A comprehensive analysis system for human health data

Abstract

The present invention relates to the field of health data analysis, and particularly to a comprehensive analysis system for human health data, including: a feature acquisition unit for determining the video frame type according to the motion domain variability and the same-frame reference value, determining the frame extraction method according to the frame extraction condition, and acquiring the data features of the monitoring data collected by the changing limb points in the extracted video frames; a method determination unit for determining the monitored motion state according to the same feature difference reference value and the feature dimension reference value, and determining the feature processing method according to the monitored motion state; a first processing unit for determining the first feature selection method according to the set conditions; a second processing unit for determining the feature category according to the individual influence value and the behavior-related value, and determining the second feature selection method according to the determination conditions; a motion recognition unit; the present invention can screen for effective features to improve the behavior recognition accuracy, thereby reducing the motion risk.

Description

Technical Field

[0001] The present invention relates to the field of health data analysis, and in particular to a comprehensive analysis system for human health data. Background Art

[0002] Wearable devices can only collect human health data, but cannot accurately identify the wearer's exercise type, and thus cannot warn of various risk factors that may exist in the wearer's health during exercise. Therefore, behavioral information is often identified through a human behavior recognition model. However, due to the complexity of human movement, the feature dimension collected based on the motion data is large, which increases the computational complexity of the human behavior recognition model, which not only leads to a long recognition time, but also may affect the accuracy of behavior recognition. Therefore, how to select effective features to conduct targeted training on the human behavior recognition model to improve the accuracy of behavior recognition and thereby reduce exercise risks is a technical problem that needs to be urgently solved by technical personnel in this field.

[0003] Chinese patent publication number CN114863566A discloses a method for identifying human motion behavior, including the following steps: Step 1, respectively use a motion capture system to collect data on multiple human actions, and obtain raw data including three-axis accelerometer data, three-axis gyroscope data, three-axis magnetometer data, and three-axis attitude angle data; Step 2, construct a CRF conditional random field model applied to text sequence data; Step 3, convert the time series of raw data of human motion containing all joints into feature vectors, and mark the corresponding human motion behavior types, and use the feature vectors and human motion behavior types as corpus to train the CRF model; Step 4, given the time series feature vectors of human motion data to be classified, human motion behavior recognition is performed to obtain the corresponding human motion behavior type. It can be seen that the above technical solution has the following problems: the data collection points are not selected in a targeted manner, resulting in a high redundancy of the collected raw data, and the feature vectors are not screened, resulting in a high feature dimension, which in turn leads to a long recognition time, and may also affect the accuracy of behavior recognition, and cannot accurately provide health reminders. Summary of the invention

[0004] To this end, the present invention provides a comprehensive analysis system for human health data to overcome the problems in the prior art that there is no targeted selection of data collection points, resulting in high redundancy of the collected original data, and no screening of feature vectors, resulting in high feature dimensions, which in turn leads to too long recognition time and may also affect the accuracy of behavior recognition and fail to provide accurate health reminders.

[0005] To achieve the above object, the present invention provides a human health data comprehensive analysis system, comprising:

[0006] A feature acquisition unit is used to determine the video frame type according to the motion domain variation degree and the same-frame reference value, and determine the frame extraction method as uniformly extracting video frames or extracting video frames according to similar type-two video frame combinations and each type-one video frame according to the frame extraction condition, and obtain the data features of the monitoring data collected by the changing limb points in the extracted video frames;

[0007] A method determination unit is connected to the feature acquisition unit and is used to determine the monitoring motion state according to the same feature difference reference value and the feature dimension reference value, and determine the feature processing method as obtaining the mapping values corresponding to the respective data features and determining the first feature selection method according to the set conditions or determining the second feature selection method according to the determination conditions;

[0008] A first processing unit is connected to the method determination unit and is used to determine the first feature selection method as selecting data features according to the difference threshold or the ratio correlation degree according to the set conditions;

[0009] A second processing unit is connected to the method determination unit and is used to determine the data feature category according to the individual influence value and the behavior correlation value, and determine the second feature selection method as selecting data features according to the redundant feature combination or the association threshold according to the determination conditions;

[0010] A motion recognition unit is respectively connected to the first processing unit and the second processing unit and is used to perform model training according to the selected features, and give a health reminder according to the user's basic information when the user is exercising.

[0011] Further, the feature acquisition unit determines the video frame type according to the motion domain variation degree and the same-frame reference value, and the video frame types include:

[0012] Type-one video frames in which the motion domain variation degree is greater than or equal to the preset motion domain variation degree and the same-frame reference value is less than the preset same-frame reference value;

[0013] Type-two video frames in which the motion domain variation degree is less than the preset motion domain variation degree or the same-frame reference value is greater than or equal to the preset same-frame reference value.

[0014] Further, the feature acquisition unit responds to different frame extraction conditions to determine the frame extraction method;

[0015] The frame extraction condition to which the feature acquisition unit responds is that the proportion of type-one video frames is greater than or equal to the preset proportion of type-one video frames and the distribution coefficient of type-one video frames is greater than or equal to the preset distribution coefficient of type-one video frames, and the frame extraction method is to uniformly extract video frames, where,

[0016] Each video frame is equally divided into N parts according to the time sequence, and the video frames at each equal division point are recorded as the extracted frames;

[0017] The frame extraction condition for which the feature acquisition unit responds is that the proportion of a certain type of video frame is less than a preset proportion of a certain type of video frame or the distribution coefficient of a certain type of video frame is less than a preset distribution coefficient of a certain type of video frame. The frame extraction method is to extract video frames according to similar combinations of a certain type of video frame and each certain type of video frame. Among them,

[0018] Determine similar combinations of a certain type of video frame according to the similarity coefficient of limb points, randomly extract a corresponding single certain type of video frame from each similar combination of a certain type of video frame, and record it as the extracted certain type of video frame. And sort each certain type of video frame according to the time sequence to obtain a reference sequence. Divide the reference sequence into M equal parts, and record the certain type of video frame and the extracted certain type of video frame at each equal division point as the extracted frames.

[0019] Furthermore, the state analysis unit determines the monitored motion state according to the same feature difference reference value and the feature dimension reference value. The monitored motion states include:

[0020] The first monitored motion state where the same feature difference reference value is greater than or equal to a preset same feature difference reference value or the feature dimension reference value is greater than or equal to a preset feature dimension reference value;

[0021] The second monitored motion state where the same feature difference reference value is less than a preset same feature difference reference value and the feature dimension reference value is less than a preset feature dimension reference value.

[0022] Furthermore, the method determination unit responds to different monitored motion states to determine the feature processing method;

[0023] When the method determination unit responds to the first monitored motion state, the feature processing method is to obtain the mapping value corresponding to each data feature and determine the first feature selection method according to the set conditions;

[0024] When the method determination unit responds to the second monitored motion state, the feature processing method is to determine the second feature selection method according to the determination conditions.

[0025] Furthermore, the mapping value corresponding to each data feature is determined according to the feature ratio. The feature ratio is determined according to the ratio of the first difference to the second difference;

[0026] The mapping value and the feature ratio have a positive correlation.

[0027] Furthermore, the set condition for which the first processing unit responds is that the distribution difference degree of the mapping value is greater than or equal to a preset distribution difference degree of the mapping value. The first feature selection method is to select data features according to the difference threshold. Among them,

[0028] Determine a similar set according to the difference threshold, and randomly extract several data features from each similar set; the difference threshold is the absolute value of the difference between the mapping values corresponding to two features.

[0029] Furthermore, the setting condition for which the first processing unit responds is that the mapping value distribution difference degree is less than a preset mapping value distribution difference degree, and the first feature selection method is to select data features according to the ratio correlation degree, where,

[0030] Determine a similarity set according to the ratio correlation degree, and randomly select several data features from each similarity set; the ratio correlation degree is determined according to the absolute value of the difference between the ratio correlation degrees corresponding to two data features, and the difference ratio is determined according to the ratio of the between-class difference degree to the within-class difference degree.

[0031] Furthermore, the second processing unit determines the data feature category according to the individual influence value and the behavior-related value, and the data feature category includes:

[0032] One type of data feature where the individual influence value is greater than or equal to a preset individual influence value or the behavior-related value is less than a preset behavior-related value;

[0033] Two types of data features where the individual influence value is less than a preset individual influence value and the behavior-related value is greater than or equal to a preset behavior-related value.

[0034] Furthermore, the second processing unit determines the second feature selection method according to the determination condition;

[0035] The determination condition for which the second processing unit responds is that the proportion of one type of data feature is greater than or equal to a preset proportion of one type of data feature and the feature richness is greater than or equal to a preset feature richness, and the second feature selection method is to select data features according to the redundant feature combination, where,

[0036] Determine the redundant feature combination according to the feature correlation degree, and randomly select the first preset number of data features from each redundant feature combination;

[0037] The determination condition for which the second processing unit responds is that the proportion of one type of data feature is less than a preset proportion of one type of data feature or the feature richness is less than a preset feature richness, and the second feature selection method is to select data features according to the association threshold, where,

[0038] Select the second preset number of data features in the order from large to small according to the association threshold.

[0039] Compared with the prior art, the beneficial effect of the present invention lies in that, in the technical scheme of the present invention, the video frame type is determined according to the action domain variability and the same-frame reference value, and the complexity of the human body motion in the video frame and the similarity between the human body motion and the human body motion in other video frames are effectively reflected through the action domain variability and the same-frame reference value, and then the frame extraction condition is effectively determined according to the video frame type, and the number and density of a type of video frames are effectively reflected through the frame extraction condition, and then different frame extraction methods are adaptively selected, so that the selection of the frame extraction method is more in line with the actual application scenario, and then the changed limb points in the extracted frame extracted according to the frame extraction method are used as data monitoring points, avoiding the defect of the prior art that the data collection points are not selected in a targeted manner, resulting in a high redundancy of the collected original data, thereby effectively reducing the cost of data collection and processing, and at the same time ensuring that the monitoring data of the data monitoring point is highly representative of key actions or posture changes, which is conducive to improving the representativeness of several data features obtained by feature extraction.

[0040] Furthermore, in the present invention, the monitored motion state is determined according to the same feature difference reference value and the feature dimension reference value, and the magnitude difference of the data features and the number of data features are effectively reflected through the same feature difference reference value and the feature dimension reference value, and then different feature processing methods are adaptively selected according to the monitored motion state, so that the selection of feature processing method is more in line with the actual application scenario, avoiding the defect of not screening the feature vector in the prior art, resulting in a high data feature dimension, and also avoiding the increase in computational complexity and the risk of overfitting due to excessively high data feature dimensions, thereby improving the recognition accuracy and generalization ability of the model, and then effectively warning of various risk factors that may exist in the wearer's health during exercise.

[0041] Furthermore, the present invention effectively reflects the distribution of mapping values by setting conditions, and then adaptively selects different first feature selection methods according to the set conditions, so that the selection of the first feature selection method is more in line with the actual application scenario, avoiding the disadvantage of not being able to effectively select data features when the difference in the distribution of mapping values is small, which is conducive to reducing redundant information and retaining key data features so that the model can more accurately capture the potential rules in the data, thereby improving the accuracy of behavior recognition.

[0042] Furthermore, in the present invention, the data feature category is determined based on the individual influence value and the behavior-related value. The individual influence value and the behavior-related value effectively reflect the influence degree of individual differences on the data feature and the correlation degree between the data feature and the same data feature in other behaviors, making the determination of the data feature category more in line with the actual application scenario. The determination condition effectively reflects the quantity situation of a type of data feature and the richness of non-zero data features. Then, different second feature selection methods are adaptively selected according to the determination condition, which is beneficial to removing redundant data features, reducing the overfitting of the model to specific data, and also reducing the complexity and computational amount of the model, thereby improving the training speed and behavior recognition accuracy of the model. Furthermore, health reminders are provided based on human behaviors and basic information. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a unit connection diagram of the human health data comprehensive analysis system of the present invention;

[0044] Figure 2 It is a flowchart of determining the frame extraction method according to the frame extraction condition of the present invention;

[0045] Figure 3 It is a flowchart of determining the feature processing method according to the monitored motion state of the present invention;

[0046] Figure 4 It is a flowchart of determining the first feature selection method according to the set condition of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] In order to make the objectives and advantages of the present invention clearer, the present invention will be further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0048] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present invention and do not limit the protection scope of the present invention.

[0049] It should be noted that in the description of the present invention, the terms indicating directions or positional relationships such as "upper", "lower", "left", "right", "inner", "outer", etc. are based on the directions or positional relationships shown in the drawings. This is only for convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention.

[0050] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installation", "connection", and "coupling" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0051] Please refer to Figures 1 to 4 As shown, the present invention provides a comprehensive human health data analysis system, including:

[0052] A feature acquisition unit, configured to determine the video frame type according to the motion domain variability and the same-frame reference value, and determine the frame extraction method as uniformly extracting video frames or extracting video frames according to the combination of similar type-II video frames and each type-I video frame according to the frame extraction condition, and acquire the data features of the monitoring data collected by the changing limb points in the extracted video frames;

[0053] A method determination unit, connected to the feature acquisition unit, configured to determine the monitoring motion state according to the same feature difference reference value and the feature dimension reference value, and determine the feature processing method as obtaining the mapping values corresponding to the respective data features and determining the first feature selection method according to the setting condition or determining the second feature selection method according to the determination condition;

[0054] A first processing unit, connected to the method determination unit, configured to determine the first feature selection method as selecting data features according to the difference threshold or the ratio correlation degree according to the setting condition;

[0055] A second processing unit, connected to the method determination unit, configured to determine the data feature category according to the individual influence value and the behavior correlation value, and determine the second feature selection method as selecting data features according to the redundant feature combination or the association threshold according to the determination condition;

[0056] A motion recognition unit, respectively connected to the first processing unit and the second processing unit, configured to perform model training according to the selected features, and give a health reminder according to the user's basic information when the user is exercising.

[0057] The present invention further includes a data acquisition unit for collecting health data information, which is respectively connected to the feature acquisition unit and the motion recognition unit. The health data information includes monitoring information and basic information. The monitoring information includes a number of monitoring videos, and a single monitoring video is a monitoring video of a single user performing a motion behavior. The motion behaviors include running, walking, squatting and standing up, etc. The changing limb points are the limb points whose positions have changed compared with the limb point positions in the standard state. The standard state is the state where the user stands at attention in the video frame. Randomly select a point on each joint of the human body, and these points are the limb points, including the head, shoulders, elbows, wrists, hips, knees and ankles, etc. The recognition of the limb points can be determined through machine vision and deep learning networks. The basic information includes a number of health parameters, and the health parameters include but are not limited to age, blood pressure, heart rate and blood oxygen saturation, which are easily understood by those skilled in the art and will not be elaborated here. The monitoring data includes a number of data units. A single data unit is when a single user performs a motion behavior, and an inertial sensor motion capture device is fixed at the corresponding data monitoring point to obtain the monitoring parameters corresponding to each time point monitored in real time during the duration of the motion behavior. The monitoring parameters include motion acceleration, angular velocity and direction, etc. Among them, the time points are set by the user himself, and a method for setting time points is provided. Set a time point every 1s, that is, record the monitoring parameters every 1s.

[0058] When obtaining the data features of the monitoring data collected by the changing limb points in the extracted video frames, the acquisition methods that can be used include Fourier transform and wavelet transform, etc. The obtained data features include but are not limited to mean, variance, skewness, kurtosis, autocorrelation coefficient, interquartile range and root mean square. The user can obtain a number of data features according to his own needs, which are easily understood by those skilled in the art and will not be elaborated here.

[0059] In the present invention, a health behavior library is provided. The health behavior library stores the warning parameter ranges of each health parameter corresponding to each motion behavior, so as to give a health reminder when the motion recognition unit recognizes a certain motion behavior of the human body and the health parameters in the user's basic information are not within the corresponding warning parameter ranges;

[0060] In the present invention, a number of historical records are correspondingly set. Any historical record records at least the action domain variation, the same frame reference value, the proportion of a certain type of video frame, the distribution coefficient of a certain type of video frame and the reference value of the same feature difference, etc. in the historical process of at least one data feature selection. And each historical record corresponds to a qualified mark, and the qualified mark records whether the recognition speed and recognition accuracy of the human behavior recognition model meet the user's requirements. The qualified mark can be recorded manually.

[0061] Specifically, the feature acquisition unit determines the video frame type according to the action domain variability and the same-frame reference value. The video frame types include:

[0062] A type of video frame in which the action domain variability is greater than or equal to the preset action domain variability and the same-frame reference value is less than the preset same-frame reference value;

[0063] A type of video frame in which the action domain variability is less than the preset action domain variability or the same-frame reference value is greater than or equal to the preset same-frame reference value.

[0064] Among them, the action domain variability = the number of changing limb points + the change amplitude reference value. The number of changing limb points is the total amount of changing limb points. It can be understood that the video frame is a rectangular image. Taking the bottom left corner point of the rectangle as the coordinate origin, the straight line extending to the right along the bottom edge of the rectangle starting from the coordinate origin as the x-axis, and the straight line extending upward along the left side of the rectangle starting from the coordinate origin as the y-axis, a rectangular coordinate system is established. The limb point position is the coordinate of the limb point in the rectangular coordinate system. For a single limb point, this limb point is denoted as the target limb point, and the shortest distance between the position of the target limb point in the video frame and the corresponding limb point position in the standard state is denoted as the change distance. The change amplitude reference value is the sum of the change distances of each limb point;

[0065] The confirmation method of the same-frame reference value is as follows: For a surveillance video, a single video frame in the surveillance video is denoted as the target video frame, and other video frames excluding the target video frame are denoted as reference video frames. The reference video frames with the same changing limb points as the target video frame and the change amplitude difference value less than the preset change amplitude difference value are denoted as the same frames. The same-frame reference value = (the number of same frames + 1) / the number of all video frames in the surveillance video. The change amplitude difference value is the absolute value of the difference between the change amplitude reference values corresponding to two video frames. The value of the preset change amplitude difference value can be determined by the user according to the actual application scenario. The greater the user's requirement for the similarity of actions in the reference video frame and the target video frame, the smaller the value of the preset change amplitude difference value. A value of the preset change amplitude difference value is provided, and the preset change amplitude difference value is 5% of the change amplitude reference value corresponding to the target video frame;

[0066] The values of the preset action domain variability and the preset same-frame reference value can be determined by the user according to the actual application scenario. The higher the user's requirement for the complexity of actions in a type of video frame, the greater the value of the preset action domain variability and the smaller the value of the preset same-frame reference value. A value of the preset action domain variability and the preset same-frame reference value is provided. The average value of the action domain variability corresponding to the historical records that can meet the user's requirements is denoted as the preset action domain variability, and the average value of the same-frame reference value corresponding to the historical records that can meet the user's requirements is denoted as the preset same-frame reference value.

[0067] Specifically, the feature acquisition unit determines the frame extraction method in response to different frame extraction conditions;

[0068] The frame extraction condition to which the feature acquisition unit responds is that the proportion of a certain type of video frame is greater than or equal to the preset proportion of a certain type of video frame and the distribution coefficient of a certain type of video frame is greater than or equal to the preset distribution coefficient of a certain type of video frame. The frame extraction method is to evenly extract video frames. Among them,

[0069] Each video frame is divided into N equal parts according to the time sequence, and the video frames at each equal division point are recorded as the extracted frames;

[0070] The frame extraction condition to which the feature acquisition unit responds is that the proportion of a certain type of video frame is less than the preset proportion of a certain type of video frame or the distribution coefficient of a certain type of video frame is less than the preset distribution coefficient of a certain type of video frame. The frame extraction method is to extract video frames according to the similar combination of a certain type of video frame and each certain type of video frame. Among them,

[0071] The similar combination of a certain type of video frame is determined according to the limb point similarity coefficient. A single certain type of video frame corresponding to each similar combination of a certain type of video frame is randomly extracted and recorded as the extracted certain type of video frame. Each certain type of video frame is sorted according to the time sequence to obtain a reference sequence. The reference sequence is divided into M equal parts, and the certain type of video frame and the extracted certain type of video frame at each equal division point are recorded as the extracted frames.

[0072] Among them, the frame extraction conditions include the first frame extraction condition and the second frame extraction condition. The first frame extraction condition is that the proportion of a certain type of video frame is greater than or equal to the preset proportion of a certain type of video frame and the distribution coefficient of a certain type of video frame is greater than or equal to the preset distribution coefficient of a certain type of video frame. The second frame extraction condition is that the proportion of a certain type of video frame is less than the preset proportion of a certain type of video frame or the distribution coefficient of a certain type of video frame is less than the preset distribution coefficient of a certain type of video frame;

[0073] The confirmation method of the proportion of a certain type of video frame is that for a surveillance video, the proportion of a certain type of video frame = the number of a certain type of video frames in the surveillance video / the number of video frames in the surveillance video;

[0074] The distribution coefficient of a certain type of video frame = adjacent difference threshold + distance paragraph value. The adjacent difference threshold is the average value of the average adjacent distances corresponding to each certain type of video frame in a single surveillance video; for a single certain type of video frame, this certain type of video frame is recorded as the target certain type of video frame, the average value of the distances between the adjacent certain type of video frames and the target certain type of video frame is recorded as the adjacent distance average value, and the average value of the adjacent distance average values corresponding to each abnormal frequency is recorded as the adjacent difference threshold; the distance between two certain type of video frames is the absolute value of the difference in the surveillance time corresponding to the two certain type of video frames; in the time direction, the distance paragraph value = the surveillance time corresponding to the last certain type of video frame - the surveillance time corresponding to the first certain type of video frame;

[0075] The values of a preset first - type video frame ratio and a preset first - type video frame distribution coefficient can be determined by the user according to the actual application scenario. The greater the user's demand for the monitoring representativeness of the data monitoring point, the smaller the values of the first - type video frame ratio and the preset first - type video frame distribution coefficient. Provide a set of values for the first - type video frame ratio and the preset first - type video frame distribution coefficient. Denote the average value of the first - type video frame ratios corresponding to the historical records that can meet the user's needs as the preset first - type video frame ratio, and denote the average value of the first - type video frame distribution coefficients corresponding to the historical records that can meet the user's needs as the preset first - type video frame distribution coefficient;

[0076] The values of N and M can be determined by the user according to actual needs. The greater the user's demand for the monitoring representativeness of the data monitoring point, the larger the values of N and M. Provide a set of values for N and M. N is 40% of the total number of video frames in a single monitored video, and M is 40% of the total number of video frames in the reference sequence.

[0077] For two second - type video frames, denote the number of changing limb points corresponding to the first second - type video frame as A, and denote the number of changing limb points corresponding to the other second - type video frame as B. The limb - point similarity coefficient = the number of identical changing limb points in the two video frames / the larger value of A and B;

[0078] Determine the similar second - type video frame combinations according to the limb - point similarity coefficient, including: According to the order of the monitoring time corresponding to the second - type video frames from early to late, conduct combined analysis on each second - type video frame in a monitored video. When conducting combined analysis on a single second - type video frame, denote this second - type video frame as the target second - type video frame, denote the other second - type video frames excluding the target second - type video frame as the reference second - type video frames, denote the reference second - type video frames with a limb - point similarity coefficient greater than the preset limb - point similarity coefficient as a set of similar second - type video frames, and conduct combined analysis on the second - type video frames not denoted as similar second - type video frame combinations until all second - type video frames are denoted as similar second - type video frame combinations, then stop the combined analysis.

[0079] Specifically, the state analysis unit determines the monitored motion state according to the same - feature difference reference value and the feature - dimension reference value. The monitored motion states include:

[0080] A first monitored motion state where the same - feature difference reference value is greater than or equal to the preset same - feature difference reference value or the feature - dimension reference value is greater than or equal to the preset feature - dimension reference value;

[0081] A second monitored motion state where the same - feature difference reference value is less than the preset same - feature difference reference value and the feature - dimension reference value is less than the preset feature - dimension reference value.

[0082] Among them, each changed limb point extracted from each video frame is recorded as a data detection point. For a motion behavior, a set of data characteristics of the monitoring data of each data detection point corresponding to a single monitoring video corresponding to a motion behavior is recorded as a feature set;

[0083] The same feature difference reference value is the maximum value among the between-class difference degrees corresponding to each data characteristic in a motion behavior. The confirmation method of the between-class difference degree is that each feature set corresponding to the motion behavior is recorded as a reference set. For a single data characteristic, the data characteristic is recorded as the target characteristic. The between-class difference degree = the maximum value of the target characteristic in each reference set - the minimum value of the target characteristic in each reference set; the feature dimension reference value is the number of data characteristics corresponding to a motion behavior;

[0084] For the values of the preset same feature difference reference value and the preset feature dimension reference value, the user can determine them according to the actual application scenario. It can be understood that the larger the values of the preset same feature difference reference value and the preset feature dimension reference value are, the greater the user's demand for selecting the feature processing method as the second feature selection method according to the determination condition is. Provide a set of values for the preset same feature difference reference value and the preset feature dimension reference value, detect the historical records of the feature processing method of obtaining the mapping values corresponding to each data characteristic and determining the first feature selection method according to the setting conditions, and record the average value of the same feature difference reference values corresponding to the historical records that can meet the user's needs as the preset same feature difference reference value, and record the average value of the feature dimension reference values corresponding to the historical records that can meet the user's needs as the preset feature dimension reference value.

[0085] Specifically, the method determination unit responds to different monitoring motion states to determine the feature processing method;

[0086] When the method determination unit responds to the first monitoring motion state, the feature processing method is to obtain the mapping values corresponding to each data characteristic and determine the first feature selection method according to the setting conditions;

[0087] When the method determination unit responds to the second monitoring motion state, the feature processing method is to determine the second feature selection method according to the determination condition.

[0088] Among them, the setting conditions include a first setting condition and a second setting condition. The first setting condition is that the mapping value distribution difference degree is greater than or equal to the preset mapping value distribution difference degree, and the second setting condition is that the mapping value distribution difference degree is less than the preset mapping value distribution difference degree;

[0089] The determination conditions include a first determination condition and a second determination condition. The first determination condition is that the proportion of a certain type of data feature is greater than or equal to a preset proportion of the certain type of data feature and the feature richness is greater than or equal to a preset feature richness. The second determination condition is that the proportion of the certain type of data feature is less than the preset proportion of the certain type of data feature or the feature richness is less than the preset feature richness.

[0090] Specifically, the mapping value corresponding to each data feature is determined according to the feature ratio, and the feature ratio is determined according to the ratio of the first difference to the second difference.

[0091] The mapping value and the feature ratio are in a positive correlation relationship.

[0092] Among them, for a single data feature, the value corresponding to the feature is denoted as the feature value, the mapping value = 2×feature ratio - 1, the feature ratio = first difference / second difference, the first difference = feature value - minimum feature value, the second difference = maximum feature value - minimum feature value, and the maximum feature value and the minimum feature value are respectively the maximum value and the minimum value among the values corresponding to each data feature in a single feature set.

[0093] Specifically, the setting condition for which the first processing unit responds is that the mapping value distribution difference degree is greater than or equal to a preset mapping value distribution difference degree, and the first feature selection method is to select data features according to a difference threshold. Among them,

[0094] Similar sets are determined according to the difference threshold, and several data features in each similar set are randomly selected; the difference threshold is the absolute value of the difference between the mapping values corresponding to two features.

[0095] Among them, the mapping value distribution difference degree is the average value of the minimum differences corresponding to each mapping value. The confirmation method of the minimum difference is as follows: for a feature set, a mapping value is denoted as the target mapping value, and the other mapping values excluding the target mapping value are denoted as reference mapping values. The absolute value of the difference between each reference mapping value and the target mapping value is detected and denoted as the reference value, and the minimum value among the reference values is denoted as the minimum difference.

[0096] For the value of the preset mapping value distribution difference degree, the user can determine it according to the actual application scenario. It can be understood that the larger the value of the preset mapping value distribution difference degree, the greater the user's need to determine similar sets according to the difference threshold. Provide a value of the preset mapping value distribution difference degree, detect the historical records corresponding to determining similar sets according to the difference threshold, and denote the average value of the mapping value distribution difference degrees corresponding to the historical records that can meet the user's needs as the preset mapping value distribution difference degree.

[0097] Determine similar sets according to the difference threshold, including: perform similarity analysis on each data feature in a feature set in ascending order of mapping values. When performing similarity analysis on a single data feature, denote this data feature as the first target feature, denote other features except the first target feature as the first reference features, and denote the first reference features and the first target feature whose difference threshold from the first target feature is less than the preset difference threshold as a similar set, and perform similarity analysis on the data features not denoted as similar sets until all data features are denoted as similar sets, then stop the similarity analysis.

[0098] The value of the preset difference threshold can be determined by the user according to the actual application scenario. The greater the user's requirement for the degree of relevance of the data features in the similar set, the smaller the value of the preset difference threshold. Provide a value of the preset difference threshold, and denote the average value of the difference thresholds corresponding to the historical records that can meet the user's requirements as the preset difference threshold.

[0099] Specifically, the setting condition for which the first processing unit responds is that the mapping value distribution difference degree is less than the preset mapping value distribution difference degree, and the first feature selection method is to select data features according to the ratio relevance degree, where

[0100] Determine similar sets according to the ratio relevance degree, and randomly select several data features from each similar set; the ratio relevance degree is determined according to the absolute value of the difference between the ratio relevance degrees corresponding to two data features, and the difference ratio is determined according to the ratio of the between-class difference degree to the within-class difference degree.

[0101] Among them, the difference ratio = between-class difference degree / within-class difference degree. The method for confirming the within-class difference degree is to denote a data feature in the feature set as the target reference feature, and denote the absolute value of the difference between other data features in the feature set that do not include the target reference feature and the target reference feature as the reference absolute value. The within-class difference degree = the maximum value among the reference absolute values - the minimum value among the reference absolute values;

[0102] Determine similar sets according to the ratio relevance degree, including: perform set analysis on each data feature in a feature set in ascending order of the ratio relevance degree. When performing set analysis on a single data feature, denote this data feature as the second target feature, denote other data features except the second target feature as the second reference features, and denote the second reference features and the second target feature whose ratio relevance degree from the second target feature is less than the preset ratio relevance degree as a similar set, and perform set analysis on the data features not denoted as similar sets until all data features are denoted as similar sets, then stop the set analysis.

[0103] The value of the preset ratio correlation can be determined by the user according to the actual application scenario. The greater the user's demand for the degree of correlation of data features in the similar set, the smaller the value of the preset ratio correlation. A value of the preset ratio correlation is provided, and the average value of the ratio correlations corresponding to the historical records that can meet the user's needs is recorded as the preset ratio correlation.

[0104] Specifically, the second processing unit determines the data feature category according to the individual influence value and the behavior-related value. The data feature category includes:

[0105] A type of data feature where the individual influence value is greater than or equal to the preset individual influence value or the behavior-related value is less than the preset behavior-related value;

[0106] A type of data feature where the individual influence value is less than the preset individual influence value and the behavior-related value is greater than or equal to the preset behavior-related value.

[0107] Among them, for each feature set corresponding to a single motion behavior, the age of the user corresponding to each feature set is detected. For a single data feature, the feature values corresponding to this data feature are recorded as target feature values. The average value of the target feature values corresponding to the maximum age is recorded as K1, and the average value of the target feature values corresponding to the minimum age is recorded as K2. The individual influence value = |K1 - K2| / (the maximum age - the minimum age),

[0108] The confirmation method of the behavior-related value is as follows: For a motion behavior, this motion behavior is recorded as the target motion behavior, and the other motion behaviors excluding the target motion behavior are recorded as reference motion behaviors. For a single data feature in the target motion behavior, this data feature is recorded as the reference feature. The feature mean values corresponding to each motion behavior are detected. The feature mean value is the average value of the feature values corresponding to the reference feature in a single motion behavior. The reference motion behaviors with a difference in feature mean values less than the preset feature mean value difference from the feature mean value of the target motion behavior are recorded as similar motion behaviors. The behavior-related value = the number of similar motion behaviors / the number of all motion behaviors. The feature mean value difference is the absolute value of the difference between the feature mean values corresponding to the two motion behaviors;

[0109] The values of the preset individual influence value and the preset behavior-related value can be determined by the user according to the actual application scenario. It can be understood that the smaller the value of the preset individual influence value and the larger the value of the preset behavior-related value, the greater the user's demand for determining the data feature as a type of data feature. A value of the preset individual influence value and a value of the preset behavior-related value are provided. The historical records corresponding to each type of data feature are detected, and the average value of the individual influence values corresponding to the historical records that can meet the user's needs is recorded as the preset individual influence value, and the average value of the behavior-related values corresponding to the historical records that can meet the user's needs is recorded as the preset behavior-related value.

[0110] Specifically, the second processing unit determines a second feature selection method according to a determination condition;

[0111] The determination condition to which the second processing unit responds is that the proportion of a type of data feature is greater than or equal to a preset proportion of a type of data feature and the feature richness is greater than or equal to a preset feature richness. The second feature selection method is to select data features according to a redundant feature combination, where

[0112] determine a redundant feature combination according to the feature correlation, and randomly select a first preset number of data features from each redundant feature combination;

[0113] The determination condition to which the second processing unit responds is that the proportion of a type of data feature is less than a preset proportion of a type of data feature or the feature richness is less than a preset feature richness. The second feature selection method is to select data features according to a correlation threshold, where

[0114] select a second preset number of data features in descending order of the correlation threshold.

[0115] Among them, for a feature set, the proportion of a type of data feature = the number of a type of data features in the feature set / the total number of data features in the feature set, and the feature richness = the number of non-zero data features in the feature set / the total number of data features in the feature set;

[0116] For the values of the preset proportion of a type of data feature and the preset feature richness, the user can determine them according to the actual application scenario. It can be understood that the smaller the values of the preset proportion of a type of data feature and the preset feature richness, the greater the user's need to select the second feature selection method to select data features according to a redundant feature combination. Provide a set of values for the preset proportion of a type of data feature and the preset feature richness, detect the historical records where the second feature selection method is to select data features according to a correlation threshold, and record the average value of the proportion of a type of data feature corresponding to the historical records that can meet the user's needs as the preset proportion of a type of data feature, and record the average value of the feature richness corresponding to the historical records that can meet the user's needs as the preset feature richness;

[0117] For any two data features, the calculation formula for the feature correlation σ corresponding to the two data features is:

[0118]

[0119] m is the number of feature sets corresponding to a single motion behavior; x k and y k are respectively the values of two data features in the kth feature set in the feature set corresponding to a single motion behavior, is the average value of the values of the data feature corresponding to x k corresponding to, is the average value of the values of the data feature corresponding to yk The average value of the corresponding data feature values, where k = 1, 2, 3, ……, m;

[0120] Determine redundant feature combinations according to feature relevance, including: performing redundancy analysis on each data feature in a feature set. When performing redundancy analysis on a single data feature, denote this data feature as the third target feature, denote the other data features except the third target feature as the third reference features, and denote the third reference features whose feature relevance to the third target feature is less than the preset feature relevance and the third target feature as a redundant feature combination, and continue to perform redundancy analysis on the data features not denoted as redundant feature combinations until all data features are denoted as redundant feature combinations, then stop the redundancy analysis.

[0121] Association threshold = individual influence value - behavior correlation value. The values of the first preset quantity and the second preset quantity can be determined by the user according to their own needs. Provide a set of values for the first preset quantity and the second preset quantity. The first preset quantity is 20% of the number of data features in a single redundant feature combination, and the preset quantity is 20% of the number of data features in a single feature set.

[0122] So far, the technical solution of the present invention has been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the protection scope of the present invention.

[0123] The above are only the preferred embodiments of the present invention and are not used to limit the present invention; for those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A comprehensive analysis system for human health data, characterized in that, Including: A feature acquisition unit, configured to determine the video frame type according to the motion domain variability and the same-frame reference value, and determine the frame extraction method as uniformly extracting video frames or extracting video frames according to similar type-II video frame combinations and each type-I video frame according to the frame extraction condition, and acquire the data features of the monitoring data collected by the changing limb points in the extracted video frames; A method determination unit, connected to the feature acquisition unit, configured to determine the monitoring motion state according to the same feature difference reference value and the feature dimension reference value, and determine the feature processing method as obtaining the mapping values corresponding to the respective data features and determining the first feature selection method according to the setting condition or determining the second feature selection method according to the determination condition; A first processing unit, connected to the method determination unit, configured to determine the first feature selection method as selecting data features according to the difference threshold or the ratio correlation according to the setting condition; A second processing unit, connected to the method determination unit, configured to determine the data feature category according to the individual influence value and the behavior correlation value, and determine the second feature selection method as selecting data features according to the redundant feature combination or the correlation threshold according to the determination condition; A motion recognition unit, respectively connected to the first processing unit and the second processing unit, configured to perform model training according to the selected features, and perform health reminders according to the user's basic information when the user is exercising; The feature acquisition unit determines the video frame type according to the motion domain variability and the same-frame reference value, and the video frame type includes: Type-I video frames with a motion domain variability greater than or equal to a preset motion domain variability and a same-frame reference value less than the preset same-frame reference value; Type-II video frames with a motion domain variability less than the preset motion domain variability or a same-frame reference value greater than or equal to the preset same-frame reference value; The feature acquisition unit responds to different frame extraction conditions to determine the frame extraction method; The frame extraction condition responded by the feature acquisition unit is that the proportion of type-I video frames is greater than or equal to the preset proportion of type-I video frames and the distribution coefficient of type-I video frames is greater than or equal to the preset distribution coefficient of type-I video frames, and the frame extraction method is to uniformly extract video frames, where Each video frame is equally divided into N parts according to the time sequence, and the video frames at each equal division point are recorded as the extracted frames; The frame extraction condition responded by the feature acquisition unit is that the proportion of type-I video frames is less than the preset proportion of type-I video frames or the distribution coefficient of type-I video frames is less than the preset distribution coefficient of type-I video frames, and the frame extraction method is to extract video frames according to similar type-II video frame combinations and each type-I video frame, where The similar type-II video frame combinations are determined according to the limb point similarity coefficient, a single type-II video frame corresponding to each similar type-II video frame combination is randomly extracted and recorded as the extracted type-II video frame, and each type-I video frame is sorted according to the time sequence to obtain a reference sequence, and the reference sequence is equally divided into M parts, and the type-I video frames and the extracted type-II video frames at each equal division point are recorded as the extracted frames.

2. The comprehensive analysis system for human health data according to claim 1, wherein The state analysis unit determines the monitoring motion state according to the same feature difference reference value and the feature dimension reference value, and the monitoring motion state includes: The first monitored motion state where the reference value of the same feature difference is greater than or equal to the preset reference value of the same feature difference or the reference value of the feature dimension is greater than or equal to the preset reference value of the feature dimension; The second monitored motion state where the reference value of the same feature difference is less than the preset reference value of the same feature difference and the reference value of the feature dimension is less than the preset reference value of the feature dimension.

3. The comprehensive human health data analysis system according to claim 2, wherein The method determination unit responds to different monitored motion states to determine the feature processing method; When the method determination unit responds to the first monitored motion state, the feature processing method is to obtain the mapping values corresponding to each data feature and determine the first feature selection method according to the set conditions; When the method determination unit responds to the second monitored motion state, the feature processing method is to determine the second feature selection method according to the determination conditions.

4. The comprehensive human health data analysis system according to claim 3, wherein The mapping values corresponding to each data feature are determined according to the feature ratio, and the feature ratio is determined according to the ratio of the first difference to the second difference; The mapping value and the feature ratio have a positive correlation.

5. The comprehensive human health data analysis system according to claim 4, characterized in that, The set condition for which the first processing unit responds is that the distribution difference degree of the mapping values is greater than or equal to the preset distribution difference degree of the mapping values, and the first feature selection method is to select data features according to the difference threshold. Among them, Determine the similarity set according to the difference threshold, and randomly select several data features from each similarity set; the difference threshold is the absolute value of the difference between the mapping values corresponding to the two features.

6. The comprehensive human health data analysis system according to claim 5, characterized in that The set condition for which the first processing unit responds is that the distribution difference degree of the mapping values is less than the preset distribution difference degree of the mapping values, and the first feature selection method is to select data features according to the ratio correlation degree. Among them, Determine the similarity set according to the ratio correlation degree, and randomly select several data features from each similarity set; the ratio correlation degree is determined according to the absolute value of the difference between the ratio correlation degrees corresponding to the two data features, and the difference ratio is determined according to the ratio of the between-class difference degree to the within-class difference degree.

7. The comprehensive human health data analysis system according to claim 3, wherein The second processing unit determines the data feature category according to the individual influence value and the behavior-related value. The data feature category includes: One type of data feature where the individual influence value is greater than or equal to the preset individual influence value or the behavior-related value is less than the preset behavior-related value; Two types of data features where the individual influence value is less than the preset individual influence value and the behavior-related value is greater than or equal to the preset behavior-related value.

8. The comprehensive human health data analysis system according to claim 7, characterized in that The second processing unit determines the second feature selection method according to the determination conditions; The determination condition for which the second processing unit responds is that the proportion of the first type of data features is greater than or equal to the preset proportion of the first type of data features and the feature richness is greater than or equal to the preset feature richness. The second feature selection method is to select data features according to the redundant feature combination. Among them, Determine the redundant feature combination according to the feature correlation degree, and randomly select the first preset number of data features from each redundant feature combination; The determination condition for which the second processing unit responds is that the proportion of the first type of data features is less than the preset proportion of the first type of data features or the feature richness is less than the preset feature richness. The second feature selection method is to select data features according to the association threshold. Among them, Select the second preset number of data features in descending order according to the association threshold.

Images