CN113362925B - Automatic good limb position control method and device - Google Patents

Abstract

The embodiment of the specification discloses an automatic limb position control method and equipment, wherein the method comprises the following steps: the controller of the mattress sends a good limb position use request of a user to the medical care node, so that the medical care node can call case information of the user conveniently, and when the medical care node confirms the good limb position use request, the good limb position control platform determines a good limb position control scheme of the user according to the case information and the body characteristic information of the user; the good limb position control platform determines a region to be controlled of the mattress and a control instruction according to a good limb position control scheme, wherein the control instruction comprises inflation or deflation operation of the region to be controlled, inflation time or deflation time of the region to be controlled and inflation quantity of the region to be controlled; the good limb position control platform sends the good limb position control scheme to the controller of the mattress, so that the controller of the mattress controls the area to be controlled of the mattress according to the control instruction, and a user is assisted to complete the good limb position control. And a customized good limb position control scheme is generated according to the actual conditions of different users, so that the automatic control of the good limb position is realized.

Description

A kind of automatic good limb position control method and equipment

technical field

The present specification relates to the field of medical technology, and in particular, to a method and device for automatic good limb position control.

Background technique

A good limb position is to place the limb in a position or maintain a posture in order to maintain the good function of the limb, which is one of the important measures for early antispasm. This good limb position can make the joint after hemiplegia relatively stable, can effectively prevent the typical spastic pattern of the upper limb flexor and lower limb extensor, and is also one of the methods to prevent the occurrence of pathological movement patterns in the future. In addition, the limbs of patients with neurological diseases (mainly including cerebrovascular accident, brain trauma, etc.) may experience spasm (upper limb flexor, lower limb extensor) during the functional recovery process. Therefore, the placement, maintenance and regular replacement of the good limb position of patients with such diseases are very important for the recovery of the disease.

At present, most of the patients' good limb positions need to be adjusted by family members and nurses, and bedding, pillows, etc. are used to assist in fixing good limb positions, which adds extra workload and greater physical load to family members and nurses. Automatic control method of good limb position.

SUMMARY OF THE INVENTION

One or more embodiments of this specification provide an automatic good limb position control method and device, which are used to solve the following technical problems: in the prior art, the conscience position is placed and fixed on items such as pillows, and there is a lack of an automatic good limb position control method. Methods.

One or more embodiments of this specification adopt the following technical solutions:

One or more embodiments of the present specification provide an automatic good limb position control method, the method includes: a mattress controller sends a user's good limb position use request to a medical care node, so that the medical care node can call the User's case information, wherein the request includes user identification and physical feature information; when the medical and nursing node confirms the request for using the good limb position, the good limb position control platform according to the user's case information and the body feature information, and determine the user's good limb position control scheme, wherein the good limb position control scheme includes the amount of inflation in the to-be-controlled area, the position of the good limb position, and the time corresponding to the position of the good limb position ; The good limb position control platform determines the mattress to be controlled area and control instructions according to the good limb position control scheme, wherein the control instructions include the inflatable or deflated operation of the to-be-controlled area, the inflation time of the to-be-controlled area or The deflation time and the inflation volume of the area to be controlled; the good limb position control platform sends the good limb position control scheme to the controller of the mattress, so that the controller of the mattress can follow the control The instruction controls the to-be-controlled area of the mattress, so as to assist the user to complete the control of the good limb position.

Further, before the sending the good limb position control scheme to the controller of the mattress, the method further includes: setting the area to be controlled of the mattress as a body placement area and a body adjacent area, so that the The controller of the mattress controls the adjacent area of the body to execute the control instruction, and assists the user to complete the control of the good limb position; wherein the to-be-controlled area includes: the left shoulder area, the right shoulder area, the left upper limb area, the right upper limb area A region, a left lower extremity region, and a right lower extremity region, each of which includes one or more of the body placement regions and at least one of the body adjacent regions.

Further, determining the user's good limb position control scheme according to the user's case information and the physical feature information, specifically includes: constructing an initial neural network model, collecting medical institution case information based on web crawler technology, The physical feature information and the relevant data of the good limb position control plan, according to the medical institution case information, body feature information and the relevant data of the good limb position control plan, train the initial neural network model, and obtain the required good limb position. Control model; input the user's case information and the body feature information into the good limb position control model that meets the requirements, and determine the user's good limb position control scheme.

Further, according to the medical institution case information, body feature information and relevant data of the good limb position control scheme, the initial neural network model is trained to obtain a good limb position control model that meets the requirements, specifically including: Group medical institution case information, body feature information and data of good limb position control plan to construct a dataset, and divide the dataset into training set and validation set according to a preset ratio; construct the network structure of the convolutional neural network model , wherein, layers 0-95 of the convolutional neural network are deep convolutional layers, which are composed of convolution blocks and single-layer convolutional layers, and layers 96-126 of the convolutional neural network are feature interaction layers; preset Training parameters, wherein the training parameters include: initial learning rate, weight decay rate and maximum number of iterations, and input the medical institution case information, body feature information and relevant data of the good limb position control scheme in the training set into the convolutional neural network Loop iterative training is carried out in the network model; the medical institution case information, body feature information and good limb position control scheme in the verification set are input into the good limb position control model, and the loss function value in the training process of the model is recorded. Loss function value, adjust the connection weight between each neuron in the neural network model, when the loss function value reaches the preset loss function threshold or the number of training times reaches the preset maximum number of iterations, a good limb position that meets the requirements is obtained A control model; wherein, the loss function is a cross-entropy loss function.

Further, determining the user's good limb position control scheme according to the user's case information and the body feature information specifically includes: determining the user's first limb position according to the user's body feature information. A good limb position control scheme, wherein the body feature information includes: the user's height and the user's weight, and the first good limb position control scheme includes: the initial position where the user is placed and the inflation volume of the area to be controlled; The first good limb position control scheme of the user is determined according to the physical feature information of the user, and specifically includes: a relationship between a preset body index and the inflation volume, wherein the body index is the weight of the user and the When the body index is less than the first preset threshold, the inflation volume is the first inflation volume; when the body index is greater than the first preset threshold and less than the second preset threshold when the inflation volume is the second inflation volume; when the body index is greater than the second preset threshold, the inflation intensity is the third inflation volume; the first inflation volume is greater than the second inflation volume , the second inflation amount is greater than the third inflation amount.

Further, after determining the first good limb position control scheme of the user according to the physical feature information of the user, the method further includes: according to the case information of the user, determining the use of good limb position Type and corresponding second good limb position control scheme, wherein the second good limb position control scheme includes the position of good limb position and the time of placing the good limb position, different good limb positions use The types correspond to different placement positions of the good limb positions and the time of placing the good limb positions.

Further, determining the area to be controlled and the control instruction of the mattress according to the good limb position control scheme specifically includes: according to the position of the good limb position in the good limb position control scheme, determining the area to be controlled and the control instruction. The mattress inflation or deflation operation according to the invention, wherein the position of the good limb position includes: supine position, lying position on the healthy side, and lying position on the affected side; and the position change time in the good limb position control scheme is determined according to the position change time. Mattress inflation time.

Further, before the controller of the mattress controls the to-be-controlled area of the mattress according to the control instruction, the method further includes: the controller of the mattress constructs coordinates with the mattress preset position as an origin system, and record the first position coordinates of the first preset part of the user and the second position coordinates of the second preset part; after the assisting the user to complete the control of the good limb position, the method further includes: the The controller of the mattress records the third position coordinate of the first preset part and the fourth position coordinate of the second preset part; The three position coordinates and the fourth position coordinates determine whether the user uses the good limb position correctly, which specifically includes: according to the difference between the first position coordinates and the second position coordinates, determining that the user has completed the good limb position The position situation before the position is used to reflect the position of each part of the body before the user completes the good limb position; according to the difference between the third position coordinate and the fourth position coordinate, determine the position The position of the user after completing the good limb position, and the position status is used to reflect the position of each part of the body after the user completes the good limb position; The position of the user after completing the good limb position is determined, whether the user moves according to the good limb position control scheme, and whether the current good limb position of the user is in the correct position.

Further, before the controller of the mattress controls the area to be controlled of the mattress according to the control instruction, the good limb position control platform obtains the skeleton key point at the first preset time according to the human body key point prediction network. The first position specifically includes: collecting the image data of the user before completing the control of the good limb position, inputting the image data into the human body key point prediction network, outputting the predicted probability that each pixel point in the image is a bone key point, and A heat map is formed, and the heat map is converted into the numerical coordinates of the skeleton key points by the soft-argmax function and output, wherein the human body key point prediction network is based on the CPN feature pyramid network and introduces the CBAM attention model to obtain, the CPN The feature pyramid network includes the GlobalNet model and the RefineNet model; the predicted position of the skeleton key point is obtained according to the first position and the instantaneous speed at the first preset time, and the first position and the predicted position are obtained by Q=AQ ¹ +BP+C to obtain the position trajectory of the skeleton key point; wherein, the Q is a feature vector, used to represent the state information of the skeleton key point at the second preset moment, the state information of the skeleton key point It includes the speed information of the skeleton key point and the coordinate information of the skeleton key point, and the A is a state transformation matrix, which is used to associate the state information of the skeleton key point at the first preset moment with the The state of the skeleton key point at the second preset time, the first preset time is the previous time of the second preset time; according to the position trajectory of the skeleton key point, it is judged whether the user uses a good limb correctly bit.

One or more embodiments of the present specification provide an automatic good limb position control device, including:

at least one processor; and,

a memory communicatively coupled to the at least one processor; wherein,

The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to: a controller of the mattress to A use request is sent to the medical care node, so that the medical care node can call the user's case information, wherein the request includes the user identification and physical feature information; when the medical care node confirms the use request of the good limb position, The good limb position control platform determines the user's good limb position control scheme according to the user's case information and the physical feature information, wherein the good limb position control scheme includes the air volume of the area to be controlled, the good limb position position information and the time corresponding to the position of the good limb position; the good limb position control platform determines the area to be controlled and the control instruction of the mattress according to the good limb position control scheme, wherein the control instruction includes The inflation or deflation operation of the area to be controlled, the inflation time or deflation time of the area to be controlled, and the inflation amount of the area to be controlled; the good limb position control platform sends the good limb position control plan to the mattress The controller, so that the controller of the mattress controls the area to be controlled of the mattress according to the control instruction, so as to assist the user to complete the control of the good limb position.

The above-mentioned at least one technical solution adopted in the embodiments of this specification can achieve the following beneficial effects: determining the corresponding good limb position control scheme according to the user's case information and physical feature information, and realizing the user-specific good limb position customization according to the user's actual situation According to the good limb position control scheme, the mattress to be controlled area and control instructions are determined, and the inflatable or deflated operation of the to-be-controlled area is realized through the mattress controller to achieve the technical effect of helping the user to realize the automatic control of the good limb position.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present specification or the prior art, the following briefly introduces the accompanying drawings required in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments described in this specification. For those of ordinary skill in the art, other drawings can also be obtained according to these drawings without creative labor. In the attached image:

1 is an application scenario diagram of an automatic good limb position control method provided by an embodiment of the present specification;

2 is a schematic flowchart of an automatic good limb position control method provided by an embodiment of the present specification;

FIG. 3 is a schematic structural diagram of the area to be controlled of the mattress provided by the embodiment of the present specification;

FIG. 4 is a schematic diagram of the left lower limb region provided in the embodiment of the present specification;

FIG. 5 is a diagram of the positional relationship between a patient and a mattress according to an embodiment of the present specification;

FIG. 6 is a schematic structural diagram of an automatic good limb position control device according to an embodiment of the present specification.

Detailed ways

In order to make those skilled in the art better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings in the embodiments of this specification. Obviously, the described The embodiments are only some of the embodiments of the present specification, but not all of the embodiments. Based on the embodiments of the present specification, all other embodiments obtained by persons of ordinary skill in the art without creative efforts shall fall within the protection scope of the present specification.

A good limb position is to place the limb in a position or maintain a posture in order to maintain the good function of the limb. Sexual position. This good limb position can make the joint after hemiplegia relatively stable, can effectively prevent the typical spastic pattern of the upper limb flexor and lower limb extensor, and is also one of the methods to prevent the occurrence of pathological movement patterns in the future. Clinically, according to the different conditions of the patients, the requirements for the placement of the good limbs are also different. For example, before recovering from anesthesia, surgical patients should take a lateral position in different directions depending on the surgical site, so that oral secretions and vomitus can easily flow out, preventing vomit reflux caused by anesthesia drugs and causing suffocation; for example, cardiopulmonary resuscitation patients , should be kept supine, and the head position should be kept back to ensure the patient's airway is smooth; for example, the same stroke patient, if the patient is a conscious stroke patient, the best posture is to sit on a chair. The patient is a comatose stroke patient, and the best posture is to lie down; for another example, stroke patients usually suffer from different degrees of sensory impairment, limb impairment and cognitive impairment, which affect the patient's self-care ability and life ability. For example, it is necessary to take scientific good limb placement for rehabilitation care. For hemiplegic patients with cerebral infarction, good limb position can improve the different degrees of limb dysfunction left by patients after cerebral infarction, and can reduce the heavy burden on individuals, families and society.

The embodiment of the present specification provides an automatic good limb position control method, and FIG. 1 is an application scenario diagram of the good limb position control method provided by the embodiment of the present specification. As shown in FIG. 1 , it includes a healthy limb position control platform 1 , a medical care node 2 and a mattress controller 3 , wherein the healthy limb position control platform includes a server 11 . The patient triggers the operation of using the good limb position on the mattress controller 3. The mattress controller 3 generates a request for the use of the good limb position, and sends the request for the use of the good limb position to the medical care node 2. The limb position use request is confirmed, and the patient's case information is called according to the use request, and the case information and the confirmed good limb position use request are sent to the good limb position control platform 1, and the server 11 generates a good limb position control plan, and according to the good limb position control plan The limb position control plan determines the area to be controlled and the control command of the mattress, and sends the good limb position control plan to the controller 3 of the mattress. The mattress controller controls the area to be controlled on the mattress according to the control command, and assists the patient to complete the good limb position. control.

FIG. 2 is a schematic flowchart of an automatic good limb position control method provided by the embodiment of the present specification. As shown in FIG. 2 , an automatic good limb position control method mainly includes:

Step S201, the mattress controller sends the user's good limb position use request to the medical care node, so that the medical care node can call the user's case information, wherein the request includes the user identification and physical feature information. In one embodiment of this specification, After the user triggers the operation of using the good limb position, the mattress controller generates a use request, and sends the user's good limb position use request to the medical care node, where the request includes the user's identity and body feature information. It should be noted that the user identification refers to the identification of the patient who is to use a good limb position, which may be an ID number or other unique identification, which is not specifically limited in this manual. In addition, the body feature information is collected by the information collection module of the mattress, and the body feature information includes the patient's height, weight, trunk length, limb length, shoulder width and other information. It should be noted that different patients have different body feature information. After collecting the body feature information, a good limb position usage plan can be customized according to the patient's physical features, bringing a better user experience to the patient and further obtaining better treatment. Effect.

In an embodiment of this specification, the user who triggers the operation using the good limb position may be the patient's family member or the patient's nursing staff. When the patient is awake, the user who triggers the operation using the good limb position may also be the patient himself. In addition, the user's trigger operation may be in the form of clicking a button of the mattress controller, clicking on the touch screen of the mattress controller, or in the form of voice recognition, which is convenient for the user to use.

In an embodiment of this specification, after the mattress controller sends the request for using the good limb position to the medical care node, the medical care node calls the user's case information according to the user ID in the good limb position use request. It should be noted that, The user's case information includes past medical history, treatment process records, etc. Through the past medical history in the user's case information, the use plan of the good limb position can be adjusted according to the actual situation of the patient. For example, patient A is a surgical patient and needs to lie on the left side and the right side, but patient A's past medical history shows that the left lower extremity has been operated on half a year ago, and heavy pressure should be avoided. Therefore, for patient A, The best way to use the limb position is to lie on the right side to prevent suffocation caused by aspiration of postoperative vomitus.

Step S202, when the medical and nursing node confirms the request for using the good limb position, the good limb position control platform determines the user's good limb position control scheme according to the user's case information and physical feature information, wherein the good limb position control scheme includes the area to be controlled. Inflation volume, position information of good limb position, and time corresponding to the position of good limb position.

In one embodiment of this specification, after receiving the user's request for using the good limb position, the medical care node confirms the user's request for using the good limb position, and the good limb position control platform is based on the confirmation of the user's request by the medical care node according to the user's The case information and physical feature information determine the user's good limb position control scheme. It should be noted that the medical care node can determine whether the user is suitable for using the good limb position at the current stage according to the actual situation of the user, and confirm the user's request for using the good limb position if it is determined that the user is suitable for using the good limb position.

In one embodiment of this specification, the good limb position control platform can determine the good limb position control scheme for use according to the user's case information and body feature information, and the control scheme includes the position of the good limb position. It should be noted that, The good limb position includes supine position and lateral position, and lateral position also includes healthy side lying position and affected side lying position. The affected side lying position can be the side of the user's body that is not aware of it, or it can refer to the post-operative operation. part side. Next, taking the hemiplegic patient with cerebral infarction as an example, the standard for the placement of the good limbs will be explained. The first is the standard placement of the supine position. Put a thin pillow on the patient's head, a long pillow on the scapula on the affected side and the upper limb, supination and abduction of the upper arm by 20° to 40°, straighten the elbow and wrist, stretch the fingers, and place the palm upward, and place the lower hip, hip, and outer thigh. Pillow, slightly padded under the knees, toes up. The second is the standard placement of the lying position on the affected side, keeping the affected side down, the healthy side up, pillow on the head, pillow on the back, abduction of the affected arm, forward extension and supination, forward extension of the affected shoulder, and supination of the forearm, The upper limbs and the trunk are at 90°, the elbows and wrists are straight, the palms are upward, the lower limbs on the affected side are placed on the bed in a slightly flexed position, the healthy legs are flexed, the hips are flexed, and the knees are placed forward on the long pillow. on the pillow or on the torso. The last is the standard placement position of the unaffected side lying position. Keep the healthy side down and the affected side up. The head is pillowed, and the upper limbs of the affected side are stretched on the pillow, so that the scapula of the affected side is extended forward and outward by 90°. ～100°, the forearm is pronated, the fingers are extended, the palm is down, the lower limb on the affected side is in a slightly flexed position, and placed on the long pillow, the ankle joint on the affected side cannot be inverted and hung on the edge of the pillow to prevent varus and sagging of the foot.

In an embodiment of the present specification, the good limb position control scheme further includes the amount of inflation in the area to be controlled and the time corresponding to the position of the good limb position. Different inflation volumes are determined according to the physical condition of the patient, which has enhanced the user's comfort when using a good limb position, and avoided secondary injury caused by squeezing the patient's limb due to excessive inflation. In addition, the control plan also includes the placement time corresponding to the good limb position. Since most of the patients who use the good limb position are hemiplegia patients or patients with inconvenient movement, they cannot move normally. Set each good limb position according to the actual situation of the patient. The release time can avoid the occurrence of complications such as bedsores and deep vein thrombosis caused by the patient staying in one position for a long time.

Specifically, step S202 specifically includes: determining the user's good limb position control scheme according to the user's case information and body feature information, which specifically includes: constructing an initial neural network model, and collecting medical institution case information and body feature information based on web crawling technology. With the relevant data of the good limb position control scheme, the initial neural network model is trained according to the medical institution case information, body feature information and the relevant data of the good limb position control scheme, and the good limb position control model that meets the requirements is obtained; The user's good limb position control scheme is determined by inputting the information of physical characteristics and physical characteristics into the good limb position control model that meets the requirements. According to the medical institution case information, body feature information and the relevant data of the good limb position control plan, train the initial neural network model to obtain a good limb position control model that meets the requirements, which specifically includes: combining multiple sets of medical institution case information and body feature information As well as the data of the good limb position control scheme, the data set is constructed, and the data set is divided into training set and verification set according to the preset ratio; the network structure of the convolutional neural network model is constructed, among which, the 0-95 of the convolutional neural network The layer is a deep convolution layer, which consists of a convolution block and a single-layer convolution layer. The 96-126 layers of the convolutional neural network are the feature interaction layers; the training parameters are preset, and the training parameters include: initial learning rate, weight decay rate and the maximum number of iterations, input the medical institution case information, body feature information and relevant data of the good limb position control plan in the training set into the convolutional neural network model for cyclic iterative training; The information and the good limb position control scheme are input into the good limb position control model, the loss function value during the model training process is recorded, and the connection weight between each neuron in the neural network model is adjusted according to the loss function value. When the loss function value reaches When the preset loss function threshold or the number of training times reaches the preset maximum number of iterations, a good limb position control model that meets the requirements is obtained; wherein, the loss function is a cross-entropy loss function.

In an embodiment of the present specification, a good limb position control scheme of the user is obtained through a good limb position control model. First, build an initial neural network model. The initial neural network model can be a convolutional neural network structure. The 0-95 layers of the convolutional neural network are deep convolutional layers, which are composed of convolutional blocks and single-layer convolutional layers. , 96-126 layers are feature interaction layers. After that, based on the web crawler technology, the medical institution case information, the user's physical feature information and the data related to the good limb position control plan are collected. Programs or scripts, which are widely used in Internet search engines or other similar websites, can automatically collect the content of all pages that they can access, so as to obtain or update the content and retrieval methods of these websites. The collected data of multiple sets of medical institution case information, multiple sets of body feature information and the data of the good limb position control plan are processed, and the processed data is constructed into a data set, and the data set is divided into training according to the ratio of 7:3 set and validation set. It should be noted that the ratio is a preset random ratio.

In an embodiment of the present specification, the initial neural network model can be trained in the following manner: training parameters are preset, wherein the training parameters include: an initial learning rate, a weight decay rate, and a maximum number of iterations, and the medical Institution case information, body feature information and relevant data of the good limb position control plan are input into the convolutional neural network model for cyclic iterative training, and the medical institution case information, physical feature information and good limb position control plan in the verification set are input into the good limb position control plan. In the limb position control model, the loss function value during the model training process is recorded, and the loss function is the cross entropy loss function. According to the loss function value, adjust the connection weight between each neuron in the neural network model. When the loss function value reaches the preset loss function threshold or the number of training times reaches the preset maximum number of iterations, a good limb position control that meets the requirements is obtained. Model.

In an embodiment of the present specification, the initial neural network model can also be trained by the following method: inputting the medical institution case information, body feature information and relevant data of the good limb position control plan in the training set into the pre-built In the convolutional neural network model, multiple good limb position control models are obtained, and the medical institution case information and physical feature information in the verification set are input into the multiple good limb position control models, and the corresponding good limb position control models are calculated. Good limb position control scheme data, compare the difference between the good limb position control scheme obtained by each good limb position control model and the good limb position control scheme in the validation set. In the position control scheme, the air volume of the area to be controlled, the position information of the good limb position, and the time corresponding to the position of the good limb position are set with different weight values according to the priority. For example, the position of the good limb position occupies the largest weight. , The time weight of the position of the good limb position is the smallest, and the comparison in the present invention does not make a specific limitation. According to the comparison results, the good limb position control model with the smallest difference is selected as the good limb position control model that meets the requirements. It should be noted that this specification does not specifically limit the model training method.

In an embodiment of the present specification, after obtaining a good limb position control model that meets the requirements, the case information and physical feature information of the user to be used with good limb position are input into the good limb position control model that meets the requirements, and the corresponding information of the user is obtained. good limb position control program. Compared with manual generation, the generation of a good limb position control scheme through the model effectively saves the time of medical workers, and avoids the problem of secondary injury to the patient caused by the wrong use of the good limb position by the user.

Specifically, step S202 further includes: determining the user's first good limb position control scheme according to the user's body feature information, wherein the body feature information includes: the user's height and the user's weight, and the first good limb position control scheme Including: placing the initial position of the user and the inflation volume of the area to be controlled; determining the user's first good limb position control scheme according to the user's body feature information, specifically including: the relationship between the preset body index inflation volume, where the body index is The ratio of the user's weight to the square of the user's height; when the body index is less than the first preset threshold, the inflation amount is the first inflation amount; when the body index is greater than the first preset threshold and smaller than the second preset threshold, the inflation amount is The second inflation volume; when the body index is greater than the second preset threshold, the inflation intensity is the third inflation volume; the first inflation volume is greater than the second inflation volume, and the second inflation volume is greater than the third inflation volume.

In an embodiment of the present specification, the user's good limb position control scheme further includes a first good limb position control scheme and a second good limb position control scheme, wherein the first good limb position control scheme includes placing the user's initial position and The air volume of the area to be controlled, the second good limb position control scheme includes the position of good limb position and the time of good limb position.

In an embodiment of the present specification, the first good limb position control scheme is determined according to the user's body feature information. First, the relationship between the body index and the inflation volume is preset, and the body index is the ratio of the user's weight to the square of the user's height. The relationship between the body index and the inflation volume is as follows: when the body index is less than the first preset threshold, the inflation volume is the first inflation volume; when the body index is greater than the first preset threshold and less than the second preset threshold, the inflation volume is The second inflation amount; when the body index is greater than the second preset threshold, the inflation intensity is the third inflation amount, the first inflation amount is greater than the second inflation amount, and the second inflation amount is greater than the third inflation amount. . It should be noted that the first preset threshold may be 18.5, and the second preset threshold may be 24. For example, if the height of user A is 1.65m and the weight is 80KG, then the body index of user A is 80/(1.65 ² ), that is, the body index of user A is 29, which is greater than 24, so the amount of inflation when user A uses a good limb position For low-intensity inflation (third inflation). It should be clear to those skilled in the art that when inflating the area to be controlled, the purpose is to fix the user's body part in a certain area. The user's body can be fixed by filling the surrounding area with low-intensity air volume. On the contrary, if the user's body index is small, the area occupied by the user is small, so the surrounding area needs to be filled with high-intensity air volume to prevent the user. body pulled away. A good limb position control scheme is generated according to the user's physical feature information, which improves the user's comfort.

In an embodiment of this specification, the relationship between the length of the trunk, the length of the limbs and the width of the shoulders and the initial placement position of the user is preset: when the length of the trunk, the length of the limbs and the width of the shoulders are within the preset range, the length of the trunk, The length of the limbs and the width of the shoulders are within the normal range. At this time, the initial placement position of the user can be set according to the general placement scheme. The general placement scheme is set according to the proportion of the human body and does not require manual adjustment. When the trunk length, limb length, and shoulder width are not within the preset ranges, it means that the user's trunk length, limb length, and shoulder width are different from the normal body ratio, and manual adjustment is required based on the general placement scheme.

In one embodiment of this specification, the use type of good limb position is determined according to the user's case information. For example, when user A's case information shows that the user is a post-surgery patient and the surgical site is lower abdominal surgery, in order to avoid the user After the operation, the vomitus was sucked in the supine position. Therefore, the position of the good limb position in the type of good limb position corresponding to the user is the left side and the right side, and the placement time of each position is 30 minutes. For another example, the case information of user B shows that the patient is a hemiplegic patient, and the hemiplegia is on the left side. The recommended positions for good limbs are lying on the unaffected side, supine, and lying on the affected side. When bedsores and other conditions occur, the placement time of each position can be set to 20 minutes; it should be noted that the decubitus position on the healthy side refers to the lateral decubitus position on the non-hemiplegic side (healthy side), and the decubitus position on the affected side refers to the lateral decubitus position on the hemiplegic side. The lateral decubitus position, taking the patient with left hemiplegia as an example, the decubitus position on the healthy side refers to the lateral decubitus position, and the lateral decubitus position refers to the lateral decubitus position. That is to say, according to the determination of the use type of good limb position according to the case information, different use types of good limb position correspond to different positions of good limb position and the time of placement of good limb position, and further, the second good limb position is determined. Position control scheme, the second control scheme includes the placement position of the good limb position and the time of placing the good limb position.

According to the actual situation of the user, a good limb position control scheme is automatically generated, which saves the time of medical workers, and further improves the scientificity of the user's use of good limb position, so that the good limb position control can better help users perform rehabilitation training.

Step S203, the good limb position control platform determines the area to be controlled and the control instruction of the mattress according to the good limb position control scheme, wherein the control instruction includes the inflating or deflating operation of the area to be controlled, the inflation time or deflation time of the area to be controlled, and the time to be controlled. Regional inflation.

Specifically, step S203 specifically includes: determining the area to be controlled and the mattress inflation or deflation operation in the control instruction according to the position of the good limb position in the good limb position control scheme, wherein the position of the good limb position includes: supine position , the unaffected side lying position and the affected side lying position; and the mattress inflation time is determined according to the position change time in the good limb position control scheme.

Before step S204, the method further includes: setting the area to be controlled of the mattress as the body placement area and the area adjacent to the body, so that the controller of the mattress controls the area adjacent to the body to execute control instructions, and assists the user to complete the control of good limbs; wherein the area to be controlled Including: left shoulder area, right shoulder area, left upper body area, right upper body area, left lower body area and right lower body area, each area includes one or more body placement areas and at least one body adjacent area.

In an embodiment of the present specification, FIG. 3 provides a partition of a mattress to be controlled area provided by the embodiment of the present specification. As shown in FIG. 3, the to-be-controlled area of the mattress includes a head area 31, a left shoulder area 32, The right shoulder area 33, the left upper limb area 34, the right upper limb area 35, the left lower limb area 36 and the right lower limb area 37, it should be noted that the left and right here refer to the left and right sides corresponding to standing facing the mattress. The right side, and each area includes the body placement area and the body adjacent area. As shown in Figure 4, the left lower limb area 36 is taken as an example for illustration, area 362 and area 364 are the body placement area, area 361, area 363 and area 365 For the adjacent area of the body. Taking the area 362 as an example, the area 362 is the body placement area, and the surrounding areas (area 361 and 363) have been inflated, and the area 362 can also be filled with a small amount of gas, so that the area 362 has a bulge effect, but the bulge of the area 362 Compared with the area 361 and the area 363, the degree is smaller, which can present the effect of high on both sides and low in the middle, which can not only ensure that the user's affected limb is lifted, but also can use the height difference between the two sides to fix the limb.

In an embodiment of the present specification, the area to be controlled and the control instructions are determined according to the position of the good limb position in the good limb position control scheme. For example, as shown in Figure 5, user C (right hemiplegia refers to the patient's right The good limb position in the good limb position control scheme of the damaged side) is supine position, unaffected side lying position (left lateral position) and affected side lateral position (right lateral position), and the change time is 20 minutes, the inflation volume is medium-intensity inflation volume. 3 and 5, when the supine position is determined, the patient lies flat on the mattress, the right side of the patient actually corresponds to the left side of the mattress, and the corresponding mattress control area in the supine position is the left shoulder Area 32 , left upper extremity area 34 and left lower extremity area 36 . The patient's right shoulder, right upper extremity, and right lower extremity are placed in the body placement areas of the left shoulder area 32, the left upper extremity area 34, and the left lower extremity area 36, respectively, and the left shoulder area 32, the left upper extremity area 34, and the left lower extremity area 36 are placed Inflation is performed in the adjacent area of the body (corresponding to the dark shadow in the mattress in Figure 5), the inflation duration is 20 minutes, and the inflation volume is a medium-intensity inflation volume, so that the mattress can inflate the patient's right shoulder, The right upper extremity and right lower extremity can be lifted and immobilized, and the body placement areas of the left shoulder area 32 , the left upper extremity area 34 , and the left lower extremity area 36 can also be inflated with a small amount of air. In this way, the body placement area presents a slightly raised effect; moreover, the coordination of the body placement area and the body adjacent area with different air volumes can better fix the user's right shoulder, right upper limb and right lower limb, and ensure that The elbows and wrists are straight, the fingers are extended, and the palms are up to achieve the standard position of the supine position. When changing from the supine position to the unaffected side lying position (left lateral position), a deflation operation is performed on the left shoulder region 32, the left upper limb region 34, and the left lower limb region 36, and the control region is determined as the right shoulder region 33, The right upper limb region 35 and the right lower limb region 37 are inflated to the adjacent regions of the body (corresponding to the dark shadow in the mattress in FIG. 5 ), the duration is 20 minutes, and the inflation amount is a medium-intensity inflation amount. The body placement area of the area is filled with a small amount of gas, which can not only lift the patient's limb slightly, but also use the height difference to fix the affected limb. When changing from the unaffected side lying position (left side lying position) to the affected side lying position (right side lying position), a deflation operation is performed on the right shoulder region 33 , the right upper limb region 35 , and the right lower limb region 37 , and the control region Determine the left shoulder area 32, the right shoulder area 33, the left upper limb area 34, the right upper limb area 35 and the right lower limb area 37, and inflate the area adjacent to the body (corresponding to the dark shadow in the mattress in Figure 5) Operation, the inflation time is 20 minutes, and the inflation volume is medium-intensity inflation volume. The mattress is divided into different areas according to the physiological structure of the human body, which improves the universality of the mattress and can be applied to most users; in addition, in the process of good limb position control, first determine the area to be controlled, and then treat the area to be controlled. operation, further reducing the operating pressure of the good limb position control platform.

Step S204, the good limb position control platform sends the good limb position control scheme to the controller of the mattress, so that the mattress controller can control the area to be controlled of the mattress according to the control instruction, so as to assist the user to complete the good limb position control.

In one embodiment of the present specification, the good limb position control scheme is sent to the controller of the mattress, and the controller controls the corresponding area of the mattress to be controlled to perform inflation or deflation operations, and the inflation time or deflation of the area to be controlled. The time and the amount of inflation in the area to be controlled can assist the patient to complete the control of the good limb position.

Specifically, before the controller of the mattress controls the area to be controlled on the mattress according to the control instruction, the method further includes: the controller of the mattress constructs a coordinate system with the preset position of the mattress as an origin, and records the first preset position of the user. the first position coordinates and the second position coordinates of the second preset part; after assisting the user to complete the control of the good limb position, the method further includes: the controller of the mattress records the third position coordinates of the first preset part and the second preset position The fourth position coordinates of the part; according to the first position coordinates, the second position coordinates, the third position coordinates and the fourth position coordinates to determine whether the user uses the good limb position correctly, specifically including: according to the first position coordinates and the second position coordinates The difference value determines the position of the user before completing the good limb position; according to the difference between the third position coordinate and the fourth position coordinate, the position of the user after completing the good limb position is determined; according to the user's completion of the good limb position The position before the position and the position after the user completes the good limb position, determine whether the user moves according to the good limb position control scheme, and determine whether the user's current good limb position is in the correct position.

In an embodiment of the present specification, whether the user uses the good limb position correctly can be monitored by the following method: the controller of the mattress constructs a coordinate system with the preset position of the mattress as the origin, wherein the preset position of the mattress can be the center of the mattress The point, the upper left end point of the mattress, the lower left end point of the mattress, the upper right end point of the mattress, and the lower right end point of the mattress, can also be a position anywhere on the mattress. In addition, the coordinate system can be a three-dimensional coordinate system, and the direction of the coordinate system can be set according to the placement of the mattress. Before the controller of the mattress controls the area to be controlled on the mattress according to the control instruction, record the first position coordinates of the first preset part of the user and the second position coordinates of the second preset part, the first preset part and the second preset part. Let the parts be adjacent joints on the same side of the user's body, for example, the left shoulder joint and the left elbow joint on the left side of the user's body, the left knee joint and the left hip joint on the left side of the user's body. After the auxiliary user completes the control of the good limb position, the third position coordinate of the first preset part and the fourth position coordinate of the second preset part are recorded and recorded. That is to say, the first position coordinate and the second position coordinate are the positions of the first preset part and the second preset part before the user completes the good limb position, and according to the difference between the first position coordinate and the second position coordinate, it is determined The user's position before completing the good limb position. Similarly, the third position coordinate and the fourth position coordinate are the positions of the first preset part and the second preset part after the user completes the good limb position. According to the difference between the third position coordinate and the fourth position coordinate, it can be determined. The position of the user after completing the good limb position. According to two position conditions, it is judged whether the user moves according to the good limb position control scheme, and it is judged whether the current good limb position of the user is in the correct position.

In an embodiment of the present specification, it is also possible to monitor whether the user uses the good limb position correctly in the following manner: after assisting the patient to complete the good limb position control, the standard good limb position state parameters can also be determined according to the good limb position control scheme , and obtain the real-time state parameters of the patient after using the good limb position, compare the real-time state parameters with the standard good limb position state parameters, and determine whether the difference between the two exceeds the preset threshold. If the threshold is set, the area to be controlled and the control command of the mattress are adjusted according to the difference, so as to facilitate the control of the good limb position. Obtaining the real-time state parameters of the patient after using the good limb position includes the following steps: First, the state image of the patient is collected through the camera and the steering device, and the steering device can realize all-round collection, avoiding the inaccurate state parameters caused by incomplete photo collection. Happening. Secondly, build a good limb position monitoring model, and train the model to obtain a good limb position monitoring model that meets the requirements. The collected state image of the patient is input into a good limb position monitoring model that meets the requirements, and the current good limb position state parameters of the patient are determined.

Specifically, the method further includes: before the controller of the mattress controls the to-be-controlled area of the mattress according to the control instruction, the good limb position control platform obtains the first position of the skeleton key point at the first moment according to the human body key point prediction network, specifically Including: collecting the image data of the user before completing the good limb position control, inputting the image data to the human body key point prediction network, outputting the predicted probability of each pixel in the image being a bone key point, and forming a heat map, which is determined by the soft-argmax function. Convert the heat map to the key numerical coordinates of the skeleton and output it. The human key point prediction network is obtained by introducing the CBAM attention model based on the CPN feature pyramid network. The CPN feature pyramid network includes the GlobalNet model and the RefineNet model; The predicted position of the bone key point is obtained by a position and instantaneous velocity, and the first position and the predicted position are obtained by Q=AQ ¹ +BP+C to obtain the position trajectory of the bone key point; wherein, Q is the feature vector, used to represent the bone The state information of the key point at the second preset time, the state information of the bone key point includes the speed information of the key point of the bone and the coordinate information of the key point of the bone, A is the state transformation matrix, which is used to convert the key point of the bone at the first preset time. The state information of the point is related to the state of the skeleton key point at the second preset time, and the first preset time is the last time of the second preset time; according to the position trajectory of the skeleton key point, it is judged whether the user is using a good limb position correctly.

需要说明的是，人体关键点预测网络是在CPN特征金字塔网络的基础上引入了CBAM注意力模型得到的，其中，CPN特征金字塔网络包括整体检测骨骼关键点的GlobalNet模型和微调网络RefineNet模型，上述模型的结合，更好的发挥了CPN特征金字塔网络对深层和浅层的融合，也极大地发挥了注意力模型分配权重学习有用信息的优势，进一步提高了对人体骨骼关键点定位的准确性，可以确定出人体骨骼关键点的位置坐标，与通过人工定位得到的位置坐标相比更加精确。In an embodiment of the present specification, in an embodiment of the present specification, it is also possible to monitor whether the user uses the good limb position correctly by the following methods: before the controller of the mattress controls the area to be controlled of the mattress according to the control instruction, by The image collection device collects image data of the user at the first preset moment, and the image data may be pictures, photos, etc. The image collection device may be a rotatable camera set above the mattress to facilitate collection of panoramic images of the user. After the image data is collected, the image data is input into the human body key point prediction network, and the output image corresponds to the predicted probability of each pixel point being a bone key point to form a heat map. For example, the probability that a pixel is a bone key point is 99%. , the probability that the adjacent pixels are bone key points is 50%, and the heat map is converted into the key numerical coordinates of the bone through the soft-argmax function and output the position coordinates, where the soft-argmax function is

It should be noted that the human body key point prediction network is obtained by introducing the CBAM attention model on the basis of the CPN feature pyramid network. The CPN feature pyramid network includes the GlobalNet model for the overall detection of skeleton key points and the fine-tuning network RefineNet model. The combination of models makes better use of the fusion of deep and shallow layers of the CPN feature pyramid network, and also greatly exerts the advantages of the attention model assigning weights to learn useful information, and further improves the accuracy of positioning key points of human skeletons. The position coordinates of the key points of the human skeleton can be determined, which is more accurate than the position coordinates obtained by manual positioning.

After obtaining the first position coordinates at the first preset time, the predicted position of the corresponding bone key point is obtained according to the first position at the first preset time and the instantaneous speed of a certain bone key point, and the first position and the predicted position are calculated by The formula Q=AQ ¹ +BP+C obtains the position trajectory of the skeleton key point; wherein, Q is the feature vector, which is used to represent the state information of the skeleton key point at the second preset time, and the state information of the skeleton key point includes the skeleton key point. The speed information of the point and the coordinate information of the key point of the bone, A is the state transformation matrix, which is used to associate the state information of the key point of the bone at the first preset time with the state of the key point of the bone at the second preset time. According to the key point of the bone The position trajectory of the point determines whether the user is using a good limb position correctly. For example, the user starts to perform position change at time T1, time T2 is the time of the change process, and time T3 is the time when the position change is successful. At this time, the first preset time may be time T1, and the second preset time may be time T2. According to T1 The position coordinates of a certain skeleton key point and the instantaneous speed of the skeleton key point can obtain the position at time T2. Similarly, when the first preset time is at time T2 and the second preset time is at time T3, the time at T3 can be obtained. The position of the skeleton key point can further determine the movement trajectory of the skeleton key point, and combine the movement trajectories of multiple skeleton key points to determine whether the user uses the correct limb position correctly. For another example, taking the user from the supine position to the left lateral position as an example, the first preset moment may be the last moment of the supine position, and the second preset moment may be the first moment of the left lateral position. The position trajectory of the skeleton key point is determined at a preset time and the second preset time, and it is determined whether the movement trajectory is the movement trajectory of the skeleton key point from the supine position to the left lateral position. It should be noted that the collection of image data of the user before completing the control of the good limb position mentioned in the embodiments of this specification may refer to the process of changing from the first position to the second position when the user is in the first position before completing the control of the good limb position. image data at .

Using the above method to detect the user's good limb position improves the accuracy of the detection method, helps to timely detect the irregular use of the good limb position, and adjusts according to the user's use of the good limb position, helping the user to achieve the best possible Effect.

The embodiment of this description also provides an automatic limb position control device, as shown in FIG. 6 , comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores data that can be processed by the at least one processor. The instruction is executed by the at least one processor, so that the at least one processor can: the controller of the mattress sends the user's good limb position use request to the medical care node, so that the medical care node can call the user's case information , where the request includes the user's identity identifier and physical feature information; when the medical and nursing node confirms the request for using the good limb position, the good limb position control platform determines the user's good limb position control plan according to the user's case information and physical feature information, wherein , the good limb position control plan includes the air volume of the area to be controlled, the position information of the good limb position and the time corresponding to the position of the good limb position; the good limb position control platform determines the area to be controlled of the mattress according to the good limb position control scheme and A control instruction, wherein the control instruction includes the operation of inflating or deflating the area to be controlled, the inflation time or deflation time of the area to be controlled, and the amount of inflation in the area to be controlled; the good limb position control platform sends the good limb position control plan to the mattress controller It is convenient for the controller of the mattress to control the area to be controlled of the mattress according to the control instruction, so as to assist the user to complete the control of the good limb position.

In one or more embodiments of the present specification, the good limb position control scheme is determined according to the user's case information and body feature information, so that the use of good limb position is more unique, and the customized good limb position control scheme is further improved. The user experience, because the control plan is obtained based on the case information, the obtained plan is more scientific, which is helpful for the good limb position to play the role of helping the patient to carry out rehabilitation treatment; in addition, after obtaining the good limb position control plan, through the control The automatic control of the good limb position is carried out by means of instructions, which reduces the burden on the patient's family members and nurses.

Each embodiment in this specification is described in a progressive manner, and the same and similar parts between the various embodiments may be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the apparatus, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and reference may be made to some descriptions of the method embodiments for related parts.

The foregoing describes specific embodiments of the present specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that in the embodiments and still achieve desirable results. Additionally, the processes depicted in the figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

The above descriptions are merely one or more embodiments of the present specification, and are not intended to limit the present specification. Various modifications and variations of the one or more embodiments of this specification are possible for those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of this specification.

Claims (10)

1. an automatic good limb position control method, is characterized in that, described method comprises: The controller of the mattress sends the user's good limb position use request to the medical care node, so that the medical care node can call the user's case information, wherein the request includes the user identification and physical feature information; When the medical care node confirms the request for using the good limb position, the good limb position control platform determines the user's good limb position control scheme according to the user's case information and the physical feature information, wherein the The good limb position control scheme includes the air volume of the area to be controlled, the position information of the good limb position, and the time corresponding to the good limb position; The good limb position control platform determines the mattress to be controlled area and control instructions according to the good limb position control scheme, wherein the control instructions include the inflatable or deflated operation of the to-be-controlled area, the inflation time or the release of the to-be-controlled area. Inflation time and the inflatable volume of the area to be controlled; The good limb position control platform sends the good limb position control scheme to the controller of the mattress, so that the controller of the mattress can control the area to be controlled of the mattress according to the control instruction, so as to assist The user completes good limb position control. 2 . The automatic good limb position control method according to claim 1 , wherein before sending the good limb position control scheme to the controller of the mattress, the method further comprises: 2 . The area to be controlled of the mattress is set as a body placement area and a body adjacent area, so that the controller of the mattress controls the body adjacent area to execute the control instruction, and assists the user to complete the control of the good limb position; wherein the The area to be controlled includes: The left shoulder region, the right shoulder region, the left upper limb region, the right upper limb region, the left lower limb region, and the right lower limb region, each of which includes one or more of the body placement regions and at least one of the body adjacent regions. 3. A kind of automatic good limb position control method according to claim 1, is characterized in that, the described user's good limb position control scheme is determined according to described user's case information and described body characteristic information, specific include: Build an initial neural network model, collect medical institution case information, body feature information and relevant data of the good limb position control scheme based on the web crawler technology, and collect relevant data according to the medical institution case information, physical characteristic information and good limb position control scheme. , train the initial neural network model to obtain a good limb position control model that meets the requirements; Inputting the user's case information and the body feature information into the good limb position control model that meets the requirements, and determining the user's good limb position control scheme. 4. A kind of automatic good limb position control method according to claim 3, is characterized in that, described according to described medical institution case information, body characteristic information and the relevant data of good limb position control plan, training described initial Neural network model to obtain a good limb position control model that meets the requirements, including: Constructing a data set from multiple groups of medical institution case information, body feature information and data of a good limb position control plan, and dividing the data set into a training set and a verification set according to a preset ratio; A network structure for constructing a convolutional neural network model, wherein layers 0-95 of the convolutional neural network are deep convolutional layers, consisting of a convolutional block and a single-layer convolutional layer, and 96-95 of the convolutional neural network 126 layers are feature interaction layers; Pre-set training parameters, wherein the training parameters include: initial learning rate, weight decay rate and maximum number of iterations, input the medical institution case information, body feature information and relevant data of the good limb position control scheme in the training set into the volume Loop iterative training in the product neural network model; Input the medical institution case information, body feature information and good limb position control scheme in the verification set into the good limb position control model, record the loss function value in the training process of the model, and adjust the neural network model according to the loss function value When the loss function value reaches the preset loss function threshold or the number of training times reaches the preset maximum number of iterations, a good limb position control model that meets the requirements is obtained; wherein, the loss The function is the cross entropy loss function. 5. An automatic good limb position control method according to claim 1, wherein, according to the user's case information and the body feature information, the user's good limb position control scheme is determined, and the specific include: According to the user's body feature information, a first good limb position control scheme of the user is determined, wherein the body feature information includes: the user's height and the user's weight, and the first good limb position control scheme includes: : Place the initial position of the user and the inflation volume of the area to be controlled; The determining the first good limb position control scheme of the user according to the physical feature information of the user, specifically includes: Presetting the relationship between the body index and the inflation volume, wherein the body index is the ratio of the weight of the user to the square of the height of the user; When the body index is less than the first preset threshold, the inflation amount is the first inflation amount; When the body index is greater than the first preset threshold and less than a second preset threshold, the inflation amount is a second inflation amount; When the body index is greater than the second preset threshold, the inflation intensity is a third inflation amount; The first inflation amount is larger than the second inflation amount, and the second inflation amount is larger than the third inflation amount. 6 . The automatic good limb position control method according to claim 5 , wherein after the first good limb position control scheme of the user is determined according to the physical feature information of the user, the The method also includes: According to the case information of the user, the use type of the good limb position and the corresponding second good limb position control scheme are determined, wherein the second good limb position control scheme includes the position of the good limb position and the good limb position The time of placing the position, different usage types of the good limb position correspond to different positions of the good limb position and the time of placing the good limb position. 7 . The automatic good limb position control method according to claim 1 , wherein the determining the area to be controlled and the control instruction of the mattress according to the good limb position control scheme specifically comprises: 8 . According to the good limb position placement position in the good limb position control scheme, determine the area to be controlled and the mattress inflation or deflation operation in the control instruction, wherein the good limb position placement position includes: supine position, healthy side lying position and the lying position on the affected side; and determining the mattress inflation time according to the position change time in the good limb position control scheme. 8 . The automatic limb position control method according to claim 1 , wherein before the controller of the mattress controls the area to be controlled of the mattress according to the control instruction, the method further comprises: 9 . The controller of the mattress constructs a coordinate system with the mattress preset position as the origin, and records the first position coordinates of the first preset part of the user and the second position coordinates of the second preset part; After the assisting the user to complete the good limb position control, the method further includes: The controller of the mattress records the third position coordinate of the first preset part and the fourth position coordinate of the second preset part; According to the first position coordinates, the second position coordinates, the third position coordinates and the fourth position coordinates, it is determined whether the user is using the good limb position correctly, which specifically includes: According to the difference between the first position coordinate and the second position coordinate, the position of the user before completing the good limb position is determined, and the position status is used to reflect the user's body before the good limb position is completed. the location of each part; According to the difference between the third position coordinate and the fourth position coordinate, the position of the user after completing the healthy limb position is determined, and the position condition is used to reflect the body after the user completes the healthy limb position. the location of each part; According to the position of the user before completing the good limb position and the position of the user after completing the good limb position, it is determined whether the user moves according to the good limb position control scheme, and the user is judged Whether the current good limb position is in the correct position. 9. A kind of automatic good limb position control method according to claim 1, is characterized in that, described method also comprises: Before the controller of the mattress controls the to-be-controlled area of the mattress according to the control instruction, the good limb position control platform obtains the first position of the skeleton key point at the first preset time according to the human body key point prediction network. , including: Collect the image data of the user before completing the good limb position control, input the image data into the human body key point prediction network, output the predicted probability that each pixel point in the image is a bone key point, and form a heat map, which is composed of soft- The argmax function converts the heat map into the numerical coordinates of the skeleton key points and outputs, wherein the human body key point prediction network is obtained by introducing the CBAM attention model based on the CPN feature pyramid network, and the CPN feature pyramid network includes the GlobalNet model and RefineNet model; The predicted position of the skeleton key point is obtained according to the first position and the instantaneous speed at the first preset time, and the first position and the predicted position are obtained by Q=AQ ¹ +BP+C to obtain the skeleton key point The position trajectory of the point; wherein, the Q is a feature vector, which is used to represent the state information of the skeleton key point at the second preset moment, and the state information of the skeleton key point includes the speed information of the skeleton key point and Coordinate information of the skeleton key point, the A is a state transformation matrix, used to associate the state information of the skeleton key point at the first preset time with the skeleton key at the second preset time The state of the point, the first preset moment is the last moment of the second preset moment; It is judged whether the user is using a good limb position correctly according to the position trajectory of the skeleton key point. 10. An automatic good limb position control device, characterized in that, comprising: at least one processor; and, a memory communicatively coupled to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to: a controller of the mattress to sending a use request to a medical care node, so that the medical care node can call the user's case information, wherein the request includes user identification and physical feature information; When the medical care node confirms the request for using the good limb position, the good limb position control platform determines the user's good limb position control scheme according to the user's case information and the physical feature information, wherein the The good limb position control scheme includes the air volume of the area to be controlled, the position information of the good limb position, and the time corresponding to the good limb position; The good limb position control platform determines the mattress to be controlled area and control instructions according to the good limb position control scheme, wherein the control instructions include the inflatable or deflated operation of the to-be-controlled area, the inflation time or the release of the to-be-controlled area. Inflation time and the inflatable volume of the area to be controlled; The good limb position control platform sends the good limb position control scheme to the controller of the mattress, so that the controller of the mattress can control the area to be controlled of the mattress according to the control instruction, so as to assist The user completes good limb position control.

Images