Orthopedic nursing scheme intelligent recommendation method based on deep learning

Abstract

The invention relates to the technical field of nursing recommendation, in particular to an intelligent recommendation method for an orthopedic nursing scheme based on deep learning, and the method comprises the following steps: obtaining multi-dimensional data of an orthopedic patient; modeling the multi-dimensional data of the patient based on the time sequence model, and speculating the recovery progress of the orthopedic patient in the future time by analyzing the orthopedic disease change trend of the orthopedic patient in each recovery stage; a personalized nursing demand knowledge base is constructed, and the nursing demand knowledge base analyzes the nursing demand of the orthopedic patient in the recovery progress according to the speculated recovery progress in the future time; and according to the analyzed nursing demands of the orthopedic patient in the future, the nursing demands are integrated, and a personalized nursing scheme is generated. The method ensures that the nursing service can adapt to the dynamic demand of the patient, thereby effectively shortening the recovery time and improving the health recovery efficiency of the patient.

Description

Technical Field

[0001] The present invention relates to the technical field of nursing recommendation, and in particular to an intelligent recommendation method for orthopedic nursing plans based on deep learning. Background Art

[0002] The demand for treatment and care of fractures, joint injuries and other orthopedic-related diseases is increasing. The care process for orthopedic patients usually includes multiple aspects such as postoperative diet management, wound care, rehabilitation training and daily activity guidance. These aspects have a crucial impact on the quality and speed of patients' recovery. However, traditional orthopedic care plans are often based on experience and are difficult to dynamically adjust according to individual differences of patients.

[0003] Current orthopedic care plans are mostly standardized processes that ignore the differences in patients' recovery progress. The use of universal care plans can easily lead to patients not getting the best care results, affecting recovery efficiency. During the postoperative recovery process, the patient's health status changes dynamically, and care needs will also vary with the recovery stage. For example, care needs in the acute postoperative period are mainly focused on pain management and wound care, while the rehabilitation period focuses more on functional recovery and sports training. However, existing care plans make it difficult to monitor and analyze patients' recovery progress in real time, resulting in care plans lagging behind actual needs.

[0004] With the digitization of medical data, patients' multi-dimensional health data (such as medical imaging data, treatment records, and real-time monitoring data) can provide a rich basis for the formulation of nursing plans. However, traditional nursing plans fail to make full use of these data. The lack of intelligent analysis methods makes it difficult for nursing plans to meet the requirements of precision and scientificity. The nursing effect depends on the experience of nursing staff and lacks systematicness and foresight. Summary of the invention

[0005] The present invention provides an intelligent recommendation method for orthopedic nursing plans based on deep learning.

[0006] An intelligent recommendation method for orthopedic nursing plans based on deep learning includes the following steps:

[0007] S1, obtain multi-dimensional data of orthopedic patients, including fracture type, medical imaging data, and treatment history data;

[0008] S2, modeling the multi-dimensional data of patients based on a time series model, wherein the time series model adopts a Transformer model, and by analyzing the changing trend of orthopedic diseases of orthopedic patients at various recovery stages, the recovery progress of orthopedic patients in the future is inferred, and a dynamic health status map of orthopedic patients is constructed;

[0009] S3, constructing a personalized nursing demand knowledge base, wherein the nursing demand knowledge base analyzes the nursing demand of orthopedic patients at the recovery progress in the future according to the estimated recovery progress;

[0010] S4, based on the analyzed nursing needs of orthopedic patients in the future, integrates the nursing needs, generates personalized nursing plans, and recommends them to orthopedic patients and medical care staff.

[0011] Optionally, the S1 specifically includes:

[0012] S11, obtain the patient's fracture type data through the hospital information management system or electronic medical record system, including the patient's fracture site, fracture morphology, and time of fracture occurrence;

[0013] S12, obtaining medical imaging data of the patient through a medical imaging system, wherein the medical imaging data includes X-rays, CT scans, and MRI images, for analyzing the severity of orthopedic diseases and postoperative conditions;

[0014] S13, obtain the patient's treatment history data through treatment records, including the patient's surgical records, postoperative rehabilitation plan, drug treatment plan, and follow-up records.

[0015] Optionally, the S2 specifically includes:

[0016] S21, organize the multi-dimensional data in chronological order to form a time series data set, wherein the time series data set includes key data features of patients at different recovery stages, the recovery stages include postoperative acute stage, rehabilitation stage, and long-term recovery stage, and the key data features include:

[0017] Fracture type characteristics: including fracture site characteristics and fracture morphology characteristics;

[0018] Medical imaging data features: including X-rays, CT scans, and MRI image features;

[0019] Characteristics of treatment history data: including surgery type and surgery date;

[0020] S22, using the Transformer model to model the constructed time series data set. The Transformer model performs weighted processing on the historical data of orthopedic patients through the self-attention mechanism, captures the data dependency between different time nodes, and thus models the change trend of the patient's orthopedic disease, and infers the patient's recovery progress in the future time period;

[0021] S23, based on the Transformer model, constructs a dynamic health status map of orthopedic patients by analyzing the changing trends of orthopedic diseases in patients at various recovery stages. The dynamic health status map reflects the progression of orthopedic diseases in patients at various recovery stages, including fracture healing and joint function recovery.

[0022] Optionally, the time series data set is represented by X = {x1, x2, ..., x t}, where x t ∈R d Represents the multi-dimensional data features at time point t, T represents the length of the data sequence, and d represents the data feature dimension at each time point.

[0023] Optionally, the self-attention mechanism in the Transformer model specifically includes:

[0024] Input embedding: The input x at each time point t is mapped to the query Q t , key K t Sum value V t vector:

[0025] Q t =W q x t ,K t =W k x t ,V t =W v x t ,in, is the learned weight matrix, d k is the dimension of the embedding space;

[0026] Calculate attention weights: The self-attention mechanism calculates the query Q t With all keys K i The similarity of in, Represents the dot product of the query vector and the key vector. The higher the similarity, the higher the weight α ti The bigger;

[0027] Weighted sum: using attention weight α ti For all value vectors V i Perform weighted summation to obtain the output representation at each time point t Among them, z t ∈R dk is a weighted representation of each time point;

[0028] Feature extraction is performed through a multi-layer self-attention mechanism, and the patient’s recovery progress at time point t is finally outputted as ht :h t =LayerNorm(W o z t +b o ),in, and is the output weight matrix and bias term of the learning, d h is the dimension representing the health status, and LayerNorm represents layer normalization.

[0029] Optionally, based on the output of the Transformer model, a dynamic health status map of orthopedic patients is constructed, and the multidimensional features output by the Transformer model are aggregated and visualized. The dynamic health status map reflects the changing trends of orthopedic diseases of patients at various recovery stages.

[0030] Optionally, the S3 specifically includes:

[0031] S31, the recovery progress is represented as a continuous time series P t , P t ={p1,p2,...,p T}Where P t represents the patient's recovery status at time point t, and the length of the prediction time period is T;

[0032] S32, based on the estimated future recovery progress P t ,Build a nursing demand knowledge base, define the mapping relationship between different recovery progress and nursing needs, and the nursing demand category is represented by C t , represents the category of nursing needs required by the patient at time point t, including dietary needs, activity needs, medication management needs, and physical therapy needs.

[0033] Optionally, the nursing demand knowledge base also includes a classification model, which restores the progress P t As input, the output is a discrete care need category C t ,The goal of the classification model is to establish a mapping of recovery ,progress to care needs categories by learning historical data (including ,patients’ recovery progress and known care needs);

[0034] The classification model is based on a multi-layer perceptron, and the model structure includes:

[0035] Input layer: size is d, i.e., input feature dimension;

[0036] Hidden layer: set to h1, h2, use ReLU activation function;

[0037] Output layer: The size is n, which is the number of care demand categories.

[0038] Optionally, the calculation process of the classification model includes:

[0039] Input layer to the first hidden layer: z1 = W1·P t +b1, where is the weight matrix of the first layer, is the bias term, P t ∈R d is the recovery progress input vector, the activation function h1=ReLU(z1)=max(0,z1), the ReLU activation function applies a nonlinear transformation to each element;

[0040] From the first hidden layer to the second hidden layer: z2 = W2 h1 + b2, where is the weight matrix of the second layer, is the bias term, h1 is the activation output of the first layer, the activation function h2 = ReLU(z2), and the ReLU activation function is continued;

[0041] From the second hidden layer to the output layer: z3 = W3 h2 + b3, where is the weight matrix of the output layer, b3∈R n is the bias term, h2 is the activation output of the second layer;

[0042] The output layer uses softmax activation: Among them, the softmax function converts the output z3 into a probability distribution, which represents the probability of each care demand category:

[0043] in, represents the predicted probability of category i at time point t, n is the number of categories, z 3,i represents the original score obtained in the calculation process of the i-th nursing demand in the third layer (it can also be understood as the score calculated by the input feature or model), z 3,j It represents the original score obtained in the calculation process of the j-th nursing need at the third level, and compares the relative importance of different categories. exp represents the exponential function, which performs an exponential transformation on the original score.

[0044] Beneficial effects of the present invention:

[0045] The present invention, by combining the patient's recovery progress prediction and analyzing the patient's multi-dimensional data, can comprehensively capture the change trend of the patient's orthopedic disease at each recovery stage and generate dynamically adjusted nursing needs. Based on the patient's specific recovery progress and nursing knowledge base, the nursing needs are effectively combined with the recovery stage to form a personalized nursing plan, which improves the accuracy of nursing and ensures that the nursing service can adapt to the patient's dynamic needs, thereby effectively shortening the recovery time and improving the patient's health recovery efficiency.

[0046] The present invention innovatively solves the problem of adapting nursing needs for different patients and different recovery stages by establishing a nursing needs knowledge base and combining nursing needs with a classification model. By analyzing historical data and real-time data, it can automatically generate a nursing needs combination with priority sorting, thus realizing a scientific combination of nursing needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the present invention or 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 for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0048] Figure 1 A schematic diagram of a method flow chart of an embodiment of the present invention;

[0049] Figure 2 Schematic diagram of the classification model structure of an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0051] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0052] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0053] like Figure 1-Figure 2As shown, a method for intelligently recommending orthopedic care plans based on deep learning includes the following steps:

[0054] S1, obtain multi-dimensional data of orthopedic patients, including fracture type, medical imaging data, and treatment history data;

[0055] S2, based on the time series model, the multi-dimensional data of patients is modeled. The time series model uses the Transformer model. By analyzing the change trend of orthopedic diseases in orthopedic patients at various recovery stages (acute phase after surgery, rehabilitation phase, and long-term recovery phase), the recovery progress of orthopedic patients in the future is inferred, and a dynamic health status map of orthopedic patients is constructed;

[0056] S3, build a personalized nursing demand knowledge base, which analyzes the nursing needs of orthopedic patients at the recovery progress in the future according to the estimated recovery progress;

[0057] S4, based on the analyzed nursing needs of orthopedic patients in the future, integrates the nursing needs, generates personalized nursing plans, and recommends them to orthopedic patients and medical care staff.

[0058] S1 specifically includes:

[0059] S11, obtain the patient's fracture type data through the hospital information management system or electronic medical record system, including the patient's fracture site, fracture morphology, and time of fracture occurrence;

[0060] S12, obtaining medical imaging data of the patient through a medical imaging system, the medical imaging data including X-rays, CT scans, and MRI images, for analyzing the severity of orthopedic diseases and postoperative conditions;

[0061] S13, obtain the patient's treatment history data through treatment records, including the patient's surgical records, postoperative rehabilitation plan, drug treatment plan, and follow-up records.

[0062] S2 specifically includes:

[0063] S21, organize the multi-dimensional data in chronological order to form a time series data set. The time series data set includes key data features of patients in different recovery stages. The recovery stages include the postoperative acute stage, rehabilitation stage, and long-term recovery stage. The key data features include:

[0064] Fracture type characteristics: including fracture site characteristics and fracture morphology characteristics;

[0065] Medical imaging data features: including X-ray, CT scan and MRI image features (fracture alignment, joint recovery, bone density);

[0066] Characteristics of treatment history data: including surgery type and surgery date;

[0067] S22, the Transformer model is used to model the constructed time series dataset. The Transformer model performs weighted processing on the historical data of orthopedic patients through the self-attention mechanism, capturing the data dependency between different time nodes, thereby modeling the change trend of the patient's orthopedic disease and inferring the patient's recovery progress in the future time period;

[0068] S23, based on the Transformer model, constructs a dynamic health status map of orthopedic patients by analyzing the changing trends of orthopedic diseases in patients at various recovery stages. The dynamic health status map reflects the progression of orthopedic diseases in patients at various recovery stages, including fracture healing and joint function recovery.

[0069] The time series data set is represented as X = {x1, x2, ..., x t}, where x t ∈R d Represents the multi-dimensional data features at time point t, T represents the length of the data sequence, and d represents the data feature dimension at each time point.

[0070] The self-attention mechanism in the Transformer model specifically includes:

[0071] Input embedding: The input x at each time point t is mapped to the query Q t , key K t Sum value V t vector:

[0072] Q t =W q x t ,K t =W k x t ,V t =W v x t ,in, is the learned weight matrix, d k is the dimension of the embedding space;

[0073] Calculate attention weights: The self-attention mechanism calculates the query Q t With all keys K i The similarity of in, Represents the dot product of the query vector and the key vector. The higher the similarity, the higher the weight α ti The bigger;

[0074] Weighted sum: using attention weight α ti For all value vectors V i Perform weighted summation to obtain the output representation at each time point t Among them, z t ∈R dk is a weighted representation of each time point;

[0075] Feature extraction is performed through a multi-layer self-attention mechanism, and the patient’s recovery progress at time point t is finally outputted as h t :h t =LayerNorm(W o z t +b o ),in, and is the output weight matrix and bias term of the learning, d h is the dimension representing the health status, LayerNorm represents layer normalization;

[0076] Assume that the patient's recovery progress in the future time period is Where N is the number of time steps for prediction. The trained model is combined with the patient’s historical data h1,h2,…,h T , predict the health status of the next N time steps Among them, f(·) is the prediction function obtained through learning, and θ represents the parameters of the model.

[0077] Based on the output of the Transformer model, a dynamic health status map of orthopedic patients is constructed. The multi-dimensional features output by the Transformer model are aggregated and visualized. The dynamic health status map reflects the changing trends of orthopedic diseases in patients at various recovery stages. For example:

[0078] Fracture healing: including the healing progress of the fracture site, the alignment of the fracture line, and the stability of the fracture segment;

[0079] Recovery of joint function: including joint range of motion, muscle strength, and gait recovery.

[0080] S3 specifically includes:

[0081] S31, the recovery progress is represented as a continuous time series P t , P t ={p1,p2,...,p T}Where P t represents the patient's recovery status at time point t, and the length of the prediction time period is T;

[0082] S32, based on the estimated future recovery progress P t,Build a nursing demand knowledge base, define the mapping relationship between different recovery progress and nursing needs, and the nursing demand category is represented by C t , represents the category of care needs required by the patient at time point t, including dietary needs, activity needs, medication management needs, and physical therapy needs.

[0083] The nursing needs knowledge base also includes a classification model, which will restore the progress P t As input, the output is a discrete care need category C t ,The goal of the classification model is to establish a mapping of recovery ,progress to care needs categories by learning historical data (including ,patients’ recovery progress and known care needs);

[0084] Before model training, the recovery progress P t and the corresponding care needs category C t Prepared in a format suitable for classification, there are N sample data, each sample is (P t ,C t ), recovery progress P t is a multidimensional time series feature vector. In order to input it into the classification model, P t Flatten or aggregate (max pooling) it to obtain a fixed-length feature vector, the nursing need category C t is a discrete category label represented by onehot encoding. For example, if there are five categories of care needs, C t is C2 (i.e., physical therapy), then its corresponding onehot encoding is [0,1,0,0,0];

[0085] The classification model is based on a multi-layer perceptron, and the model structure includes:

[0086] Input layer: size is d, i.e., input feature dimension;

[0087] Hidden layer: set to h1, h2, use ReLU activation function;

[0088] Output layer: The size is n, which is the number of care demand categories.

[0089] The calculation process based on the multi-layer perceptron classification model includes:

[0090] Input layer to the first hidden layer: z1 = W1·P t +b1, where is the weight matrix of the first layer, is the bias term, P t ∈R dis the recovery progress input vector, the activation function h1=ReLU(z1)=max(0,z1), the ReLU activation function applies a nonlinear transformation to each element;

[0091] From the first hidden layer to the second hidden layer: z2 = W2 h1 + b2, where is the weight matrix of the second layer, b2∈R h2 is the bias term, h1 is the activation output of the first layer, the activation function h2 = ReLU(z2), and the ReLU activation function is continued;

[0092] From the second hidden layer to the output layer: z3 = W3 h2 + b3, where is the weight matrix of the output layer, b3∈R n is the bias term, h2 is the activation output of the second layer;

[0093] The output layer uses softmax activation: Among them, the softmax function converts the output z3 into a probability distribution, which represents the probability of each care demand category:

[0094] in, represents the predicted probability of category i at time point t, n is the number of categories, z 3,i represents the original score obtained in the calculation process of the i-th nursing demand in the third layer (it can also be understood as the score calculated by the input feature or model), z 3,j It represents the original score obtained in the calculation process of the j-th nursing need at the third level, and compares the relative importance of different categories. exp represents the exponential function, which performs an exponential transformation on the original score.

[0095] Use the cross entropy loss function to train the model: Among them, y i is the true label (onehot encoding), is the predicted class probability.

[0096] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0097] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for intelligent recommendation of orthopedic nursing plans based on deep learning, characterized in that: The following steps are involved: S1, obtain multi-dimensional data of orthopedic patients, including fracture type, medical imaging data, and treatment history data; S2, modeling the multi-dimensional data of patients based on a time series model, wherein the time series model adopts a Transformer model, and by analyzing the changing trend of orthopedic diseases of orthopedic patients at various recovery stages, the recovery progress of orthopedic patients in the future is inferred, and a dynamic health status map of orthopedic patients is constructed; S3, constructing a personalized nursing demand knowledge base, wherein the nursing demand knowledge base analyzes the nursing demand of orthopedic patients at the recovery progress in the future according to the estimated recovery progress; S4, based on the analyzed nursing needs of orthopedic patients in the future, integrates the nursing needs, generates personalized nursing plans, and recommends them to orthopedic patients and medical care staff.

2. According to claim 1, a method for intelligently recommending orthopedic nursing plans based on deep learning is characterized in that: The S1 specifically includes: S11, obtain the patient's fracture type data through the hospital information management system or electronic medical record system, including the patient's fracture site, fracture morphology, and time of fracture occurrence; S12, obtaining medical imaging data of the patient through a medical imaging system, wherein the medical imaging data includes X-rays, CT scans, and MRI images, for analyzing the severity of orthopedic diseases and postoperative conditions; S13, obtain the patient's treatment history data through treatment records, including the patient's surgical records, postoperative rehabilitation plan, drug treatment plan, and follow-up records.

3. The method for intelligently recommending orthopedic nursing care plans based on deep learning according to claim 1, characterized in that: The S2 specifically includes: S21, organize the multi-dimensional data in chronological order to form a time series data set, wherein the time series data set includes key data features of patients at different recovery stages, the recovery stages include postoperative acute stage, rehabilitation stage, and long-term recovery stage, and the key data features include: Fracture type characteristics: including fracture site characteristics and fracture morphology characteristics; Medical imaging data features: including X-ray, CT scan and MRI image features; Treatment history data characteristics: including surgery type and surgery date; S22, using the Transformer model to model the constructed time series data set. The Transformer model performs weighted processing on the historical data of orthopedic patients through the self-attention mechanism, captures the data dependency between different time nodes, and thus models the change trend of the patient's orthopedic disease, and infers the patient's recovery progress in the future time period; S23, based on the Transformer model, constructs a dynamic health status map of orthopedic patients by analyzing the changing trends of orthopedic diseases in patients at various recovery stages. The dynamic health status map reflects the progression of orthopedic diseases in patients at various recovery stages, including fracture healing and joint function recovery.

4. The method for intelligently recommending orthopedic nursing care plans based on deep learning according to claim 3, characterized in that: The time series data set is represented as X={x1,x2,...,x t }, where x t ∈R d Represents the multi-dimensional data features at time point t, T represents the length of the data sequence, and d represents the data feature dimension at each time point.

5. The method for intelligently recommending orthopedic nursing care plans based on deep learning according to claim 4, characterized in that: The self-attention mechanism in the Transformer model specifically includes: Input embedding: The input x at each time point t is mapped to the query Q t , key K t Sum value V t vector: Q t =W q x t ,K t =W k x t ,V t =W v x t ,in, is the learned weight matrix, d k is the dimension of the embedding space; Calculate attention weights: The self-attention mechanism calculates the query Q t With all keys K i The similarity of ti : in, Represents the dot product of the query vector and the key vector. The higher the similarity, the higher the weight α ti The bigger; Weighted sum: using attention weight α ti For all value vectors V i Perform weighted summation to obtain the output representation at each time point t in, is a weighted representation of each time point; Feature extraction is performed through a multi-layer self-attention mechanism, and the patient’s recovery progress at time point t is finally outputted as h t :h t =LayerNorm(W o z t +b o ),in, and is the output weight matrix and bias term of the learning, d h is the dimension representing the health status, and LayerNorm represents layer normalization.

6. The method for intelligently recommending orthopedic nursing care plans based on deep learning according to claim 5, characterized in that: Based on the output of the Transformer model, a dynamic health status map of orthopedic patients is constructed, and the multidimensional features output by the Transformer model are aggregated and visualized. The dynamic health status map reflects the changing trends of orthopedic diseases in patients at various recovery stages.

7. The method for intelligently recommending orthopedic nursing care plans based on deep learning according to claim 1, characterized in that: The S3 specifically includes: S31, the recovery progress is represented as a continuous time series P t , where P t represents the patient's recovery status at time point t, and the length of the prediction time period is T; S32, based on the estimated future recovery progress P t ,Build a nursing demand knowledge base, define the mapping relationship between different recovery progress and nursing needs, and the nursing demand category is represented by C t , represents the category of nursing needs required by the patient at time point t, including dietary needs, activity needs, medication management needs, and physical therapy needs.

8. The method for intelligently recommending orthopedic nursing care plans based on deep learning according to claim 7, characterized in that: The nursing demand knowledge base also includes a classification model, which restores the progress P t As input, the output is a discrete care need category C t ,The goal of the classification model is to establish a mapping of ,recovery progress to care need categories by learning from historical ,data; The classification model is based on a multi-layer perceptron, and the model structure includes: Input layer: size is d, i.e., input feature dimension; Hidden layer: set to h1, h2, use ReLU activation function; Output layer: The size is n, which is the number of care demand categories.

9. The method for intelligently recommending orthopedic nursing care plans based on deep learning according to claim 8, characterized in that: The calculation process of the classification model includes: Input layer to the first hidden layer: z1 = W1·P t +b1, where is the weight matrix of the first layer, is the bias term, P t ∈R d is the recovery progress input vector, the activation function h1=ReLU(z1)=max(0,z1), the ReLU activation function applies a nonlinear transformation to each element; From the first hidden layer to the second hidden layer: z2 = W2 h1 + b2, where is the weight matrix of the second layer, is the bias term, h1 is the activation output of the first layer, and the activation function h2 = ReLU(z2); From the second hidden layer to the output layer: z3 = W3 h2 + b3, where is the weight matrix of the output layer, b3∈R n is the bias term, h2 is the activation output of the second layer; The output layer uses softmax activation: Among them, the softmax function converts the output z3 into a probability distribution, which represents the probability of each care demand category: in, represents the predicted probability of category i at time point t, n is the number of categories, z 3,i represents the original score obtained in the calculation process of the i-th nursing need at the third level, z 3,j It represents the original score obtained in the calculation process of the j-th nursing demand in the third layer, and exp represents the exponential function, which performs exponential transformation on the original score.

Images