Intelligent assistance method and system for nursing rehabilitation

Abstract

The present invention provides an intelligent auxiliary method and system for nursing rehabilitation, and relates to the technical field of nursing rehabilitation data processing. The method comprises: detecting multiple physiological indicators and posture vectors of rehabilitation trainees; determining the first target moment when the rehabilitation trainees feel pain stimulation; determining the second target moment when the rehabilitation trainees perform the rehabilitation action of the current nursing cycle; determining the target posture vector in the rehabilitation action of the current nursing cycle; obtaining the initial posture vector of the rehabilitation trainees, and inputting the target posture vector, the initial posture vector, and the posture vector of the previous moment of the target posture vector into the trained nursing rehabilitation action generation model to obtain the action posture vector of the rehabilitation action of the next nursing cycle; generating the action video of the rehabilitation action of the next nursing cycle. According to the present invention, appropriate rehabilitation actions can be determined at different stages of rehabilitation training to improve the appropriateness and safety of rehabilitation training.

Description

Technical Field

[0001] The present invention relates to the technical field of nursing rehabilitation data processing, and in particular to an intelligent auxiliary method and system for nursing rehabilitation. Background Art

[0002] In the related art, CN117556220A discloses an intelligent auxiliary system and method for rehabilitation nursing, which relates to the field of rehabilitation nursing. It first arranges the physiological data of multiple predetermined time points according to the time dimension and the physiological sample dimension into a physiological data time series matrix, and then encodes the motion data of the multiple predetermined time points respectively to obtain a sequence of motion data encoding vectors, and then performs context global motion state association feature analysis on the sequence of motion data encoding vectors to obtain a sequence of context motion state feature vectors, and then performs embedded association feature analysis on the sequence of context motion state feature vectors and the physiological data time series matrix to obtain physiological state embedded motion state time series features, and finally, based on the physiological state embedded motion state time series features, it is determined whether the rehabilitation plan of the monitored patient needs to be adjusted. In this way, medical staff can better monitor and manage the patient's rehabilitation process.

[0003] CN114912290A belongs to the application technical field of intelligent assistance of nursing robots in stand-sit interaction, and particularly relates to the solution of the problem of obstacles to stand-sit interaction for people with physical disabilities. This method proposes a nursing robot intelligent assistance control method based on model prediction, which can solve the problem of obstacles to stand-sit interaction for people with physical disabilities and reduce the impact force of the seat on the human body during stand-sit interaction. This method adopts the following technical scheme, which includes the following steps: Step 1: Simplify the biological model of the human body, combine the simplified human body structure, construct a spatial geometric model describing the biological structure of the human body, and describe the geometric relationship between the parts of the human body during stand-sit interaction; Step 2: Construct a dynamic model of the human body for stand-sit interaction based on the vertical dimension and the human body spatial geometric model in step 1.

[0004] Based on the above related technologies, the efficiency of rehabilitation care can be improved. However, in terms of rehabilitation movements, it is difficult to accurately determine the appropriate movement postures in the related technologies, making it difficult to find rehabilitation movements that are suitable for the rehabilitation conditions of rehabilitation trainees in different rehabilitation training stages.

[0005] The information disclosed in the background technology section of this application is only intended to deepen the understanding of the general background technology of this application, and should not be regarded as an admission or any form of suggestion that the information constitutes the prior art already known to those skilled in the art. Summary of the invention

[0006] The present invention provides an intelligent auxiliary method and system for nursing rehabilitation, which can solve the technical problem that related technologies are difficult to find rehabilitation movements suitable for the rehabilitation conditions of rehabilitation trainees in different rehabilitation training stages.

[0007] According to a first aspect of the present invention, there is provided an intelligent assistance method for nursing rehabilitation, comprising:

[0008] At multiple moments in the current nursing cycle, multiple physiological indicators of the rehabilitation training personnel are detected by physiological indicator detection equipment;

[0009] At multiple moments in the current care cycle, the posture vector of the rehabilitation trainee is detected by a posture detection device, wherein the posture detection device includes a head position detection device, a neck position detection device, a shoulder position detection device, an elbow position detection device, a wrist position detection device, a waist position detection device, a knee position detection device and an ankle position detection device, and the posture vector is used to represent the relative position relationship between the posture detection devices;

[0010] At the end of the current nursing cycle, based on multiple physiological indicators obtained at multiple moments, determine the first target moment at which the rehabilitation training personnel feel pain stimulation among multiple moments;

[0011] According to the posture vectors obtained at the multiple moments, determining, among the multiple moments, a second target moment for the rehabilitation personnel to perform the rehabilitation action of the current nursing cycle;

[0012] Determine a target posture vector in the rehabilitation action of the current nursing cycle according to the first target moment, the second target moment and the posture vector corresponding to the second target moment, wherein the target posture vector is used to represent the posture of the rehabilitation trainee feeling pain stimulation when performing the rehabilitation action of the current nursing cycle;

[0013] Acquire an initial posture vector of the rehabilitation trainee, and input the target posture vector, the initial posture vector and the posture vector at the previous moment of the target posture vector into the trained nursing rehabilitation action generation model to obtain the action posture vector of the rehabilitation action of the next nursing cycle, wherein the initial posture vector is used to represent the relative position relationship between the various posture detection devices when the rehabilitation trainee is in a non-action state;

[0014] According to the motion posture vector, an action video of the rehabilitation action for the next nursing cycle is generated.

[0015] According to a second aspect of the present invention, there is provided an intelligent assistance system for nursing rehabilitation, comprising:

[0016] The physiological index module is used to detect multiple physiological indexes of rehabilitation training personnel through physiological index detection equipment at multiple moments in the current nursing cycle;

[0017] A posture vector module, used for detecting the posture vector of the rehabilitation trainee at multiple moments in the current nursing cycle through a posture detection device, wherein the posture detection device includes a head position detection device, a neck position detection device, a shoulder position detection device, an elbow position detection device, a wrist position detection device, a waist position detection device, a knee position detection device and an ankle position detection device, and the posture vector is used to represent the relative position relationship between the various posture detection devices;

[0018] The first target moment module is used to determine the first target moment at which the rehabilitation training personnel feel pain stimulation from among multiple moments at the end of the current nursing cycle according to multiple physiological indicators obtained at multiple moments;

[0019] A second target time module, used to determine, from among the multiple moments, a second target time for the rehabilitation personnel to perform the rehabilitation action of the current nursing cycle;

[0020] A target posture vector module, used to determine a target posture vector in the rehabilitation action of the current nursing cycle according to the first target moment, the second target moment and the posture vector corresponding to the second target moment, wherein the target posture vector is used to represent the posture of the rehabilitation training personnel feeling pain stimulation when performing the rehabilitation action of the current nursing cycle;

[0021] The action posture vector module is used to obtain the initial posture vector of the rehabilitation trainee, and input the target posture vector, the initial posture vector and the posture vector at the previous moment of the target posture vector into the trained nursing rehabilitation action generation model to obtain the action posture vector of the rehabilitation action of the next nursing cycle, wherein the initial posture vector is used to represent the relative position relationship between the various posture detection devices when the rehabilitation trainee is in a non-action state;

[0022] The action video module is used to generate an action video of the rehabilitation action for the next nursing cycle according to the action posture vector.

[0023] According to a third aspect of the present invention, there is provided an intelligent assistance device for nursing and rehabilitation, comprising: a processor; a memory for storing instructions executable by the processor; wherein the processor is configured to call the instructions stored in the memory to execute the intelligent assistance method for nursing and rehabilitation.

[0024] According to a fourth aspect of the present invention, there is provided a computer-readable storage medium having computer program instructions stored thereon, wherein the computer program instructions, when executed by a processor, implement the intelligent assistance method for nursing rehabilitation.

[0025] Technical effect: According to the present invention, by monitoring the physiological indicators and posture vectors of rehabilitation trainees at multiple moments, the rehabilitation progress and needs of rehabilitation trainees can be more accurately understood, thereby realizing a more personalized nursing plan. Through the physiological indicator detection device and the posture detection device, the physiological state and posture of the rehabilitation trainees can be monitored in real time, and problems can be found and adjusted in time. The rehabilitation action is adjusted according to the moment and posture when the rehabilitation trainees feel the pain stimulation, so as to find the appropriate rehabilitation action in different rehabilitation training stages, and improve the appropriateness and safety of rehabilitation training. Using the trained nursing rehabilitation action generation model, the rehabilitation action of the next nursing cycle is generated according to the target posture vector and the initial posture vector, which helps the rehabilitation trainees to perform correct action exercises. The action video of the rehabilitation action generated in the next nursing cycle can help the rehabilitation trainees understand the correct action execution method more intuitively, improve the training effect and learning efficiency. When determining the first target moment, the discrimination index can be obtained by the overall deviation degree of multiple physiological indicators at adjacent moments in the current nursing cycle from the reference standard index, so as to determine the first target moment when the rehabilitation trainees may feel the pain stimulation, which helps to accurately determine the posture of the rehabilitation trainees who feel the pain when performing the rehabilitation action, and is conducive to more accurate adjustment of the rehabilitation action. When determining the target posture vector, the relative motion vector of the posture detection device relative to the reference posture detection device in the time period from the moment before the third target moment to the third target moment, and the relative motion vector of the posture detection device relative to the reference posture detection device in the time period from the third target moment to the moment after the third target moment can be determined, and the motion direction between the two relative motion vectors can be determined to obtain the target posture discrimination parameter, so as to objectively and accurately judge whether the rehabilitation training personnel are stimulated by pain at the third target moment, and then determine the target posture vector, providing a data basis for the subsequent effective management and control of the rehabilitation actions of the nursing cycle. When training the nursing rehabilitation action generation model, the loss function can be reasonably set through the sample action posture vector, the sample initial posture vector, the sample target posture vector and the sample posture vector at the moment before the sample target posture vector, so that the action corresponding to the sample action posture vector is executed in the same way as the action corresponding to the sample target posture vector, but the amplitude is slightly smaller, and the action corresponding to the sample action posture vector is more in line with the action characteristics of the human body, so that the action corresponding to the vector is easy to execute, which helps to improve the effect of rehabilitation training.

[0026] It should be understood that the above general description and the following detailed description are exemplary and explanatory only and do not limit the present invention. Other features and aspects of the present invention will become more apparent from the following detailed description of exemplary embodiments with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other embodiments can be obtained based on these drawings without creative work.

[0028] Figure 1 A schematic diagram of a process flow of an intelligent assistance method for nursing rehabilitation according to an embodiment of the present invention is exemplarily shown;

[0029] Figure 2 A block diagram of an intelligent assistance system for nursing rehabilitation according to an embodiment of the present invention is exemplarily shown. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0032] Figure 1 A flow chart of an intelligent assistance method for nursing rehabilitation according to an embodiment of the present invention is exemplarily shown, and the method includes:

[0033] Step S101, at multiple moments in the current nursing cycle, multiple physiological indicators of the rehabilitation training personnel are detected by physiological indicator detection equipment;

[0034] Step S102, at multiple moments in the current nursing cycle, the posture vector of the rehabilitation trainee is detected by a posture detection device, wherein the posture detection device includes a head position detection device, a neck position detection device, a shoulder position detection device, an elbow position detection device, a wrist position detection device, a waist position detection device, a knee position detection device and an ankle position detection device, and the posture vector is used to represent the relative position relationship between the posture detection devices;

[0035] Step S103, at the end of the current nursing cycle, according to the multiple physiological indicators obtained at multiple moments, determine the first target moment at which the rehabilitation trainee feels pain stimulation among the multiple moments;

[0036] Step S104, determining a second target time for the rehabilitation personnel to perform the rehabilitation action of the current nursing cycle from among the multiple time points according to the posture vectors obtained at the multiple time points;

[0037] Step S105, determining a target posture vector in the rehabilitation action of the current nursing cycle according to the first target moment, the second target moment, and the posture vector corresponding to the second target moment, wherein the target posture vector is used to represent the posture of the rehabilitation trainee feeling pain stimulation when performing the rehabilitation action of the current nursing cycle;

[0038] Step S106, obtaining an initial posture vector of the rehabilitation trainee, and inputting the target posture vector, the initial posture vector and the posture vector at the previous moment of the target posture vector into the trained nursing rehabilitation action generation model to obtain the action posture vector of the rehabilitation action of the next nursing cycle, wherein the initial posture vector is used to represent the relative position relationship between the various posture detection devices when the rehabilitation trainee is in a non-action state;

[0039] Step S107, generating an action video of the rehabilitation action for the next nursing cycle according to the action posture vector.

[0040] According to the intelligent auxiliary method for nursing rehabilitation of the embodiment of the present invention, by monitoring the physiological indicators and posture vectors of the rehabilitation trainees at multiple times, the rehabilitation progress and needs of the rehabilitation trainees can be understood more accurately, thereby realizing a more personalized nursing plan. Through the physiological indicator detection device and the posture detection device, the physiological state and posture of the rehabilitation trainees can be monitored in real time, and problems can be found and adjusted in time. The rehabilitation movements are adjusted according to the moment and posture when the rehabilitation trainees feel the pain stimulation, so as to find the appropriate rehabilitation movements in different rehabilitation training stages, and improve the appropriateness and safety of rehabilitation training. Using the trained nursing rehabilitation movement generation model, the rehabilitation movements of the next nursing cycle are generated according to the target posture vector and the initial posture vector, helping the rehabilitation trainees to practice the correct movements. The action video of the rehabilitation movement generated for the next nursing cycle can help the rehabilitation trainees understand the correct action execution method more intuitively, and improve the training effect and learning efficiency.

[0041] According to one embodiment of the present invention, in step S101, each detection cycle can be set to 10 minutes, 15 minutes, etc., and the interval between adjacent moments can be set to 1 second, 3 seconds, etc., and the present invention is not limited to this. Use equipment specially designed to measure physiological indicators, such as a sphygmomanometer, a heart rate monitor, a thermometer, etc. The physiological indicator detection equipment can monitor changes in various physiological indicators of the human body. Physiological indicators may include multiple parameters such as blood pressure, heart rate, and body temperature. The multiple physiological indicators obtained from each detection are gradually analyzed and recorded to understand the changes in the physiological indicators of the rehabilitation trainees over the training time.

[0042] According to one embodiment of the present invention, in step S102, during the rehabilitation training, the rehabilitation trainee carries various posture detection devices, i.e., position sensors, which help monitor the posture changes of the rehabilitation trainee. The posture vector is the relative position vector of each posture detection device and the reference posture detection device. For example, the reference posture detection device can be set as a neck position detection device, then the posture vector is a vector composed of elements of the relative position relationship between other position detection devices and the neck position detection device (the coordinates of the three dimensions of the positions of other position detection devices, and the difference between the coordinates of the three dimensions of the positions of the neck position detection device), for example, a vector composed of elements of the relative position relationship between the elbow position detection device and the neck position detection device, elements of the relative position relationship between the wrist position detection device and the neck position detection device, etc. Among them, the reference posture detection device is a head position detection device or a neck position detection device.

[0043] According to an embodiment of the present invention, in step S103, the pain stimulation felt by the rehabilitation training personnel may be manifested in aspects such as increased blood pressure, increased heart rate, and increased body temperature, and the moment when these conditions are met may be determined as the first target moment.

[0044] According to one embodiment of the present invention, step S103 includes: obtaining reference standard indicators corresponding to multiple physiological indicators; determining the first target moment at which the rehabilitation training personnel feel pain stimulation based on relative deviation indicators between the multiple physiological indicators at each moment and the reference standard indicators.

[0045] According to one embodiment of the present invention, the reference standard index may be a plurality of standard indexes such as high pressure 120 mmHg, low pressure 80 mmHg, heart rate 80 beats per minute, body temperature 36.5 degrees Celsius, etc. The relative deviation index may evaluate the degree of deviation between the physiological index of the rehabilitation trainee and the reference standard index at each moment. The first target moment may correspond to a physiological index that significantly deviates from the reference standard index, indicating that the rehabilitation trainee may feel pain stimulation.

[0046] According to an embodiment of the present invention, the first target moment at which the rehabilitation training personnel feel pain stimulation is determined based on the relative deviation indexes of multiple physiological indicators at each moment and the reference standard indicators, including: obtaining the discrimination index at the i-th moment according to formula (1): ,

[0047] (1),

[0048] in, is the jth physiological index at the i-th moment in the current nursing cycle, is the jth physiological index at the i+1th moment in the current nursing cycle, is the reference standard index of the jth physiological index, m is the number of types of physiological indexes, is the relative deviation index of the jth physiological index at the i-th moment in the current nursing cycle, j≤m, and j and m are both positive integers; if the discrimination index at the i-th moment is greater than or equal to the preset discrimination threshold, the i-th moment is determined as the first target moment.

[0049] According to one embodiment of the present invention, in formula (1), It is the product of the relative deviation index between various physiological indicators at the i-th moment in the current nursing cycle and their corresponding reference standard indicators, indicating the overall deviation degree of various physiological indicators of rehabilitation training personnel from the reference standard indicators at the i-th moment in the current nursing cycle. It is the product of the relative deviation index between each physiological index at the i+1th moment in the current nursing cycle and its corresponding reference standard index, indicating the overall deviation degree of each physiological index of the rehabilitation training personnel from the reference standard index at the i+1th moment in the current nursing cycle. The difference between the above two items is the discrimination index at the i-th moment. If the discrimination index at the i-th moment is greater than or equal to the preset discrimination threshold, it means that the overall degree of deviation of the physiological index of the rehabilitation training personnel from the reference standard index at the i+1th moment in the current nursing cycle is significantly improved relative to the i-th moment. It may be due to the occurrence of pain stimulation at the i-th moment, coupled with the delayed effect of nerve conduction, resulting in a significant increase in the overall deviation degree at the i+1th moment. For example, pain stimulation is felt at the i-th moment, and then a significant increase in heart rate occurs at the i+1th moment. Therefore, the i-th moment may be subjected to pain stimulation, and the i-th moment is determined as the first target moment, that is, the moment when pain stimulation may be received.

[0050] In this way, the discrimination index can be obtained through the overall deviation degree of multiple physiological indicators at adjacent moments in the current care cycle from the reference standard indicators, so as to determine the first target moment when the rehabilitation training personnel may feel the pain stimulation, which helps to accurately determine the posture of the rehabilitation training personnel who feel pain when performing rehabilitation movements, and is conducive to more accurate adjustment of rehabilitation movements.

[0051] According to one embodiment of the present invention, in step S104, the posture vector describes the relative positional relationship between different body parts and the head or neck, and the execution of the rehabilitation action of the current nursing cycle by the rehabilitation personnel can be analyzed. In the current nursing cycle, the moment when the rehabilitation personnel is performing the rehabilitation action can be determined, that is, the second target moment.

[0052] According to one embodiment of the present invention, step S104 includes: obtaining the motion posture vector of the rehabilitation action of the current nursing cycle; determining, among the posture vectors obtained at multiple moments, a first posture vector whose cosine similarity with the motion posture vector of the rehabilitation action of the current nursing cycle is greater than or equal to a preset similarity threshold; and determining the moment corresponding to the first posture vector as the second target moment.

[0053] According to one embodiment of the present invention, the action posture vector of the rehabilitation action of the current nursing cycle is a vector for describing the posture of the rehabilitation action of the current nursing cycle. When the rehabilitation trainee performs the rehabilitation action, the posture vector of the rehabilitation trainee is similar to the action posture vector. In other words, if it is detected that the posture vector of the rehabilitation trainee is similar to the action posture vector, it can be said that the rehabilitation trainee is performing the rehabilitation action. The cosine similarity between the posture vector and the action posture vector of the rehabilitation action of the current nursing cycle represents the degree of similarity between the two vectors. The closer the cosine similarity is to 1, the more similar the posture vector is to the action posture vector of the rehabilitation action of the current nursing cycle. When the cosine similarity is greater than or equal to a preset similarity threshold, the posture vector is determined as a first posture vector, and the moment corresponding to the first posture vector is determined as the second target moment, that is, the moment of performing the rehabilitation action.

[0054] According to one embodiment of the present invention, step S105 includes: determining a third target moment in the intersection of a set of first target moments and a set of second target moments; obtaining a posture vector corresponding to the third target moment, a posture vector corresponding to a moment before the third target moment, and a posture vector corresponding to a moment after the third target moment; determining a target posture discrimination parameter of the third target moment based on the posture vector corresponding to the third target moment, the posture vector corresponding to a moment before the third target moment, and the posture vector corresponding to a moment after the third target moment; if the target posture discrimination parameter of the third target moment is greater than or equal to a preset posture discrimination threshold, determining the posture vector corresponding to the third target moment as the target posture vector in the rehabilitation action of the current nursing cycle.

[0055] According to one embodiment of the present invention, the third target moment is a moment in the intersection of the set of the first target moment and the set of the second target moment, that is, the moment when the rehabilitation trainee is performing the rehabilitation training action and may feel pain stimulation. It can be determined whether the rehabilitation trainee is performing the opposite direction movement before and after the third target moment by the posture vector corresponding to the third target moment, the posture vector corresponding to the moment before the third target moment, and the posture vector corresponding to the moment after the third target moment. If the rehabilitation trainee performs the opposite direction movement before and after a third target moment, combined with the factor that the overall deviation degree of the physiological indicators of the rehabilitation trainee at the third target moment increases, it can be known that at the third target moment, the rehabilitation trainee feels the pain stimulation and moves the limbs in the opposite direction to reduce the pain. Therefore, the third target moment is the moment when the rehabilitation trainee feels the pain stimulation in the process of performing the rehabilitation action, and the posture vector corresponding to the third target moment is the target posture vector.

[0056] According to an embodiment of the present invention, determining the target posture discrimination parameter at the third target moment according to the posture vector corresponding to the third target moment, the posture vector corresponding to the moment before the third target moment, and the posture vector corresponding to the moment after the third target moment includes: determining the target posture discrimination parameter at the kth third target moment according to formula (2): ,

[0057] (2),

[0058] in, is a vector composed of elements in the posture vector corresponding to the kth third target moment, representing the relative position relationship between the sth posture detection device and the reference posture detection device, wherein the reference posture detection device is a head position detection device or a neck position detection device, S is the total number of posture detection devices, is a vector composed of elements representing the relative positional relationship between the s-th attitude detection device and the reference attitude detection device in the attitude vector corresponding to the moment before the k-th third target moment, is a vector composed of elements in the posture vector corresponding to the moment after the kth third target moment, representing the relative position relationship between the sth posture detection device and the reference posture detection device, s≤S-1, and s and S are both positive integers, and if is a conditional function.

[0059] According to one embodiment of the present invention, in formula (2), represents the relative motion vector of the s-th posture detection device relative to the reference posture detection device during the period from the moment before the third target moment to the third target moment, It represents the relative motion vector of the s-th posture detection device relative to the reference posture detection device in the time period from the third target moment to the moment after the third target moment. Indicates that when the product of the above two relative motion vectors is less than or equal to 0, the conditional function value is 1, otherwise, the conditional function value is 0. That is, if the angle between the motion directions of the above two relative motion vectors is greater than or equal to 90°, that is, the motion directions of the two relative motion vectors are opposite or nearly opposite before and after the third target moment, then the conditional function value is recorded as 1, otherwise, the conditional function value is recorded as 0. The average value of the conditional function values ​​corresponding to each posture detection device can be used as the target posture discrimination parameter. Furthermore, taking the rehabilitation personnel's arm performing rehabilitation movements as an example, one arm wears three posture detection devices (elbow position detection device, wrist position detection device, shoulder position detection device), then the preset posture discrimination threshold can be set to or At a third target moment, if the angle between the motion directions of the two relative motion vectors corresponding to two or three posture detection devices is greater than or equal to 90°, it can be determined that the rehabilitation personnel feel pain stimulation and make an arm retraction action. The posture vector corresponding to the third target moment is determined as the target posture vector in the rehabilitation action of the current nursing cycle.

[0060] In this way, the relative motion vector of the posture detection device relative to the reference posture detection device in the time period from the moment before the third target moment to the third target moment, as well as the relative motion vector of the posture detection device relative to the reference posture detection device in the time period from the third target moment to the moment after the third target moment can be determined, and the motion direction between the two relative motion vectors can be determined to obtain the target posture discrimination parameters, thereby objectively and accurately judging whether the rehabilitation training personnel are subjected to pain stimulation at the third target moment, and then determining the target posture vector, providing a data basis for the subsequent effective management and control of the rehabilitation movements of the nursing cycle.

[0061] According to one embodiment of the present invention, in step S106, the initial posture vector of the rehabilitation trainee in a non-motion state is collected and recorded. The initial posture vector describes the relative position relationship between the various posture detection devices of the rehabilitation trainee in a non-motion state, and the body proportions of the rehabilitation trainee can be understood. The motion trajectory of the rehabilitation action can be understood through the posture vector of the previous moment of the target posture vector, so as to better optimize the rehabilitation action. The trained nursing rehabilitation action generation model can be a neural network model, which can generate the action posture vector of the rehabilitation action of the next nursing cycle through the initial posture vector, the target posture vector, and the posture vector of the previous moment of the target posture vector. The action posture vector is a posture vector corresponding to an action with a slightly smaller action amplitude than that corresponding to the target posture vector.

[0062] According to one embodiment of the present invention, the training steps of the nursing rehabilitation action generation model include: inputting the sample initial posture vector, the sample target posture vector and the sample posture vector at the previous moment of the sample target posture vector into the nursing rehabilitation action generation model to obtain the sample action posture vector; obtaining the loss function of the nursing rehabilitation action generation model according to the sample action posture vector, the sample initial posture vector, the sample target posture vector and the sample posture vector at the previous moment of the sample target posture vector; training the nursing rehabilitation action generation model according to the loss function to obtain a trained nursing rehabilitation action generation model.

[0063] According to one embodiment of the present invention, the nursing rehabilitation action generation model is trained through a loss function, and the model parameters can be continuously adjusted to minimize the loss function, and finally a trained nursing rehabilitation action generation model is obtained, which can accurately generate the action posture vector of the rehabilitation action of the next nursing cycle.

[0064] According to one embodiment of the present invention, the loss function of the nursing rehabilitation action generation model is obtained according to the sample action posture vector, the sample initial posture vector, the sample target posture vector and the sample posture vector at the previous moment of the sample target posture vector, including: obtaining the loss function of the nursing rehabilitation action generation model according to formula (3): ,

[0065] (3),

[0066] in, is a vector composed of elements in the sample target posture vector representing the relative position relationship between the sth posture detection device and the reference posture detection device, wherein the reference posture detection device is a head position detection device or a neck position detection device, S is the total number of posture detection devices, is a vector composed of elements representing the relative position relationship between the sth posture detection device and the reference posture detection device in the sample posture vector at the previous moment of the sample target posture vector, is a vector composed of elements in the sample action posture vector representing the relative position relationship between the sth posture detection device and the reference posture detection device, is the vertical angle between the vector composed of the elements in the sample target posture vector representing the relative position relationship between the s-th posture detection device and the reference posture detection device and the vector composed of the elements in the sample initial posture vector representing the relative position relationship between the s-th posture detection device and the reference posture detection device, is the vertical angle between the vector composed of the elements representing the relative position relationship between the s-th posture detection device and the reference posture detection device in the sample posture vector at the previous moment of the sample target posture vector and the vector composed of the elements representing the relative position relationship between the s-th posture detection device and the reference posture detection device in the sample initial posture vector, is the horizontal angle between the vector composed of the elements in the sample target posture vector representing the relative position relationship between the s-th posture detection device and the reference posture detection device and the vector composed of the elements in the sample initial posture vector representing the relative position relationship between the s-th posture detection device and the reference posture detection device, is the horizontal angle between the vector composed of the elements representing the relative position relationship between the s-th posture detection device and the reference posture detection device in the sample posture vector at the previous moment of the sample target posture vector and the vector composed of the elements representing the relative position relationship between the s-th posture detection device and the reference posture detection device in the sample initial posture vector, and is the preset weight, and , s≤S-1, and s and S are both positive integers, if is a conditional function.

[0067] According to one embodiment of the present invention, in formula (3), It indicates that the position of the s-th posture detection device relative to the reference posture detection device has not changed compared with the previous moment. is the Euclidean distance between the relative position relationship between the sth posture detection device and the reference posture detection device in the sample action posture vector and the relative position relationship between the sth posture detection device and the reference posture detection device in the sample target posture vector, which represents the error output by the nursing rehabilitation action generation model. That is, when the relative position relationship between the sth posture detection device and the reference posture detection device does not change, it means that the part wearing the posture detection device does not move during the rehabilitation action. Therefore, the vector composed of the elements of the relative position relationship between the posture detection device and the reference posture detection device generated by the nursing rehabilitation action generation model is theoretically consistent with If there is a deviation, As the loss function value in this case, and shrink during the training process, so that and Closer.

[0068] According to one embodiment of the present invention, In the case of is the length of the vector composed of elements in the sample target posture vector representing the relative position relationship between the sth posture detection device and the reference posture detection device. is the length of the vector composed of elements representing the relative positional relationship between the sth posture detection device and the reference posture detection device in the sample posture vector at the previous moment of the sample target posture vector. express The x-axis coordinate of the target value. express The y-axis coordinate of the target value. express The target value of the z-axis coordinate. Therefore, That is to say The adjustment goal is to make Close to the adjustment target. The adjustment target is the vector between the sample target posture vector and the sample posture vector at the previous moment of the sample target posture vector, and through the setting of the above-mentioned length and angle, the position corresponding to the vector is located on the arc between the position corresponding to the sample target posture vector and the position corresponding to the sample posture vector at the previous moment of the sample target posture vector, which not only makes the movement amplitude corresponding to the vector smaller than the movement amplitude corresponding to the sample target posture vector, but also makes the execution mode of the movement corresponding to the vector the same as the execution mode of the movement corresponding to the sample target posture vector, but the amplitude is slightly smaller, and the characteristics located on the above-mentioned arc are also more in line with the movement characteristics of the human body, so that the movement corresponding to the vector is easy to execute, which helps to improve the effect of rehabilitation training.

[0069] According to one embodiment of the present invention, the conditional functions of each posture detection device (excluding the reference posture detection device) are summed to obtain the loss function of the nursing rehabilitation action generation model. The parameters of the nursing rehabilitation action generation model are adjusted by the loss function to minimize the loss function. After multiple trainings, the trained nursing rehabilitation action generation model can be obtained and can be used to generate the action posture vector of the rehabilitation action of the next nursing cycle.

[0070] In this way, the loss function can be reasonably set through the sample action posture vector, the sample initial posture vector, the sample target posture vector and the sample posture vector at the previous moment of the sample target posture vector, so that the action corresponding to the sample action posture vector is executed in the same way as the action corresponding to the sample target posture vector, but with a slightly smaller amplitude, and the action corresponding to the sample action posture vector is more in line with the movement characteristics of the human body, making the action corresponding to the vector easy to execute, which helps to improve the effect of rehabilitation training.

[0071] According to one embodiment of the present invention, in step S107, computer graphics or animation software is used to convert the process of the rehabilitation training personnel's rehabilitation action from the initial posture vector to the action posture vector of the rehabilitation action of the next nursing cycle into an action video, which is convenient for the rehabilitation training personnel to imitate and execute, and helps to improve the rehabilitation training effect.

[0072] According to the intelligent auxiliary method for nursing rehabilitation of the embodiment of the present invention, by monitoring the physiological indicators and posture vectors of the rehabilitation trainees at multiple times, the rehabilitation progress and needs of the rehabilitation trainees can be understood more accurately, thereby realizing a more personalized nursing plan. Through the physiological indicator detection device and the posture detection device, the physiological state and posture of the rehabilitation trainees can be monitored in real time, and problems can be found and adjusted in time. The rehabilitation movements are adjusted according to the moment and posture when the rehabilitation trainees feel the pain stimulation, so as to find the appropriate rehabilitation movements in different rehabilitation training stages, and improve the appropriateness and safety of rehabilitation training. Using the trained nursing rehabilitation movement generation model, the rehabilitation movements of the next nursing cycle are generated according to the target posture vector and the initial posture vector, helping the rehabilitation trainees to practice the correct movements. The action video of the rehabilitation movement generated for the next nursing cycle can help the rehabilitation trainees understand the correct action execution method more intuitively, and improve the training effect and learning efficiency. When determining the first target moment, the discrimination index can be obtained by the overall degree of deviation of multiple physiological indicators at adjacent moments in the current nursing cycle from the reference standard index, thereby determining the first target moment when the rehabilitation trainee may feel pain stimulation, which helps to accurately determine the posture in which the rehabilitation trainee feels pain when performing rehabilitation movements, and is conducive to more accurate adjustment of rehabilitation movements. When determining the target posture vector, the relative motion vector of the posture detection device relative to the reference posture detection device in the time period from the previous moment to the third target moment, and the relative motion vector of the posture detection device relative to the reference posture detection device in the time period from the third target moment to the next moment of the third target moment can be determined, and the motion direction between the two relative motion vectors can be determined to obtain the target posture discrimination parameter, thereby objectively and accurately judging whether the rehabilitation trainee is subjected to pain stimulation at the third target moment, and then determining the target posture vector, providing a data basis for the subsequent effective management and control of rehabilitation movements in the nursing cycle. When training the nursing rehabilitation action generation model, the loss function can be reasonably set through the sample action posture vector, the sample initial posture vector, the sample target posture vector and the sample posture vector at the previous moment of the sample target posture vector, so that the action corresponding to the sample action posture vector is executed in the same way as the action corresponding to the sample target posture vector, but with a slightly smaller amplitude, and the action corresponding to the sample action posture vector is more in line with the movement characteristics of the human body, so that the action corresponding to the vector is easy to execute, which helps to improve the effect of rehabilitation training.

[0073] Figure 2 A block diagram of an intelligent assistance system for nursing rehabilitation according to an embodiment of the present invention is exemplarily shown, and the system includes:

[0074] The physiological index module is used to detect multiple physiological indexes of rehabilitation training personnel through physiological index detection equipment at multiple moments in the current nursing cycle;

[0075] A posture vector module, used for detecting the posture vector of the rehabilitation trainee at multiple moments in the current nursing cycle through a posture detection device, wherein the posture detection device includes a head position detection device, a neck position detection device, a shoulder position detection device, an elbow position detection device, a wrist position detection device, a waist position detection device, a knee position detection device and an ankle position detection device, and the posture vector is used to represent the relative position relationship between the various posture detection devices;

[0076] The first target moment module is used to determine the first target moment at which the rehabilitation training personnel feel pain stimulation from among multiple moments at the end of the current nursing cycle according to multiple physiological indicators obtained at multiple moments;

[0077] A second target time module, used to determine, from among the multiple moments, a second target time for the rehabilitation personnel to perform the rehabilitation action of the current nursing cycle;

[0078] A target posture vector module, used to determine a target posture vector in the rehabilitation action of the current nursing cycle according to the first target moment, the second target moment and the posture vector corresponding to the second target moment, wherein the target posture vector is used to represent the posture of the rehabilitation training personnel feeling pain stimulation when performing the rehabilitation action of the current nursing cycle;

[0079] The action posture vector module is used to obtain the initial posture vector of the rehabilitation trainee, and input the target posture vector, the initial posture vector and the posture vector at the previous moment of the target posture vector into the trained nursing rehabilitation action generation model to obtain the action posture vector of the rehabilitation action of the next nursing cycle, wherein the initial posture vector is used to represent the relative position relationship between the various posture detection devices when the rehabilitation trainee is in a non-action state;

[0080] The action video module is used to generate an action video of the rehabilitation action for the next nursing cycle according to the action posture vector.

[0081] According to one embodiment of the present invention, there is provided an intelligent assistance device for nursing and rehabilitation, comprising: a processor; a memory for storing instructions executable by the processor; wherein the processor is configured to call the instructions stored in the memory to execute the intelligent assistance method for nursing and rehabilitation.

[0082] According to one embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the intelligent assistance method for nursing rehabilitation is implemented.

[0083] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0084] It should be understood by those skilled in the art that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The purpose of the present invention has been fully and effectively achieved. The functional and structural principles of the present invention have been demonstrated and explained in the embodiments, and the embodiments of the present invention may be deformed or modified in any way without departing from the principles.

[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent assistance method for nursing rehabilitation, It is characterized in that include: At multiple moments in the current nursing cycle, multiple physiological indicators of the rehabilitation training personnel are detected by physiological indicator detection equipment; At multiple moments in the current care cycle, the posture vector of the rehabilitation trainee is detected by a posture detection device, wherein the posture detection device includes a head position detection device, a neck position detection device, a shoulder position detection device, an elbow position detection device, a wrist position detection device, a waist position detection device, a knee position detection device and an ankle position detection device, and the posture vector is used to represent the relative position relationship between the posture detection devices; At the end of the current nursing cycle, based on multiple physiological indicators obtained at multiple moments, determine the first target moment at which the rehabilitation training personnel feel pain stimulation among multiple moments; According to the posture vectors obtained at the multiple moments, determining, among the multiple moments, a second target moment for the rehabilitation personnel to perform the rehabilitation action of the current nursing cycle; Determine a target posture vector in the rehabilitation action of the current nursing cycle according to the first target moment, the second target moment and the posture vector corresponding to the second target moment, wherein the target posture vector is used to represent the posture of the rehabilitation trainee feeling pain stimulation when performing the rehabilitation action of the current nursing cycle; Acquire an initial posture vector of the rehabilitation trainee, and input the target posture vector, the initial posture vector and the posture vector at the previous moment of the target posture vector into the trained nursing rehabilitation action generation model to obtain the action posture vector of the rehabilitation action of the next nursing cycle, wherein the initial posture vector is used to represent the relative position relationship between the various posture detection devices when the rehabilitation trainee is in a non-action state; generating an action video of a rehabilitation action for the next nursing cycle according to the action posture vector; At the end of the current care cycle, based on multiple physiological indicators obtained at multiple moments, the first target moment at which the rehabilitation training personnel feel pain stimulation is determined at multiple moments, including: Obtain reference standard indicators corresponding to multiple physiological indicators; Based on the relative deviation indicators of multiple physiological indicators at each moment and the reference standard indicators; According to the relative deviation indexes of multiple physiological indexes at each moment and the reference standard indexes, the first target moment when the rehabilitation training personnel feel the pain stimulation is determined; According to the relative deviation indicators of multiple physiological indicators at each moment and the reference standard indicators, the first target moment when the rehabilitation training personnel feel the pain stimulation is determined, including: According to the formula ， Get the discrimination index at the i-th moment ,in, is the jth physiological index at the i-th moment in the current nursing cycle, is the jth physiological index at the i+1th moment in the current nursing cycle, is the reference standard index of the jth physiological index, m is the number of types of physiological indexes, is the relative deviation index of the jth physiological index at the i-th moment in the current nursing cycle, j≤m, and both j and m are positive integers; If the discrimination index of the i-th moment is greater than or equal to the preset discrimination threshold, the i-th moment is determined as the first target moment.

2. The intelligent assistance method for nursing rehabilitation according to claim 1, It is characterized in that According to the posture vectors obtained at multiple moments, determining the second target moment for the rehabilitation personnel to perform the rehabilitation action of the current nursing cycle in the multiple moments, including: Obtaining the action posture vector of the rehabilitation action in the current nursing cycle; Determine, among the posture vectors obtained at multiple moments, a first posture vector whose cosine similarity with the action posture vector of the rehabilitation action of the current nursing cycle is greater than or equal to a preset similarity threshold; The time corresponding to the first posture vector is determined as the second target time.

3. The intelligent assistance method for nursing rehabilitation according to claim 1, It is characterized in that Determining a target posture vector in a rehabilitation action of a current nursing cycle according to the first target moment, the second target moment, and a posture vector corresponding to the second target moment includes: Determining a third target time in the intersection of the set of first target times and the set of second target times; Acquire a posture vector corresponding to a third target moment, a posture vector corresponding to a moment before the third target moment, and a posture vector corresponding to a moment after the third target moment; Determine a target posture discrimination parameter at the third target moment according to a posture vector corresponding to the third target moment, a posture vector corresponding to a moment before the third target moment, and a posture vector corresponding to a moment after the third target moment; If the target posture discrimination parameter at the third target moment is greater than or equal to the preset posture discrimination threshold, the posture vector corresponding to the third target moment is determined as the target posture vector in the rehabilitation action of the current nursing cycle.

4. The intelligent assistance method for nursing rehabilitation according to claim 3, It is characterized in that Determining a target posture discrimination parameter at the third target moment according to a posture vector corresponding to the third target moment, a posture vector corresponding to a moment before the third target moment, and a posture vector corresponding to a moment after the third target moment, comprising: According to the formula ， Determine the target attitude discrimination parameters at the kth third target moment ,in, is a vector composed of elements in the posture vector corresponding to the kth third target moment, representing the relative position relationship between the sth posture detection device and the reference posture detection device, wherein the reference posture detection device is a head position detection device or a neck position detection device, S is the total number of posture detection devices, is a vector composed of elements representing the relative positional relationship between the s-th attitude detection device and the reference attitude detection device in the attitude vector corresponding to the moment before the k-th third target moment, is a vector composed of elements in the posture vector corresponding to the moment after the kth third target moment, representing the relative position relationship between the sth posture detection device and the reference posture detection device, s≤S-1, and s and S are both positive integers, and if is a conditional function.

5. The intelligent assistance method for nursing rehabilitation according to claim 1, It is characterized in that The training steps of the nursing rehabilitation action generation model include: Inputting the sample initial posture vector, the sample target posture vector and the sample posture vector at the previous moment of the sample target posture vector into the nursing rehabilitation action generation model to obtain the sample action posture vector; Obtaining a loss function of a nursing rehabilitation action generation model according to the sample action posture vector, the sample initial posture vector, the sample target posture vector, and the sample posture vector at a previous moment of the sample target posture vector; According to the loss function, the nursing rehabilitation action generation model is trained to obtain a trained nursing rehabilitation action generation model.

6. The intelligent assistance method for nursing rehabilitation according to claim 5, It is characterized in that According to the sample action posture vector, the sample initial posture vector, the sample target posture vector and the sample posture vector at the previous moment of the sample target posture vector, a loss function of the nursing rehabilitation action generation model is obtained, including: According to the formula ， Obtaining the loss function of the nursing rehabilitation action generation model ,in, is a vector composed of elements in the sample target posture vector representing the relative position relationship between the sth posture detection device and the reference posture detection device, wherein the reference posture detection device is a head position detection device or a neck position detection device, S is the total number of posture detection devices, is a vector composed of elements representing the relative position relationship between the sth posture detection device and the reference posture detection device in the sample posture vector at the previous moment of the sample target posture vector, is a vector composed of elements in the sample action posture vector representing the relative position relationship between the sth posture detection device and the reference posture detection device, is the vertical angle between the vector composed of the elements in the sample target posture vector representing the relative position relationship between the s-th posture detection device and the reference posture detection device and the vector composed of the elements in the sample initial posture vector representing the relative position relationship between the s-th posture detection device and the reference posture detection device, is the vertical angle between the vector composed of the elements representing the relative position relationship between the s-th posture detection device and the reference posture detection device in the sample posture vector at the previous moment of the sample target posture vector and the vector composed of the elements representing the relative position relationship between the s-th posture detection device and the reference posture detection device in the sample initial posture vector, is the horizontal angle between the vector composed of the elements in the sample target posture vector representing the relative position relationship between the s-th posture detection device and the reference posture detection device and the vector composed of the elements in the sample initial posture vector representing the relative position relationship between the s-th posture detection device and the reference posture detection device, is the horizontal angle between the vector composed of the elements representing the relative position relationship between the s-th posture detection device and the reference posture detection device in the sample posture vector at the previous moment of the sample target posture vector and the vector composed of the elements representing the relative position relationship between the s-th posture detection device and the reference posture detection device in the sample initial posture vector, and is the preset weight, and , s≤S-1, and s and S are both positive integers, if is a conditional function.

7. An intelligent assistance system for nursing rehabilitation for executing the method according to any one of claims 1 to 6, It is characterized in that include: The physiological index module is used to detect multiple physiological indexes of rehabilitation training personnel through physiological index detection equipment at multiple moments in the current nursing cycle; A posture vector module, used for detecting the posture vector of the rehabilitation trainee at multiple moments in the current nursing cycle through a posture detection device, wherein the posture detection device includes a head position detection device, a neck position detection device, a shoulder position detection device, an elbow position detection device, a wrist position detection device, a waist position detection device, a knee position detection device and an ankle position detection device, and the posture vector is used to represent the relative position relationship between the various posture detection devices; The first target moment module is used to determine the first target moment at which the rehabilitation training personnel feel pain stimulation from among multiple moments at the end of the current nursing cycle according to multiple physiological indicators obtained at multiple moments; A second target time module, used to determine, from among the multiple moments, a second target time for the rehabilitation personnel to perform the rehabilitation action of the current nursing cycle; A target posture vector module, used to determine a target posture vector in the rehabilitation action of the current nursing cycle according to the first target moment, the second target moment and the posture vector corresponding to the second target moment, wherein the target posture vector is used to represent the posture of the rehabilitation training personnel feeling pain stimulation when performing the rehabilitation action of the current nursing cycle; The action posture vector module is used to obtain the initial posture vector of the rehabilitation trainee, and input the target posture vector, the initial posture vector and the posture vector at the previous moment of the target posture vector into the trained nursing rehabilitation action generation model to obtain the action posture vector of the rehabilitation action of the next nursing cycle, wherein the initial posture vector is used to represent the relative position relationship between the various posture detection devices when the rehabilitation trainee is in a non-action state; The action video module is used to generate an action video of the rehabilitation action for the next nursing cycle according to the action posture vector.

8. A computer-readable storage medium, It is characterized in that Computer program instructions are stored thereon, and when the computer program instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images