's movement posture and scale data during training are analyzed in combination with the posture change characteristics of normal people when performing corresponding movements, and the patient's limb coordination ability during training is evaluated based on the correlation between various indicators. Then, the patient's ADL training data is comprehensively evaluated in combination with the fluctuation degree of the patient's limb coordination ability recovery, thereby providing a reference for medical staff.

[0051] This embodiment proposes an ADL ability training data management system for patients with neurological diseases. The system includes a memory and a processor. The processor executes a computer program stored in the memory to achieve the following: Figure 1 The specific steps are as follows:

[0052] Step S1 : obtaining posture images and different types of scales during the ADL ability training process of normal people and patients with neurological diseases.

[0053] First, a Kinect sensor camera is used to capture the posture image and depth image of the front of the body of a patient with a neurological disease during the ADL ability training process, and a Kinect sensor camera is used to capture the posture image and depth image of multiple normal people while they are doing the same action, wherein the camera is calibrated based on the Zhang calibration method. It should be noted that normal people are healthy people who do not suffer from neurological diseases, and the posture images collected in this embodiment are RGB images. In this embodiment, the number of normal people is 5. In specific applications, the implementer can set it according to the specific situation. The posture image corresponds one to one with the pixels in the depth image. This embodiment collects the depth image in order to obtain the depth value of each position in the posture image.

[0054] At the same time, different scales are obtained for neurological disease patients and healthy individuals during ADL training. In this embodiment, the scales include the FMA rating scale and the Barthel index scale. The FMA rating scale is used to assess the individual's limb function, and the Barthel index scale is used to assess the individual's ability to perform daily activities. It should be noted that these scales are assessed by medical staff, and multiple scale assessments are conducted for both neurological disease patients and healthy individuals. The frequency of posture image acquisition and scale assessment is set by the implementer based on specific circumstances.

[0055] Thus, this embodiment has acquired posture images, depth values ​​at each position in the posture images, and different types of scales during ADL training for multiple healthy individuals and patients with neurological disorders. It should be noted that this embodiment uses a patient with a neurological disorder as an example; the method provided in this embodiment can also be used to process patients with other neurological disorders.

[0056] Step S2, based on the difference in position distribution, depth difference, and corresponding relative occurrence time between the joint points in the image to be evaluated and the reference image, evaluate the likelihood that the image to be evaluated and the reference image are the same action, wherein the image to be evaluated is a posture image of a patient with a neurological disease, and the reference image is a posture image of a normal person; and use the likelihood value to screen a control image for the image to be evaluated.

[0057] For each collected pose image, we convert it to grayscale. Based on the grayscale results, we semantically identify the human body region. We then use the Canny operator to obtain the human body edge contour, i.e., the outermost edge. All pixels within the contour constitute the human body region. The Kincet SDK is then used to obtain multiple joint points within the human body region. This method allows us to extract multiple joint points from each pose image. The number of joint points in different pose images is the same.

[0058] Neurological disease patients often suffer from decreased joint and muscle coordination and control, as well as poor balance. This can lead to them being unable to accurately and quickly complete designated movements, resulting in significant deviations from the target trajectory, discontinuities in movement completion, and slow completion speeds. Therefore, we analyze the differences between the movement trajectory of neurological disease patients and the target trajectory using collected posture images to assess the degree to which the patient's movement trajectory meets the standard. The target trajectory is the trajectory of a normal person with no muscle impairment.

[0059] Ideally, when different examinees complete the same action, the timing of their corresponding images in the overall video of the examinee is similar, the positions and depth information of their different joints relative to themselves are relatively similar, and the limb rotation angles are also relatively similar. Therefore, based on the above analysis, the similarity between the posture images of patients with neurological diseases and those of normal people is analyzed to determine the possibility that the corresponding two frames of images are the same action, and then screen the reference images that belong to the same action as each posture image of the patient with neurological disease.

[0060] Specifically, for any joint point in any posture image: first, the ratio of the Euclidean distance between the joint point and the coordinate center point of the human body region in which it is located to the Euclidean distance between the two joint points with the greatest distance in the human body region is determined as the distance index of the joint point; the direction vector from the joint point to its adjacent joint point is used as the direction vector of the joint point; it should be noted that: if the number of adjacent joint points of the joint point is greater than 1, the direction vector from the joint point to each of its adjacent joint points is obtained separately, and the sum of these direction vectors is used as the direction vector of the joint point. Then, the ratio between the depth value of the joint point and the depth mean of all relevant nodes in the image where the joint point is located is determined as the depth index of the joint point; the depth mean of all relevant nodes in a posture image is the average value of the depth values ​​of all relevant nodes in the posture image. Using this method, the direction vector and depth index of each joint point in each collected posture image can be obtained.

[0061] For patients with neurological diseases, multiple posture images were collected. For each healthy person, multiple posture images were also collected. All posture images of patients with neurological diseases were recorded as images to be evaluated, and the posture images of healthy people were recorded as reference images. In other words, there are multiple images to be evaluated and multiple reference images.

[0062] Next, this embodiment is described by taking an image to be evaluated and a reference image as an example. Other images to be evaluated and other reference images can be processed using the method provided in this embodiment.

[0063] Specifically, any joint point in the image to be evaluated is recorded as a candidate joint point, and the matching value of the candidate joint point and each joint point in the reference image is obtained according to the cosine value between the direction vector of the candidate joint point and the direction vector of each joint point in the reference image, the difference in the distance index between the candidate joint point and each joint point in the reference image, and the difference in the depth index. The cosine value is positively correlated with the matching value, and the difference in the distance index and the difference in the depth index are both negatively correlated with the matching value.

[0064] In this embodiment, the difference in distance indices of two joint points is the absolute value of the difference between the distance indices of the two joint points, and the difference in depth indices of two joint points is the absolute value of the difference between the depth indices of the two joint points.

[0065] Among them, a positive correlation relationship indicates that the dependent variable will increase as the independent variable increases, and the dependent variable will decrease as the independent variable decreases. It can be an additive relationship, a multiplicative relationship, etc., which is determined by actual application; a negative correlation relationship indicates that the dependent variable will decrease as the independent variable increases, and the dependent variable will increase as the independent variable decreases. It can be a subtractive relationship, a division relationship, etc., which is determined by actual application.

[0066] In this embodiment, a specific calculation formula for the matching value is given: the i-th joint point in the image to be evaluated and the i-th joint point in the reference image The matching value of a joint point can be expressed as:

[0067]

[0068] in, Indicates the difference between the i-th joint point in the image to be evaluated and the i-th joint point in the reference image. The matching value of the joint points, Indicates the distance index between the i-th joint point in the image to be evaluated and the i-th joint point in the reference image. The difference in distance indicators of the joint points, Indicates the depth index of the i-th joint point in the image to be evaluated and the depth index of the i-th joint point in the reference image. The difference in depth index of each joint point, Represents the direction vector of the i-th joint point in the image to be evaluated and the direction vector of the i-th joint point in the reference image The cosine value between the direction vectors of the joint points.

[0069] In this embodiment, 0.01 is added to the denominator of the matching value calculation formula to prevent the denominator from being 0. In specific applications, the implementer can set it according to the specific situation. The smaller the difference in the distance index between the i-th joint point and the The smaller the difference in the depth index of the joint points, the greater the difference between the direction vector of the i-th joint point in the image to be evaluated and the direction vector of the i-th joint point in the reference image. The larger the cosine value between the direction vectors of the i-th joint points, the closer the i-th joint point is to the i-th joint point. The higher the similarity of the features presented by the joint points, that is, the closer the i-th joint point in the image to be evaluated is to the i-th joint point in the reference image, the higher the similarity of the features presented by the joint points. The higher the matching value of each joint point, the better.

[0070] Using the above method, the matching value between the candidate node and each joint point in the reference image can be calculated. The larger the matching value, the more the corresponding two joint points should be regarded as a matching joint point pair. Therefore, the joint point with the largest matching value with the candidate joint point in the reference image is determined as the matching joint point of the candidate joint point. The candidate joint point and its matching key point constitute a matching joint point pair.

[0071] By using the above method, all relevant nodes in the image to be evaluated are processed to obtain multiple matching joint point pairs, each of which is composed of a joint point in the image to be evaluated and a joint point in the reference image.

[0072] Next, we will still use the candidate joint point as an example to illustrate that based on the difference in relative occurrence time between the image to be evaluated where the candidate joint point is located and the reference image, and the matching values ​​of all matching joint point pairs in the image to be evaluated where the candidate joint point is located and the reference image, we can obtain the possibility value that the image to be evaluated and the reference image are the same action; specifically, the ratio of the duration between the initial moment and the acquisition moment of the image where the joint point is located to the total duration of completing the current ADL ability training is recorded as the duration ratio of the corresponding image. Based on the difference in duration ratio between the image to be evaluated where the candidate joint point is located and the reference image, and the average value of the matching values ​​of all matching joint point pairs formed by the key points in the image to be evaluated where the candidate joint point is located and the reference image, we can obtain the possibility value that the image to be evaluated and the reference image are the same action. The average value of the matching values ​​is positively correlated with the possibility value, and the difference in duration ratio is negatively correlated with the possibility value. The initial moment is the first moment of the ADL ability training.

[0073] In this embodiment, the calculation formula of the likelihood value is given, and the k-th posture image of the neurological disease patient and the k-th posture image of the neurological disease patient are The probability value of the reference image for the same action can be expressed as:

[0074]

[0075] in, Represents the k-th posture image and the k-th posture image of the patient with neurological disease The probability value of the reference image being the same action, Represents the k-th posture image and the k-th posture image of the patient with neurological disease The average value of the matching values ​​of all matching joint point pairs composed of joint points in the reference image, represents the duration ratio of the k-th posture image of patients with neurological diseases, Indicates the The duration of the reference image, Indicates the absolute value sign.

[0076] The time ratio of the k-th posture image of a patient with a neurological disease to the k-th posture image is The smaller the difference between the two images, the smaller the absolute value, and the more similar the timing of the two images in the overall video. In this embodiment, 0.01 is added to the denominator of the probability value calculation formula to prevent the denominator from being 0. In specific applications, the implementer can set it according to specific circumstances. The average value of the matching values ​​of all the matching joint point pairs composed of the joint points in the kth reference image is used to reflect the relationship between the kth posture image of the neurological disease patient and the kth posture image. The overall matching degree of the joint points in the kth reference image. The smaller the difference in the duration ratio between the kth reference images, the better the The larger the average value of the matching values ​​of all the matching joint point pairs composed of the joint points in the kth reference image, the closer the kth posture image of the neurological disease patient is to the kth posture image. The higher the similarity between the kth and the kth reference images, the more likely these two images are pictures of the same action, i.e., the kth posture image of a neurological patient is similar to the kth posture image of a neurological patient. The higher the probability that the two reference images are of the same action, the higher the probability that the two reference images are of the same action.

[0077] Using the above method, it is possible to obtain the likelihood that the k-th posture image of a neurological patient and each reference image are performing the same action. Since this embodiment collects image data from multiple normal individuals, each normal individual has multiple reference images. Therefore, for any normal individual, the reference image corresponding to the maximum likelihood value of performing the same action as the k-th posture image of the neurological patient among all reference images of the normal individual is used as the comparison image for the k-th posture image of the neurological patient. It should be noted that if multiple reference images among all reference images of the normal individual have the maximum likelihood value of performing the same action as the k-th posture image of the neurological patient, then the first reference image in chronological order is used as the comparison image for the k-th posture image of the neurological patient.

[0078] By using the above method, the probability values ​​of all images to be evaluated of a neurological disease patient and each reference image representing the same action are calculated, and multiple reference images for each image to be evaluated can be screened based on the calculated probability values.

[0079] Step S3, comprehensively considering the possibility that the image to be evaluated and all the control images are the same action and the acquisition time of the image to be evaluated, to obtain the movement completion index of the neurological disease patient at each moment; according to the difference in the change characteristics of each type of indicator and other indicators of the neurological disease patient at each moment and each type of indicator, determine the coordination coefficient at each moment, where the indicators include the movement completion index and the score values ​​of different types of scales.

[0080] The more images of a neurological patient successfully matching a normal person's posture when completing a specified action, and the higher the degree of matching, the higher the patient's motion completion index in this test. Based on this characteristic, the probability that the image being evaluated and all control images represent the same action, as well as the time the image was captured, is then combined to determine the neurological patient's motion completion index at each moment.

[0081] For any image to be evaluated: the average probability value of the image to be evaluated and all its reference images representing the same action is calculated, and this average value is recorded as the first average value; the product of the first average value and the number of frames of the reference image of the image to be evaluated of the neurological patient is calculated; the ratio of this product to the detection time corresponding to the posture image of the neurological patient during the ADL ability training process is determined as the neurological patient's movement completion index at the time the image to be evaluated is captured; the greater the ratio between this product and the detection time corresponding to the posture image of the neurological patient during the ADL ability training process, the more similar the neurological patient is to a normal person when completing the specified action, that is, the greater the movement completion index. Using this method, the movement completion index of the neurological patient at each moment can be obtained.

[0082] The exercise completion index and the score values ​​of different scales are used as indicators of different categories, that is, multiple categories of indicators are obtained.

[0083] Next, a moment in the ADL ability training process of a neurological disease patient is taken as an example for description. The method provided in this embodiment can be used to process other moments.

[0084] Any moment during the ADL ability training of patients with neurological diseases is recorded as the moment to be analyzed.

[0085] For any type of indicator: the difference between the indicator of this type of patient at the next moment after the time to be analyzed and the indicator of this type at the time to be analyzed is used as the change value of this type of indicator at the time to be analyzed; if the change value is a negative number, the time to be analyzed is determined as the time of abnormal fluctuation of this type of indicator.

[0086] Label any one type of indicator as a first type of indicator, and label any other type of indicator except the first type of indicator as a second type of indicator. Record the number of times at and before the time to be analyzed when both the first type of indicator and the second type of indicator experienced abnormal fluctuations as a first number, and record the number of times at and before the time to be analyzed when the first type of indicator experienced abnormal fluctuations as a second number. Determine the ratio of the first number to the second number as the abnormal consistency coefficient between the first type of indicator and the second type of indicator at the time to be analyzed. Using this method, the abnormal consistency coefficient between the first type of indicator and each other type of indicator at the time to be analyzed can be obtained.

[0087] Furthermore, based on the abnormal consistency coefficient between the first category indicator at the time to be analyzed and each other category of indicators, and the difference in change values ​​between the first category indicator at the time to be analyzed and before the time when the first category indicator and each other category of indicators are in abnormal fluctuation at the same time and the corresponding other indicators, the correlation coefficient between the first category indicator at the time to be analyzed and all other indicators is obtained.

[0088] In this embodiment, a specific calculation formula for the correlation coefficient is given. The correlation coefficient between the first type of indicator and all other indicators at the time to be analyzed can be expressed as:

[0089]

[0090] in, It represents the correlation coefficient between the first category of indicators and all other indicators at the time to be analyzed, It represents the average value of the abnormal consistency coefficient between the first category indicator and all other categories of indicators at the time to be analyzed. Indicates the number of indicator types, It indicates the number of times when the first category indicator and the xth category indicator other than the first category indicator are in abnormal fluctuation at the same time before and at the time to be analyzed. It means the first category indicator at the Tth abnormal fluctuation moment when the first category indicator and the xth category indicator other than the first category indicator are at the same time in the abnormal fluctuation moment. It means the xth indicator at the Tth abnormal fluctuation moment when the first indicator and the xth indicator except the first indicator are at the same time in the abnormal fluctuation moment before the moment to be analyzed. Indicates the absolute value sign, Represents the normalization function, which is used to make The value of is in the range of (0, 1).

[0091] It is used to reflect the cumulative sum of the differences in the degree of abnormality when the first category indicator and the other indicators are in a state of abnormal fluctuation. The smaller its value is, the greater the average value of the abnormal consistency coefficients of the first category indicator and all other categories of indicators at the time to be analyzed, which means that the correlation between the first category indicator and all other indicators at the time to be analyzed is stronger, and the larger the correlation coefficient is.

[0092] By using the above method, the correlation coefficient between each type of indicator and all other indicators at the time to be analyzed can be obtained.

[0093] The product of each type of indicator at the time to be analyzed and the correlation coefficient of each type of indicator with all other indicators is recorded as the first eigenvalue of each type of indicator at the time to be analyzed. Each type of indicator at the time to be analyzed has a first eigenvalue; the cumulative sum of the first eigenvalues ​​of all indicators at the time to be analyzed is determined as the coordination coefficient of the time to be analyzed.

[0094] By using the above method, the coordination coefficient at each moment during the ADL ability training process of patients with neurological diseases can be obtained.

[0095] Step S4: determining an effective evaluation value of the training according to the change of the coordination coefficient during the ADL ability training of the neurological disease patient.

[0096] During the recovery process of ADL ability in patients with neurological diseases, there may be instability in the recovery of their ADL due to factors such as disease progression or recurrence of symptoms. For example, after a stroke, the recovery of a patient's cognitive function may fluctuate, leading to fluctuations in the recovery of their ADL ability. Therefore, the effective evaluation value of the training is evaluated based on the changes in the coordination coefficient during ADL ability training for patients with neurological diseases.

[0097] If, as of the current moment, the patient's limb coordination ability recovers better during the ADL ability training, that is, the more times the limb coordination coefficient increases and the greater the increase amplitude, then as of the current moment, the patient's daily living ability recovery is more stable and the recovery degree is better.

[0098] This embodiment, still using the moment to be analyzed as an example, determines the coordination ability growth index for that moment by the ratio of the coordination coefficient at the moment to be analyzed to the coordination coefficient at the moment immediately preceding it. If the coordination ability growth index is greater than 1, the moment to be analyzed is considered a growth moment; if the coordination ability growth index is less than or equal to 1, the moment to be analyzed is considered a decline moment. Using this method, all moments except the first moment during ADL training for patients with neurological diseases are evaluated and divided into two categories: growth moments and decline moments. Consecutive growth moments constitute a growth period, while consecutive decline moments constitute a decline period.

[0099] If the growth period accounts for a larger proportion of the total training time up to the present, and the amplitude of change in limb coordination ability is greater than that in the reduction period, and the interval between the rising periods is shorter, then the recovery of daily living ability in the current state will be more stable.

[0100] Based on this feature, a recovery index is obtained according to the time distribution of the growth period and the relative size relationship between the range of the growth index of the growth period and the range of the growth index of the reduction period. Specifically, a first ratio between the average value of the range of the growth index of all growth periods and the average value of the range of the growth index of all reduction periods, and a second ratio between the total duration of all growth periods and the total duration of the neurological disease patient detection process are calculated; the recovery index is determined based on the maximum time interval between adjacent growth periods, the first ratio, and the second ratio. The first ratio and the second ratio are both positively correlated with the recovery index, and the maximum time interval is negatively correlated with the recovery index.

[0101] In this embodiment, a specific calculation formula for the recovery index is given, and the recovery index can be expressed as:

[0102]

[0103] in, represents the recovery index, represents the total duration of all growth periods during the detection of patients with neurological diseases, represents the total duration of the neurological disease patient testing process, represents the maximum time interval between all adjacent growth periods during the detection of neurological disease patients, It represents the average value of the range of the growth index of all growth periods during the detection of patients with neurological diseases. It represents the average value of the range of the growth index of all the reduction periods during the examination of patients with neurological diseases.

[0104] It represents the first ratio. The larger the ratio is, the greater the change in limb coordination ability is during the growth period compared to the decrease period. represents a second ratio, which represents the ratio of the growth period to the total duration of the neurological disease patient detection process. The larger the first ratio, the larger the second ratio, and the smaller the maximum time interval between all adjacent growth periods in the neurological disease patient detection process, the greater the recovery index.

[0105] After determining the recovery index, the effective evaluation value of the training is evaluated by combining the relative size relationship between the coordination coefficient at the initial moment of ADL ability training and the coordination coefficient at the current moment and the recovery index.

[0106] Specifically, the ratio of the coordination coefficient at the initial moment of ADL ability training to the coordination coefficient at the current moment is calculated, and this ratio is recorded as the third ratio. The larger the third ratio is, the better the coordination of the patient's limbs in the current state is compared with the initial training; the product of the third ratio and the recovery index is determined as the effective evaluation value of the training. When the third ratio is larger and the recovery index is larger, the effective evaluation value of the training is larger.

[0107] The effective evaluation value of the training can provide a reference for medical staff. Subsequent medical staff can judge whether the training method needs to be adjusted based on the effective evaluation value of the training and experience.

[0108] This embodiment first evaluates the likelihood that the image to be evaluated and the reference image represent the same action based on the differences in positional distribution, depth, and relative occurrence time between the joint points in posture images of neurological patients undergoing ADL training and those in posture images of healthy individuals. Specifically, the likelihood that the joint points in the posture images of the neurological patients undergoing ADL training and those in the posture images of healthy individuals represent the same action is determined by comprehensively analyzing temporal features and the features presented by the joint points in the images. The likelihood is then used to screen control images that represent the same action as the posture images of the neurological patients undergoing ADL training. Furthermore, the movement completion status of the neurological patients is evaluated, and a movement completion index is obtained at each moment. The movement completion index and the scores of different scales are used as different indicators. Based on the differences in the changing characteristics of each indicator type and other indicators at each moment, as well as the numerical distribution of each indicator type, an effective evaluation value for the neurological patients' training is determined. This achieves a comprehensive evaluation of the ADL training data of neurological patients, improves the accuracy of the evaluation of the patients' ADL training data, and provides a reference for medical personnel.

[0109] In other embodiments, a device for managing ADL ability training data for patients with neurological diseases is provided, comprising a memory and a processor. The memory is used to store executable program code, and the processor is used to call and execute the executable program code from the memory, causing the device to execute the method performed by the aforementioned ADL ability training data management system for patients with neurological diseases. The device can specifically be a chip, component, or module, and the chip may include a connected processor and memory; wherein the memory is used to store instructions, and when the processor calls and executes the instructions, the chip can execute the method performed by the ADL ability training data management system for patients with neurological diseases provided in the aforementioned embodiments.

[0110] In other embodiments, a computer program product is also provided. When the computer program product is run on a computer, the computer is caused to execute the above-mentioned related steps to implement the ADL ability training data management method based on neurological disease patients provided in the above embodiment.

[0111] In other embodiments, a computer-readable storage medium is also provided, which stores computer program code. When the computer program code runs on a computer, the computer executes the above-mentioned related method steps to implement the method performed by the ADL ability training data management system for neurological disease patients provided in the above embodiment.

[0112] Among them, the provided devices, computer program products, and computer-readable storage media are all used to execute the corresponding methods provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0113] It should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A data management system for ADL ability training of patients with neurological diseases, comprising a memory and a processor, characterized in that: The processor executes the computer program stored in the memory to implement the following steps: Obtain posture images and different types of scales during ADL ability training of normal people and patients with neurological diseases; Based on the differences in the position distribution, depth, and relative occurrence time between the joint points in the image to be evaluated and the reference image, the likelihood value of the image to be evaluated and the reference image representing the same action is evaluated, where the image to be evaluated is a posture image of a patient with a neurological disease and the reference image is a posture image of a normal person; and the likelihood value is used to select a reference image for the image to be evaluated; The probability that the image to be evaluated and all control images represent the same action, as well as the time at which the image to be evaluated was captured, is used to determine the movement completion index for each moment in the neurological disease patient. Based on the differences in the changing characteristics of each indicator type and other indicators at each moment in the neurological disease patient, as well as each indicator type, the coordination coefficient for each moment is determined. The indicators include the movement completion index and the score values ​​of different scales. According to the changes in coordination coefficient during ADL ability training of patients with neurological diseases, the effective evaluation value of the training is determined; Determining the coordination coefficient at each moment based on the difference in the change characteristics of each type of indicator and other indicators of the neurological disease patient at each moment and each type of indicator includes: For any type of indicator: the difference between the indicator of the neurological disease patient at the next moment after the time to be analyzed and the indicator of the any type at the time to be analyzed is used as the change value of the indicator of the any type at the time to be analyzed; if the change value is a negative number, the time to be analyzed is determined to be the time of abnormal fluctuation of the indicator of the any type; Any one type of index is marked as a first type of index, and any other type of index except the first type of index is marked as a second type of index; the number of moments before and before the time to be analyzed when both the first type of index and the second type of index are at abnormal fluctuation moments is recorded as a first number, and the number of moments before and after the time to be analyzed when the first type of index is at abnormal fluctuation moments is recorded as a second number; the ratio of the first number to the second number is determined as the abnormal consistency coefficient between the first type of index and the second type of index at the time to be analyzed; The correlation coefficient between the first category indicator and all other indicators at the time to be analyzed is obtained based on the abnormal consistency coefficient between the first category indicator and each other category of indicators at the time to be analyzed and before, and the difference in change values ​​between the first category indicator and the corresponding other indicators at the time of abnormal fluctuation at the time to be analyzed and before. According to the correlation coefficient of each type of indicator at the time to be analyzed and all other indicators, as well as the coordination coefficient of each type of indicator at the time to be analyzed, the coordination coefficient of the time to be analyzed is obtained; the time to be analyzed is any time during the ADL ability training of the neurological disease patient; Determining the effective evaluation value of the training according to the change of the coordination coefficient during the ADL ability training of the neurological disease patient includes: The ratio of the coordination coefficient at the time to be analyzed to the coordination coefficient at the moment immediately before the time to be analyzed is determined as the coordination ability growth index at the time to be analyzed; if the coordination ability growth index is greater than 1, the time to be analyzed is regarded as an increasing time; if the coordination ability growth index is less than or equal to 1, the time to be analyzed is regarded as a decreasing time; A recovery index is obtained based on the time distribution of the growth period and the relative size relationship between the range of the growth index of the growth period and the range of the growth index of the decrease period; the growth period is composed of continuous growth moments; and the decrease period is composed of continuous decrease moments; Evaluate the effective evaluation value of the training based on the relative magnitude relationship between the coordination coefficient at the initial moment of the ADL ability training and the coordination coefficient at the current moment and the recovery index; The recovery index is obtained according to the time distribution of the growth period and the relative size relationship between the range of the growth index of the growth period and the range of the growth index of the reduction period, including: Calculating a first ratio between an average value of the range of growth indices of all growth periods and an average value of the range of growth indices of all reduction periods, and a second ratio between a total duration of all growth periods and a total duration of the neurological disease patient detection process; determining the recovery index according to a maximum time interval between adjacent growth periods, the first ratio, and the second ratio, wherein both the first ratio and the second ratio are positively correlated with the recovery index, and the maximum time interval is negatively correlated with the recovery index; The effective evaluation value of the training is evaluated by combining the relative magnitude relationship between the coordination coefficient at the initial moment of the ADL ability training and the coordination coefficient at the current moment and the recovery index, including: calculating a third ratio between the coordination coefficient at the initial moment of the ADL ability training and the coordination coefficient at the current moment; The product of the third ratio and the recovery index is determined as an effective evaluation value of the training.

2. The ADL ability training data management system for neurological disease patients according to claim 1, characterized in that: The evaluating the likelihood value of the image to be evaluated and the reference image representing the same action based on the difference in position distribution, depth difference, and corresponding relative occurrence time between the joint points in the image to be evaluated and the joint points in the reference image includes: For any joint point: the ratio of the Euclidean distance between the joint point and the coordinate center point of the human body region where the joint point is located to the Euclidean distance between the two joint points with the greatest distance in the human body region is determined as the distance index of the joint point; the direction vector from the joint point to its adjacent joint point is used as the direction vector of the joint point; the ratio of the depth value of the joint point to the mean depth value of all relevant nodes in the image where the joint point is located is determined as the depth index of the joint point; Obtaining a matching value between the candidate joint point and each joint point in the reference image based on a cosine value between a direction vector of the candidate joint point and a direction vector of each joint point in the reference image, a difference in a distance index between the candidate joint point and each joint point in the reference image, and a difference in a depth index, wherein the cosine value is positively correlated with the matching value, and the difference in the distance index and the difference in the depth index are both negatively correlated with the matching value; determining the joint point in the reference image with the largest matching value with the candidate joint point as the matching joint point of the candidate joint point; the candidate joint point is any joint point in the image to be evaluated; Based on the difference in relative occurrence time between the image to be evaluated and the reference image where the candidate joint point is located, and the matching value of all matching joint point pairs in the image to be evaluated and the reference image where the candidate joint point is located, the possibility value that the image to be evaluated and the reference image are the same action is obtained; wherein the matching joint point pairs are composed of joint points in the image to be evaluated and joint points in the reference image that match the joint points in the image to be evaluated.

3. The ADL ability training data management system for neurological disease patients according to claim 2, characterized in that: The method of obtaining a probability value of the image to be evaluated and the reference image representing the same action based on the difference in relative occurrence time between the image to be evaluated and the reference image where the candidate joint point is located, and the matching value of all matching joint point pairs in the image to be evaluated and the reference image where the candidate joint point is located, includes: The ratio of the time between the initial moment and the acquisition moment of the image where the joint point is located to the total time to complete the current ADL ability training is recorded as the duration ratio of the corresponding image; Based on the difference in the duration ratio between the image to be evaluated and the reference image where the candidate joint point is located, and the average of the matching values ​​of all matching joint point pairs formed by the joint points in the image to be evaluated and the reference image where the candidate joint point is located, the possibility value of the image to be evaluated and the reference image being the same action is obtained. The average of the matching values ​​is positively correlated with the possibility value, and the difference in the duration ratio is negatively correlated with the possibility value.

4. The ADL ability training data management system for neurological disease patients according to claim 1, characterized in that: The method of screening a reference image of the image to be evaluated by using the likelihood value includes: For any normal person: the reference image corresponding to the maximum likelihood value is determined as the control image of the image to be evaluated.

5. The ADL ability training data management system for neurological disease patients according to claim 4, characterized in that: The probability value of the image to be evaluated and all the control images being the same action and the acquisition time of the image to be evaluated are combined to obtain the movement completion index of the neurological disease patient at each moment, including: For any image to be evaluated: Calculating a first average value of the likelihood values ​​of any image to be evaluated and all its reference images representing the same action; Calculate the product of the first average value and the number of frames of the control image of any image to be evaluated of the neurological disease patient; and determine the ratio of the product to the detection time corresponding to the posture image of the ADL ability training process of the neurological disease patient as the movement completion index of the neurological disease patient at the time of acquisition of any image to be evaluated.

6. The ADL ability training data management system for neurological disease patients according to claim 1, characterized in that: The coordination coefficient of the time to be analyzed is obtained based on the correlation coefficient of each type of indicator at the time to be analyzed with all other indicators, and each type of indicator at the time to be analyzed, including: The product of each type of indicator at the time to be analyzed and the correlation coefficient between each type of indicator and all other indicators is recorded as the first eigenvalue of each type of indicator at the time to be analyzed; The cumulative sum of the first characteristic values ​​of all indicators at the time to be analyzed is determined as the coordination coefficient at the time to be analyzed.

Images