Nursing monitoring method and device

Abstract

In the technical scheme provided by the embodiment of the invention, the nursing monitoring method and the nursing monitoring device are provided, the medical time sequence data are compared and monitored in real time according to the preset standard monitoring data, and the comparison result is synchronized to the nursing application end to realize direct monitoring of a patient; in addition, a plurality of medical time sequence data can form a medical data sequence, and feature extraction and feature classification are performed on the medical data sequence through a risk identification model to obtain potential risk information about the patient. According to the invention, automatic monitoring of the physiological information of the patient can be realized, the potential risk can be identified, and the all-around monitoring efficiency and integrity of the patient in the hospitalization nursing process, especially in postoperative and severe nursing are improved.

Description

Technical Field

[0001] This application relates to the technical field of medical care, and specifically provides a nursing monitoring method and device. Background Art

[0002] For postoperative patients, especially those undergoing major surgeries, their treatment requires not only considering the preoperative assessment of the surgery and the formulation of the surgical plan, but also the safety and stability of the intraoperative surgical operation, as well as the postoperative care of the patients. Compared with the inpatient care of ordinary patients, due to the physical condition of postoperative patients and the possibility of secondary diseases, their in-hospital care needs to be more timely and targeted. Currently, the care for postoperative patients mainly relies on nursing staff to conduct real-time inspections based on the physiological monitoring data of the patients at regular intervals, and to judge by asking about the physical condition of the patients. This method depends on the professional knowledge of the nursing staff and the high-frequency inquiries and inspections of the nursing staff, and cannot achieve real-time monitoring or in-depth data interpretation, making it impossible to judge potential high-risk information. Summary of the Invention

[0003] To solve the above technical problems, the embodiments of this application provide a nursing monitoring method and device, which can process the real-time medical sensing data of patients to obtain the direct nursing risks and potential nursing risks of patients, improve the integrity and accuracy of obtaining patient care information, and improve the survival rate of postoperative patients.

[0004] To achieve the above object, the technical solutions adopted in the embodiments of this application are as follows:

[0005] In a first aspect, a postoperative patient nursing monitoring method is provided. The method is applied to a monitoring system, and the monitoring system includes a medical data collection device, a server, and a nursing application terminal. The medical data collection device communicates with the nursing application terminal. The medical data collection device is used to collect the medical time-series data of patients and transmit the medical time-series data to the server. The method is applied to the server and includes: receiving the medical time-series data, comparing the pre-set standard monitoring data with the medical time-series data to determine whether it exceeds the monitoring range, and if it exceeds the monitoring range, sending an alarm reminder to the nursing application terminal; if it does not exceed the monitoring range, collecting a plurality of the medical time-series data within a standard time period to form a medical data sequence, inputting the medical data sequence into a pre-set risk identification model, processing the medical data sequence based on the risk identification model to obtain risk information, and sending the risk information to the nursing application terminal.

[0006] Further, the medical time-series data includes first-type medical time-series data and second-type medical time-series data, and the standard monitoring data includes first-type standard monitoring data and second-type standard monitoring data; determining whether it exceeds the monitoring range by comparing the preset standard monitoring data with the medical time-series data includes: comparing the preset first-type standard monitoring data and the second-type standard monitoring data with the first-type medical time-series data and the second-type medical time-series data respectively to determine whether it exceeds the monitoring range; the step of sending an alarm reminder to the nursing application end if it exceeds the monitoring range includes: sending an alarm reminder to the nursing application end when the second-type medical time-series data exceeds the monitoring range; when the second-type medical time-series data does not exceed the monitoring range, determining whether the first-type medical time-series data exceeds the monitoring range and the corresponding quantity of the exceeded monitoring range, and when the quantity of the first-type medical time-series data exceeding the monitoring range is greater than or equal to 2, sending an alarm reminder to the nursing application end.

[0007] Further, the method further includes: when the quantity of the first-type medical time-series data exceeding the monitoring range is less than 2, obtaining the generation frequency of this first-type medical time-series data within the standard time period, and when the generation frequency is greater than 30%, sending an alarm reminder to the nursing application end.

[0008] Further, the risk identification model includes a feature extraction module and a classifier module. The feature extraction module is used to extract features from the medical data sequence and obtain a feature vector, and the classifier module is used to receive the feature vector and determine the classification result.

[0009] Further, the feature extraction module includes a first feature extraction module and a second feature extraction module constructed based on a convolutional network. The first feature extraction module is constructed based on a convolutional neural network, and the second feature extraction module is constructed based on a recurrent neural network. The second feature extraction module is a bidirectional long short-term memory network optimized by a trigonometric function pedigree arithmetic optimization algorithm. The feature extraction module is obtained after training the convolutional neural network and the optimized bidirectional long short-term memory network; the first feature extraction module is used to capture the spatial relationship of the medical data sequence, and the spatial relationship includes a trend relationship, a periodic relationship, and a proximity relationship. The second feature extraction module is used to capture and fuse the spatial relationship and then extract features.

[0010] Further, the second feature extraction module includes an input layer, a backward layer, a forward layer, and an output layer. Control units are provided in the backward layer and the forward layer. The control unit includes an input gate, a forget gate, and an output gate. The control unit is used to control the new information added to the control unit and determine the updated information. The forget gate determines the information to be discarded from the control unit. The output gate determines the output information based on the current state of the control unit.

[0011] Further, the classifier module receives the feature vector, processes it through a sigmoid function to obtain the type probability corresponding to the feature vector, and corresponds the type probability to the label category probability. When the type probability is greater than the label category probability, the feature vector belongs to this label category.

[0012] Further, optimizing the bidirectional long short-term memory network based on the trigonometric function pedigree arithmetic optimization algorithm includes the following steps: generating the algorithm and the initial parameters of the bidirectional long short-term memory network, calculating the fitness value of each individual, comparing the fitness values of each agent and determining the best fitness value of the individual, updating the optimal position of the agent, determining whether the convergence condition is satisfied and performing iterative updates until the convergence condition is satisfied, and obtaining the bidirectional long short-term memory network with the minimum mean square error.

[0013] Further, training the convolutional neural network and the optimized bidirectional long short-term memory network includes the following steps: constructing the read normalized medical data sequence data to obtain a feature matrix containing the temporal correlation and spatial correlation of the medical data sequence; using the time series container row model to build. After the model network layer structure is built, determine the parameter values such as the convolution kernel and stride in the model; input the feature matrix into the convolutional neural network, and use the convolutional neural network to effectively extract the spatial features of the medical data sequence data to obtain the feature vector of the medical data sequence time series; input the feature vector extracted by the convolutional neural network into the optimized bidirectional long short-term memory network, and use the optimized bidirectional long short-term memory network to extract the long-term time dependence features of the medical data sequence; train and predict the model from the forward network and the backward network according to the initial parameters, and output the corresponding loss curve and prediction result; output the predicted medical data sequence time series value, and calculate the prediction error value until the error value converges to the minimum.

[0014] Second aspect, a postoperative patient care and monitoring device is provided. The device is applied to a monitoring system, which includes a medical data acquisition device, a server, and a nursing application terminal. The medical data acquisition device communicates with the nursing application terminal. The medical data acquisition device is used to collect the medical time-series data of the patient and transmit the medical time-series data to the server. The device is applied to the server and includes: an initial warning module, configured to receive the medical time-series data, compare the medical time-series data with pre-set standard monitoring data to determine whether it exceeds the monitoring range, and send an alarm reminder to the nursing application terminal if it exceeds the monitoring range; a risk identification module, when the medical time-series data does not exceed the monitoring range, configured to collect a plurality of the medical time-series data within a standard time period to form a medical data sequence, input the medical data sequence into a pre-set risk identification model, process the medical data sequence based on the risk identification model to obtain risk information, and send the risk information to the nursing application terminal.

[0015] Third aspect, a terminal device is provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the postoperative patient care and monitoring method described in any one of the above is implemented.

[0016] Fourth aspect, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the postoperative patient care and monitoring method described in any one of the above is implemented.

[0017] In the technical solution provided by the embodiments of the present application, a care and monitoring method and device are provided. By comparing and monitoring the medical time-series data in real time according to the pre-set standard monitoring data, and synchronizing the comparison result to the nursing application terminal, direct monitoring of the patient is achieved. And the present application can also form a medical data sequence from a plurality of the medical time-series data, and perform feature extraction and feature classification on the medical data sequence through a risk identification model to obtain potential risk information about the patient. The present application can realize automatic monitoring of the patient's physiological information, identify potential risks, and improve the all-round monitoring efficiency and integrity of patients during the hospitalization care process, especially in postoperative and critical care. Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0019] The methods, systems, and / or programs in the accompanying drawings will be further described according to exemplary embodiments. These exemplary embodiments will be described in detail with reference to the drawings. These exemplary embodiments are non-limiting exemplary embodiments, where example numbers represent similar mechanisms in the various views of the drawings.

[0020] Figure 1 It is a schematic structural diagram of a monitoring system provided by an embodiment of the present application.

[0021] Figure 2 It is a schematic flowchart of a postoperative patient care and monitoring method provided by an embodiment of the present application.

[0022] Figure 3 It is a schematic structural diagram of a postoperative patient care and monitoring device provided by an embodiment of the present application.

[0023] Figure 4 It is a schematic structural diagram of a care and monitoring device provided by an embodiment of the present application. Detailed implementation manners

[0024] To better understand the above technical solutions, the technical solutions of the present application will be described in detail below through the drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solutions of the present application, rather than limitations on the technical solutions of the present application. Without conflict, the technical features in the embodiments of the present application and the embodiments can be combined with each other.

[0025] In the following detailed description, many specific details are set forth by way of example in order to provide a thorough understanding of the relevant teachings. However, it will be apparent to those skilled in the art that the present application may be practiced without these details. In other instances, well-known methods, procedures, systems, components, and / or circuits have been described at a relatively high level without detail in order to avoid unnecessarily obscuring aspects of the present application.

[0026] Flowcharts are used in the present application to illustrate the execution processes performed by the systems according to the embodiments of the present application. It should be clearly understood that the execution processes of the flowcharts may not be executed in sequence. On the contrary, these execution processes may be executed in reverse order or simultaneously. Additionally, at least one other execution process may be added to the flowchart. One or more execution processes may be deleted from the flowchart.

[0027] Before further elaborating on the embodiments of the present invention, the nouns and terms involved in the embodiments of the present invention are described. The nouns and terms involved in the embodiments of the present invention are applicable to the following explanations.

[0028] (1) Responsive to, which is used to represent the conditions or states on which the executed operations depend. When the dependent conditions or states are met, one or more of the executed operations can be real-time or can have a set delay; without special instructions, there is no limit on the execution order of multiple executed operations.

[0029] (2) Based on, which is used to represent the conditions or states on which the executed operations depend. When the dependent conditions or states are met, one or more of the executed operations can be real-time or can have a set delay; without special instructions, there is no limit on the execution order of multiple executed operations.

[0030] The embodiments of the present application provide a postoperative patient care and monitoring method, which is used to monitor the physical characteristics of patients after surgery, and perform early warning processing based on real-time physical characteristic data to determine whether the patient has abnormal conditions and potential risk information, so as to realize automated nursing information judgment.

[0031] Similarly, this solution can also be used for in-hospital care of ICU patients and critically ill patients. For postoperative patients, ICU patients, and critically ill patients, there is always a risk of deterioration or even death at any time. Therefore, it becomes particularly important to monitor, analyze, and predict the health status of the above patients in real time. Traditional patient medical prediction methods based on statistics mainly analyze the health status of patients by constructing a physiological scoring system. This method has great limitations, cannot effectively focus on the changes in the local state of physiological indicators, and has problems such as prediction delay, low accuracy, and little clinical reference value. And the above patient medical prediction problem has a higher real-time requirement compared to ordinary medical predictions. Currently, in the prior art, the learning and analysis of medical prediction mainly rely on a large number of time series indicators, including heart rate, blood pressure, body temperature, etc. However, there is a serious imbalance in the proportion of sample data of various categories in existing medical data sets. Therefore, how to balance the sample data and improve the quality of the data set has become another major problem in medical prediction.

[0032] Based on this, the postoperative patient care and monitoring method provided by the embodiments of the present application can solve the above prior art problems. Refer to Figure 1, wherein the postoperative patient care and monitoring method of the present application is applied to the monitoring system 100, and the monitoring system 100 includes a medical data acquisition device 120 for collecting medical data of patients, and also includes a server 110 and a nursing application terminal 130. The server communicates with the medical data acquisition device and the nursing application terminal. The medical data acquisition device uploads the collected medical data to the server. The server is configured with the postoperative patient care and monitoring method in the embodiments of the present application to process the collected medical data, and sends the processing result to the nursing application terminal. The nursing application terminal is configured based on a mobile device, which can be any one of a PDA and various mobile terminals, and can also include a computer terminal in the nurse station, which will not be elaborated in the embodiments of the present application.

[0033] Among them, in order to illustrate how the monitoring method operates, please refer to Figure 2 , for the postoperative patient care and monitoring method in the embodiments of the present application, it specifically includes the following steps:

[0034] Step S210. Receive the medical time-series data, and compare it with the pre-set standard monitoring data to determine whether it exceeds the monitoring range. If it exceeds the monitoring range, send an alarm reminder to the nursing application terminal.

[0035] In the embodiments of the present application, there are two ways for data early warning in the monitoring method. The first is direct early warning, and the early warning logic is based on whether there is physiological data exceeding the early warning range in the real-time data. The second is indirect early warning, and the indirect early warning logic is based on extracting and classifying data features from the real-time data after excluding direct early warning to determine whether there is a risk and the type of risk.

[0036] Therefore, in actual operation, for early warning, it is first necessary to determine whether there is data with direct risk. Among them, for the collected medical data, the corresponding characterization information is different due to different types. For example, the information meanings represented by body temperature data and electrocardiogram data are completely different, especially the changes of the above two data and the information represented by the changes are also completely different. Therefore, if all the collected data are used as monitoring data or the same early warning weights and attentions are configured for all the collected data, it will cause waste of nursing resources. Therefore, in the embodiments of the present application, the collected medical real-time data is divided into the first type of medical time-series data and the second type of medical time-series data according to its physiological characterization weight. Correspondingly, the first type of standard monitoring data and the second type of standard monitoring data are also set for the standard monitoring data for comparing the above two types of data. It can be understood that. The first type of medical real-time data is basic data, including but not limited to data such as body temperature and pulse; the second type of medical real-time data is specific data.

[0037] Among them, for two different types of medical real-time data, because of their different natures and corresponding characterization information, the comparison methods for monitoring need to be further subdivided.

[0038] Specifically, the comparison of whether it exceeds the monitoring range is specifically determined by comparing the pre-set first-type standard monitoring data and the second-type standard monitoring data with the first-type medical time-series data and the second-type medical time-series data respectively to determine whether it exceeds the monitoring range.

[0039] Among them, when the second-type medical time-series data exceeds the monitoring range, an alarm reminder is sent to the nursing application end; when the second-type medical time-series data does not exceed the monitoring range, it is determined whether the first-type medical time-series data exceeds the monitoring range and the corresponding quantity of the exceeded monitoring range. When the quantity of the first-type medical time-series data that exceeds the monitoring range is greater than or equal to 2, an alarm reminder is sent to the nursing application end.

[0040] In order to further reduce the cost of the monitoring process, when the quantity of the first-type medical time-series data that exceeds the monitoring range is less than 2, the generation frequency of this first-type medical time-series data within the standard standard time period is obtained. When the generation frequency is greater than 30%, an alarm reminder is sent to the nursing application end.

[0041] It can be seen from the above processing process that in the embodiment of the present application, for direct early warning, not any data collected is directly used as the data for direct comparison for judgment, which reduces the nursing cost and improves the pertinence of nursing. Among them, the classification of the first-type medical real-time data and the second-type medical real-time data is set based on the patient's situation and the corresponding disease situation in different cases, which will not be elaborated in the embodiment of the present application.

[0042] Step S220. If it does not exceed the monitoring range, a plurality of the medical time-series data within the collection time period of the standard time period are collected to form a medical data sequence, and the medical data sequence is input into a pre-set risk identification model. Based on the risk identification model, the medical data sequence is processed to obtain risk information, and the risk information is sent to the nursing application end.

[0043] In the embodiment of the present application, step S210 is mainly for direct risk judgment. In the actual process, doctors and medical staff need to pay more attention to the potential physical conditions of patients, and the potential physical conditions are more important in postoperative patients and critically ill patients. By obtaining potential data, the current physical state of the patient can be more determined. Especially in the ICU, the potential data can show the survival ability of the patient. Through the mining of this data, it can assist medical staff in corresponding nursing and treatment.

[0044] Among them, this process is implemented based on a risk identification model, and this risk identification model is constructed in a machine learning manner. This structure includes two parts. The first part is a feature extraction module, and the second part is a classifier module. The feature extraction module is used to extract features from the medical data sequence and obtain a feature vector, and the classifier module is used to receive the feature vector and determine the classification result. That is, first, the feature extraction module extracts features from the medical data sequence, and the classifier module determines the probabilities corresponding to different classifications based on the extracted feature vector, and sends this probability and the corresponding classification result to the nursing application terminal. Medical staff can make a secondary manual judgment based on the medical data sequence and the corresponding classification result. Through this process, the manual judgment cost and the situation of manual judgment errors when medical staff directly process the medical data sequence are reduced.

[0045] Specifically, the feature extraction module in the embodiment of the present application includes a first feature extraction module and a second feature extraction module constructed based on a convolutional network. The first feature extraction module is constructed based on a convolutional neural network, and the second feature extraction module is constructed based on a recurrent neural network. The second feature extraction module is an optimized bidirectional long short-term memory network based on a trigonometric function pedigree arithmetic optimization algorithm. The feature extraction module is obtained by training the convolutional neural network and the optimized bidirectional long short-term memory network.

[0046] Among them, in the embodiment of the present application, the first feature extraction module is constructed using a convolutional neural network. A convolutional neural network is a network with a three-dimensional volume, and its hidden layer architecture mainly consists of one or more convolutional layers, pooling layers, fully connected layers, and output layers. Specifically, it includes a first convolutional layer, a first pooling layer, a second convolutional layer, a second pooling layer, a fully connected layer, and an output layer.

[0047] In the embodiment of the present application, when inputting the medical data sequence into the first feature extraction module, it also includes constructing features of the medical data sequence. The constructed features include trend features, periodic construction, and proximity construction. Among them, the construction is implemented using the time series container Sequential in the prior art. Through the above construction, a feature matrix X containing the temporal and spatial correlations of medical data is obtained, which is defined as:

[0048] Among them is the value of the medical data sequence collected at the i-th acquisition position at the t - m moment; is all the medical data sequence values collected at the i-th acquisition position.

[0049] Among them, the medical data sequence after feature construction includes trend data, periodic data, and proximity data. Then, the constructed data above is input into the first feature extraction module for feature extraction respectively. Then, the extracted feature vectors are input into the second feature extraction module to obtain corresponding trend features, periodic features, and proximity features. Then, the trend features, periodic features, and proximity features are fused to obtain a fused feature, and this fused feature is the output feature of the feature extraction model.

[0050] Specifically, the fusion of the trend feature, periodic feature, and proximity feature to obtain the fused feature is based on the following formula:

[0051] Y F = w a Y a + w b Y b + w c Y c , where Y F is the fused feature, Y a , Y b and Y c are the trend feature, periodic feature, and proximity feature respectively, and w a , w b and w c are the trend weight, periodic weight, and proximity weight respectively.

[0052] Among them, the second feature extraction module adopts a bidirectional long short-term memory network. The bidirectional long short-term memory network includes an input layer, a backward layer, a forward layer, and an output layer. A control unit is arranged between the backward layer and the forward layer. The control unit includes an input gate, a forget gate, and an output gate. The input gate is used to control the new information added to the control unit and determine the updated information. The forget gate determines the information that needs to be discarded from the control unit. The output gate determines the output information based on the current state of the control unit. And, to solve the problem of feature interference caused by fluctuating data, a fusion mechanism for the output data of the hidden layer is added to the output layer of the bidirectional long short-term memory network in the embodiment of the present application, which is a time series state supervision mechanism.

[0053] Among them, the state update formulas of the input gate, forget gate, and output gate at time t are as follows:

[0054] i t = σ(w i ·[h t-1 , x t + b i );

[0055] o t = σ(wo · [h t-1 , x t + b o );

[0056] f t = σ(w f · [h t-1 , x t + b f );

[0057] C t = f t × C t-1 + i t × tanh(w c · [h t-1 , x t + b c );

[0058] h t = o t × tanh(C t );where x t is the feature vector at time t, w f , w i and w o are the weights of the three gate structures, b f , b i , b o are the corresponding bias values, i t = σ(w i · [h t-1 , x t + b i ), o t = σ(w o · [h t-1 , x t + b o ), f t = σ(w f · [h t-1 , x t + b f ) are the current state calculation formulas respectively. C t = f t × C t-1 + i t × tanh(w c · [h t-1 , x t + b c ) is used to calculate the cell state of the control unit after integrating the discarded information and the candidate information. h t = o t × tanh(C t ) determines the final output information of the current state.

[0059] Among them, in order to better focus on and capture the changes in local states, the embodiments of the present application introduce a new implicit layer output data fusion mechanism - a time series state supervision mechanism. Specifically, the main processing procedure for the time series state supervision mechanism is as follows: First, expand the loop module according to the time series. After expansion, the time series module corresponds to the states at each moment of the time series indicators. Then, construct the decision weights of the time series data based on the output fluctuation values of all expanded modules. Among them, in order to construct the decision weights of the time series state supervision mechanism, it is necessary to calculate the fluctuation value for each expanded module separately.

[0060] Specifically: Denote w i as the decision weight of the time series indicator at the i-th moment, M as the number of loop modules, h i as the output data of the implicit layer of the i-th module, N as the dimension of the output data of the implicit layer, v i as the fluctuation value of the i-th module, F TSSM as the feature vector of the time series indicator data after adding the time series state supervision mechanism. The specific formula is as follows:

[0061]

[0062] Through the first feature extraction module and the second feature extraction module in the embodiments of the present application, the extraction of the feature vector in the medical data sequence can be realized. Regarding the training of the feature extraction module, it includes first initially training the second feature extraction module, and then performing secondary training on the initially trained second feature extraction module and the first feature extraction module to finally obtain a converged feature extraction module.

[0063] Among them, regarding the training of the second feature extraction module, in the embodiments of the present application, a hyperparameter optimization of the second feature extraction module is provided based on the trigonometric function pedigree arithmetic optimization algorithm to update its weights. The idea is to first initialize the algorithm parameters and the network structure, and then determine the fitness function, as shown in the following formula: where is the predicted value, y i is the true value, and n is the population size; then calculate and update the function fitness value and the optimal position, and stop the iteration after meeting the iteration end condition.

[0064] The specific training process is as follows: Generate the initialization parameters of the algorithm and the bidirectional long short-term memory network, calculate the fitness value of each individual, compare the fitness values of each agent and determine the best fitness value of the individual, update the optimal position of the agent, judge whether the convergence condition is met and perform iterative update until the convergence condition is met to obtain the optimized bidirectional long short-term memory network with the minimum mean square error.

[0065] The optimized second feature extraction module and the first feature extraction module are jointly trained to obtain the final feature extraction module. The training process is as follows: The read normalized medical data sequence is constructed to obtain a feature matrix containing temporal and spatial correlations. The model is built using the time series container Sequential. After the model network layer structure is built, the parameter values such as the convolution kernel and stride in the model are determined, and the Adam optimization algorithm is selected. The feature matrix is input into the convolutional neural network, and the convolutional neural network is used to effectively extract spatial features to obtain the feature vectors of the time series. The feature vectors are input into the bidirectional long short-term memory network, and the bidirectional long short-term memory network is used to extract time-dependent features. The model is trained and predicted from the forward network and the backward network according to parameters such as batchsize and epoch, and the corresponding loss curve and prediction results are output. The predicted time series values are output, and the prediction error value of the model is calculated.

[0066] Among them, the specific processing process of the Adam optimization algorithm is as follows:

[0067] For the input: Define the learning rate α, the initial parameter λ0, the numerical stability quantity ε, the first-order gradient decay coefficient β1, the second-order gradient decay coefficient β2, and the optimization function f(λ); the output is the final parameter λ T ; Initialization: m0 = 0, v0 = 0; Initialize the simple moving average: Calculate the gradient at the t-th step: g t = Δ λ f t (λ t-1 ); Calculate the second moment of the movement amount: Calculate the first moment of the movement amount: m t = β1m t-1 +(1 - β1)g t ; Calculate the correction of the moving deviation: Calculate the maximum value of the simple moving average:

[0068] If ρ t > 4, then calculate the correction value of the second moment of the movement amount: Calculate the variance correction term: Use adaptive momentum to update the parameters: Use non - adaptive momentum to update the parameters: λ t = λ t-1 - α t m t , and finally obtain λ T .

[0069] Among them, for the feature extraction module, the extracted feature vector is input into the classifier module to obtain a classification result.

[0070] In the embodiment of the present application, for the classifier module, the softmax function in the prior art is replaced with a sigmoid function. The processing process is to use the feature extraction result (h1′, h2′, h3′,..., h n ′) as the input, where n is the dimension of the feature vector, i.e., the number of label classes, and then it is transformed into a probability form ((p1, p2, p3,..., p n ), 0 < p i ≤ 1) through the sigmoid function. The expression of the function is as follows:

[0071] S(x) = 1 / (1 + e -x ), for a specific class k, if p k > sigmoid(d k ), then it belongs to class k.

[0072] Among them, for the loss function of the classifier module, the loss values of each classifier are integrated, and the specific formula is as follows:

[0073] Among them, p(y i = l i ) = sigmoid(d ij ) is the output of the sigmoid function, and Π is the indicator function.

[0074] Among them, for the sigmoid function threshold in the embodiment of the present application, it is obtained based on Gaussian fitting, and the sigmoid function threshold setting method based on Gaussian fitting plays a very good role in tightening the decision boundary. In traditional classification problems, the threshold of the sigmoid function is generally set to 0.5. However, in the actual application process, the threshold of the sigmoid function should often be greater than 0.5. Therefore, the application of this method is of great significance for improving the analysis accuracy of the classification model.

[0075] Among them, for the sigmoid function threshold setting method based on Gaussian fitting, it is first necessary to assume that the predicted probabilities of the sample data of each label class follow a Gaussian distribution with a mean of 1; then, the image of the Gaussian distribution is complemented, and the standard deviation of the Gaussian function after fitting is calculated; finally, the upper bound of the sigmoid function threshold is determined through the 3σ criterion. Based on the characteristics of the probability value, in order to achieve the goal of tightening the decision threshold of the sigmoid function, the lower bound of the threshold is taken as 0.5, that is, t i > 0.5. Therefore, the calculation formula of the sigmoid function threshold is: t i = max{0.5, 1 - 3σi}.

[0076] In the embodiment of the present application, the classifier module constructed by the above method is used to determine the classification probability corresponding to the feature vector, so as to determine the deep classification information of the current state, and send this information to the nursing application end, realizing the nursing monitoring of the patient.

[0077] Refer to Figure 3 , in the embodiment of the present application, a postoperative patient care monitoring device 300 is further provided. Similarly, this device is applied to the server, and the device includes:

[0078] An initial warning module 310, configured to receive the medical time series data, compare the preset standard monitoring data with the medical time series data to determine whether it exceeds the monitoring range, and send an alarm reminder to the nursing application end if it exceeds the monitoring range;

[0079] A risk identification module 320, when the medical time series data does not exceed the monitoring range, is configured to collect a plurality of the medical time series data within a standard time period to form a medical data sequence, input the medical data sequence into a preset risk identification model, process the medical data sequence based on the risk identification model to obtain risk information, and send the risk information to the nursing application end.

[0080] Refer to Figure 4 , the above method can also be integrated into the provided nursing monitoring device 400. Since the device may have relatively large differences due to different configurations or performances, it may include one or more processors 401 and a memory 402. One or more storage application programs or data may be stored in the memory 402. Among them, the memory 402 may be short-term storage or persistent storage. The application programs stored in the memory 402 may include one or more modules (not shown in the figure), and each module may include a series of computer-executable instructions in the nursing monitoring device. Further, the processor 401 may be set to communicate with the memory 402, and execute a series of computer-executable instructions in the memory 402 on the nursing monitoring device. The nursing monitoring device may further include one or more power supplies 403, one or more wired / wireless network interfaces 404, one or more input / output interfaces 405, one or more keyboards 406, etc.

[0081] In a specific embodiment, the nursing monitoring device includes a memory and one or more programs. One or more of the programs are stored in the memory, and one or more of the programs may include one or more modules. Each module may include a series of computer-executable instructions in the nursing monitoring device and is configured to be executed by one or more processors. The one or more programs include computer-executable instructions for performing the following:

[0082] Receive the medical time-series data, compare it with the pre-set standard monitoring data to determine whether it exceeds the monitoring range, and send an alarm reminder to the nursing application end if it exceeds the monitoring range;

[0083] If it does not exceed the monitoring range, collect a plurality of the medical time-series data within a standard time period to form a medical data sequence, input the medical data sequence into a pre-set risk identification model, process the medical data sequence based on the risk identification model to obtain risk information, and send the risk information to the nursing application end.

[0084] The following specifically introduces each component of the processor:

[0085] Among them, in this embodiment, the processor is an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application. For example: one or more digital signal processors (DSPs), or one or more field programmable gate arrays (FPGAs).

[0086] Optionally, the processor can execute various functions by running or executing software programs stored in the memory and calling data stored in the memory, such as executing the above Figure 2 shown method.

[0087] In a specific implementation, as an embodiment, the processor may include one or more microprocessors.

[0088] Among them, the memory is used to store the software program for executing the solution of the present application and is controlled by the processor for execution. The specific implementation method can refer to the above method embodiment and will not be elaborated here.

[0089] Optionally, the memory may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or may also be an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM), or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media, or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory may be integrated with the processor or exist independently and be coupled to the processing unit through the interface circuit of the processor. The embodiments of the present application do not make specific limitations on this.

[0090] It should be noted that the structure of the processor shown in this embodiment does not constitute a limitation on the device. The actual device may include more or fewer components than shown in the figure, or combine some components, or have a different component layout.

[0091] In addition, the technical effects of the processor can refer to the technical effects of the method described in the above method embodiments and will not be elaborated here.

[0092] It should be understood that the processor in the embodiments of the present application may be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0093] It should also be understood that the memory in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0094] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wired (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer, or a data storage device such as a server or data center that includes one or more collections of available media. The available media can be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0095] In the present application, "at least one" means one or more, and "a plurality" means two or more. "At least one of the following (items)" or a similar expression means any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can represent: a, b, c, a - b, a - c, b - c, or a - b - c, where a, b, and c can be single or multiple.

[0096] It should be understood that in various embodiments of the present application, the magnitudes of the serial numbers of the above processes do not mean the order of execution, and the order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0097] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present application.

[0098] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0099] In several embodiments provided in the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0100] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0101] In addition, in each embodiment of the present application, the functional units can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0102] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art or a part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present application. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0103] As described above, it is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims described above.

Claims

1. A postoperative patient care and monitoring method, characterized in that, The method is applied to a monitoring system, which includes a medical data acquisition device, a server, and a nursing application terminal. The medical data acquisition device communicates with the nursing application terminal. The medical data acquisition device is used to collect the medical time-series data of a patient and transmit the medical time-series data to the server. The method is applied to the server and includes: Receiving the medical time-series data, comparing the medical time-series data with preset standard monitoring data to determine whether it exceeds the monitoring range, and sending an alarm reminder to the nursing application terminal if it exceeds the monitoring range; If it does not exceed the monitoring range, collect a plurality of the medical time-series data within a standard time period to form a medical data sequence, input the medical data sequence into a preset risk identification model, process the medical data sequence based on the risk identification model to obtain risk information, and send the risk information to the nursing application terminal.

2. The postoperative patient care and monitoring method according to claim 1, wherein The medical time-series data includes first-type medical time-series data and second-type medical time-series data, and the standard monitoring data includes first-type standard monitoring data and second-type standard monitoring data. Comparing the preset standard monitoring data with the medical time-series data to determine whether it exceeds the monitoring range includes: comparing the preset first-type standard monitoring data and the second-type standard monitoring data with the first-type medical time-series data and the second-type medical time-series data respectively to determine whether it exceeds the monitoring range. The step of sending an alarm reminder to the nursing application terminal if it exceeds the monitoring range includes: sending an alarm reminder to the nursing application terminal when the second-type medical time-series data exceeds the monitoring range; when the second-type medical time-series data does not exceed the monitoring range, determine whether the first-type medical time-series data exceeds the monitoring range and the corresponding quantity of the exceeded monitoring range. If the quantity of the first-type medical time-series data that exceeds the monitoring range is greater than or equal to 2, send an alarm reminder to the nursing application terminal.

3. The postoperative patient care and monitoring method according to claim 2, wherein The method further includes: when the quantity of the first-type medical time-series data that exceeds the monitoring range is less than 2, obtain the generation frequency of this first-type medical time-series data within the standard time period, and send an alarm reminder to the nursing application terminal when the generation frequency is greater than 30%.

4. The postoperative patient care and monitoring method according to claim 3, characterized in that, The risk identification model includes a feature extraction module and a classifier module. The feature extraction module is used to extract features from the medical data sequence and obtain a feature vector, and the classifier module is used to receive the feature vector and determine the classification result.

5. The postoperative patient care and monitoring method according to claim 4, characterized in that The feature extraction module includes a first feature extraction module and a second feature extraction module constructed based on a convolutional network. The first feature extraction module is constructed based on a convolutional neural network, and the second feature extraction module is constructed based on a recurrent neural network. The second feature extraction module is a bidirectional long short-term memory network optimized by a trigonometric function pedigree arithmetic optimization algorithm. The feature extraction module is obtained by training the convolutional neural network and the optimized bidirectional long short-term memory network. The first feature extraction module is used to capture the spatial relationships of the medical data sequence, and the spatial relationships include trend relationships, periodic relationships, and proximity relationships. The second feature extraction module is used to capture and fuse the spatial relationships and then perform feature extraction.

6. The postoperative patient care and monitoring method according to claim 5, characterized in that, The second feature extraction module includes an input layer, a backward layer, a forward layer, and an output layer. A control unit is provided in the backward layer and the forward layer. The control unit includes an input gate, a forget gate, and an output gate. The control unit is used to control the new information added to the control unit and determine the updated information. The forget gate determines the information that needs to be discarded from the control unit, and the output gate determines the output information based on the current state of the control unit.

7. The postoperative patient care and monitoring method according to claim 6, characterized in that, The classifier module receives the feature vector, processes it through a sigmoid function to obtain the type probability corresponding to the feature vector, and corresponds the type probability to the label category probability. When the type probability is greater than the label category probability, the feature vector belongs to this label category.

8. The postoperative patient care and monitoring method according to claim 7, wherein, The optimization of the bidirectional long short-term memory network based on the trigonometric function pedigree arithmetic optimization algorithm includes the following steps: generating the initial parameters of the algorithm and the bidirectional long short-term memory network, calculating the fitness value of each individual, comparing the fitness values of each agent and determining the best fitness value of the individual, updating the optimal position of the agent, determining whether the convergence condition is satisfied and performing iterative updates until the convergence condition is satisfied, and obtaining the bidirectional long short-term memory network with the minimum mean square error.

9. The postoperative patient care and monitoring method according to claim 8, wherein, The training of the convolutional neural network and the optimized bidirectional long short-term memory network includes the following steps: constructing the read normalized medical data sequence data to obtain a feature matrix containing the temporal and spatial correlations of the medical data sequence; using the time series container row model to build, and after building the model network layer structure, determining the parameter values such as the convolution kernel and stride in the model; inputting the feature matrix into the convolutional neural network, and using the convolutional neural network to effectively extract the spatial features of the medical data sequence data to obtain the feature vector of the time series of the medical data sequence; inputting the feature vector extracted by the convolutional neural network into the optimized bidirectional long short-term memory network, and using the optimized bidirectional long short-term memory network to extract the long-term temporal dependence features of the medical data sequence; training and predicting the model from the forward network and the backward network according to the initial parameters, and outputting the corresponding loss curve and prediction result; outputting the predicted time series value of the medical data sequence, and calculating the prediction error value until the error value converges to the minimum.

10. A postoperative patient care and monitoring device, characterized in that, The device is applied to a monitoring system, which includes a medical data acquisition device, a server, and a nursing application terminal. The medical data acquisition device communicates with the nursing application terminal. The medical data acquisition device is used to collect the medical time-series data of a patient and transmit the medical time-series data to the server. The device is applied to the server and includes: An initial warning module, which is used to receive the medical time-series data, compare the pre-set standard monitoring data with the medical time-series data to determine whether it exceeds the monitoring range, and send an alarm reminder to the nursing application terminal if it exceeds the monitoring range; A risk identification module. When the medical time-series data does not exceed the monitoring range, it is used to collect multiple pieces of the medical time-series data within a standard time period to form a medical data sequence, input the medical data sequence into a pre-set risk identification model, process the medical data sequence based on the risk identification model to obtain risk information, and send the risk information to the nursing application terminal.

Images