Drug follow-up management method and device based on ERAS, equipment and medium
Through the drug follow-up management method in the context of ERAS, artificial intelligence technology is used to extract and fusion patient data, and a drug follow-up management report is generated, which solves the problem of low drug management efficiency in rapid rehabilitation, and improves patient recovery efficiency and Security.
Patent Information
- Application Number
- CN202411114852.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-14
- Publication Date
- 2024-11-12
AI Technical Summary
In the context of rapid rehabilitation (ERAS), the lack of effective drug follow-up management methods is difficult to assist in better implementation of the concept of rapid rehabilitation and improve the efficiency and safety of patients' postoperative recovery.
Using ERAS-based drug follow-up management method, the perioperative physiological, situational and clinical data are collected, and the convolutional neural network and long-term memory network model are used to extract and fusion the drug label prediction results, combined with the deep neural network, Automate the generation of drug follow-up management reports to reduce manual compliance.
It improves the accuracy and efficiency of drug follow-up management, reduces dependence on manpower, enhances the scientificity and personalization of drug use strategies, and promotes rapid recovery of patients.
Smart Images

Figure CN118942620A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an ERAS-based drug follow-up management method, device, equipment and medium. Background Art
[0002] Enhanced recovery after surgery (ERAS) is a new rehabilitation model that is different from the traditional one and a rehabilitation concept throughout the entire treatment process. It was first proposed by Danish surgeon Kehlet in 1997. It refers to a concept and technology based on evidence-based medicine and optimized by multidisciplinary cooperation. Through a series of perioperative optimization measures, it can reduce the stress response caused by surgery, reduce postoperative complications, and accelerate the recovery of organ and body functions, thereby accelerating the postoperative recovery of patients. Rapid recovery and minimally invasive surgery are important breakthroughs and development directions in the field of surgery in the 21st century, and are important measures that meet the current medical level and medical needs. Since the concept of ERAS was proposed, domestic and foreign researchers have continued to pay attention to and carry out research on ERAS pathways and perioperative management programs, and have implemented them systematically to varying degrees. At present, it is most widely used in multiple surgical fields such as gastrointestinal surgery, urology, and breast surgery. The rapid recovery pathway covers the entire process of patients from pre-admission to post-operative discharge, and requires a multidisciplinary team to continuously improve many of its links to improve safety, operability, and overall efficiency.
[0003] Therefore, in the context of ERAS, a drug follow-up management method is urgently needed to assist in better implementing the concept of rapid recovery. Summary of the invention
[0004] In view of the above, it is necessary to provide a drug follow-up management method, device, equipment and medium based on ERAS, which can realize effective follow-up management of drugs based on ERAS.
[0005] A drug follow-up management method based on ERAS, the drug follow-up management method based on ERAS comprising:
[0006] In response to a drug follow-up management instruction for a target user, collecting target physiological data of the target user during the perioperative period, target situational data within a specified period, and target clinical data during the perioperative period;
[0007] Performing time series segmentation on the target physiological data and the target situational data using a time point segmentation strategy to obtain continuous and time-series segmented data;
[0008] Using a first convolutional neural network model to extract features from the target physiological data to obtain target physiological features;
[0009] Using a long short-term memory network model to extract features from the time series segmentation data, to obtain target time series segmentation features;
[0010] Using a second convolutional neural network model to extract features from the target clinical data to obtain target clinical features;
[0011] Using a feature fusion model to fuse the target physiological feature, the target time series segmentation feature and the target clinical feature to obtain a target fusion feature;
[0012] Inputting the target fusion feature into a drug label prediction model trained based on a network model with a predictive function to obtain a label prediction result;
[0013] A drug follow-up management report for the target user is generated according to the tag prediction result.
[0014] According to a preferred embodiment of the present invention, the collecting of the target physiological data of the target user during the perioperative period, the target situational data within a specified period, and the target clinical data during the perioperative period includes:
[0015] Collecting physiological data of the target user within a first preset time period before the operation, during the operation, and within a second preset time period after the operation as the target physiological data;
[0016] Collecting the target user's lifestyle, disease situation and activity data during hospitalization within the specified time period as the target situation data; wherein the target situation data includes one or more of the following combinations: preoperative rehabilitation data, smoking data, preoperative bowel preparation data, catheter management data, intraoperative body temperature data, early postoperative feeding data, early postoperative activity and physical therapy data, and behavioral data;
[0017] The data used to construct the patient's health portrait are collected within the first preset time period before the operation and the second preset time period during and after the operation as the target clinical data.
[0018] According to a preferred embodiment of the present invention, before using the feature fusion model to perform feature fusion on the target physiological feature, the target time series segmentation feature and the target clinical feature, the method further includes:
[0019] Acquire a deep neural network, and initialize network parameters of the deep neural network to obtain an initial model;
[0020] Acquire training samples; wherein the training samples include a first sample set constructed by extracting features from physiological data using the first convolutional neural network model, a second sample set constructed by extracting features from the physiological data and contextual data after time series segmentation using the short-term memory network model, and a third sample set constructed by extracting features from clinical data using the second convolutional neural network model;
[0021] Continuously updating the network parameters of the initial model using the first sample set, the second sample set, and the third sample set, and obtaining the reconstruction error of the current round in each round of updating;
[0022] When it is detected that the reconstruction error is less than a preset threshold, the updating is stopped, and the currently obtained network parameters are determined as the final network parameters of the initial model to obtain the feature fusion model;
[0023] After initializing the network parameters of the deep neural network, the initialization network parameters of the initial model obtained include:
[0024]
[0025] Among them, Tr CNN-LSTM0 represents the initialization network parameters of the initial model, R represents the number of neurons in the input layer of the initial model, and W CNN-LSTM1 represents the weight from the input layer to the first hidden layer of the initial model, b CNN-LSTM1 represents the bias of the first hidden layer, W CNN-LSTM2 represents the weight from the first hidden layer to the second hidden layer of the initial model, b CNN-LSTM2 represents the bias of the second hidden layer, W CNN-LSTM3 represents the weight from the second hidden layer to the third hidden layer of the initial model, b CNN-LSTM3 represents the bias of the third hidden layer, h CNN-LSTM1 、h CNN-LSTM2 、h CNN-LSTM3 respectively represent the number of neurons in the first hidden layer, the second hidden layer, and the third hidden layer; rand represents a random number function that randomly generates 0 to 1, and zeros represents a zero function;
[0026] Among them, the reconstruction error of each round obtained during each update includes:
[0027] L1=L θ1 +L θ2 +L θ3 ;
[0028] Among them, L1 represents the reconstruction error of the current round obtained during each round of update; L θ1represents the reconstruction error of the network layer relative to each first sample obtained when each first sample in the first sample set is input into the current round model to update the network parameters in each round of update, L θ2 represents the reconstruction error of the network layer relative to each second sample obtained when each second sample in the second sample set is input into the current round model to update the network parameters in each round of updating; L θ3 It represents the reconstruction error of the network layer relative to each third sample obtained when each third sample in the third sample set is input into the current round model to update the network parameters in each round of updating.
[0029] According to a preferred embodiment of the present invention, the method further comprises:
[0030] Use the following formula to calculate L θ1 :
[0031]
[0032] Uc CNN1 =σ(W′ CNN-LSTM1 *σ(W′ CNN-LSTM2 *σ(W′ CNN-LSTM3 *X CNN +b′ CNN-LSTM3 ))+b′ CNN-LSTM2 ))+b′ CNN-LSTM1 );
[0033] Where n represents the total number of neurons, X CNN represents any of the first samples, Uc CNN1 Indicates that X CNN The network layer pair X obtained when input to the current round model to update the network parameters CNN The reconstruction value, W′ CNN-LSTM1 represents the weight transferred from the first hidden layer to the output layer of the current round model, W′ CNN-LSTM2 represents the weight transferred from the second hidden layer to the output layer, W′ CNN-LSTM3 represents the weight transferred from the third hidden layer to the output layer, b′ CNN-LSTM1 represents the bias of the output layer corresponding to the first hidden layer, b′ CNN-LSTM2 represents the bias of the output layer corresponding to the second hidden layer, b′ CNN-LSTM3 represents the bias of the output layer corresponding to the third hidden layer, and σ represents the activation function.
[0034] According to a preferred embodiment of the present invention, generating a drug follow-up management report for the target user according to the tag prediction result includes:
[0035] Obtaining a therapeutic utility simulation index, an adverse reaction simulation index, and a drug-related problem risk simulation index from the label prediction results;
[0036] Obtain clinical guidelines from a configuration database;
[0037] Matching the therapeutic utility simulation index, the adverse reaction simulation index, and the drug-related problem risk simulation index in the clinical guidelines to obtain an initial drug recommendation list;
[0038] Obtaining annotation information of the designated medical staff on the initial drug recommendation list, and obtaining the medication risk of each drug in the initial drug recommendation list;
[0039] Adding the annotation information to the initial drug recommendation list, and color-marking each drug in the initial drug recommendation list according to the medication risk of each drug, to obtain a target drug recommendation list;
[0040] Obtaining a medication follow-up strategy for each drug in the target drug recommendation list;
[0041] The drug follow-up management report is generated according to the target drug recommendation list and the medication follow-up strategy.
[0042] According to a preferred embodiment of the present invention, after generating the drug follow-up management report according to the target drug recommendation list and the medication follow-up strategy, the method further includes:
[0043] When it is detected that there is a change in the data of a drug in the target drug recommendation list, an early warning prompt is issued;
[0044] The change data include: dosage, side effects, and indications.
[0045] According to a preferred embodiment of the present invention, after generating the drug follow-up management report of the target user according to the tag prediction result, the method further includes:
[0046] Generating a follow-up strategy for the target user according to the drug follow-up management report;
[0047] According to the follow-up strategy, obtain the follow-up information of the target user, the terminal information of the terminal device used by the target user to upload the follow-up information, and the business data encapsulation type corresponding to the follow-up strategy as data to be processed;
[0048] Encrypting the data to be processed to obtain ciphertext data;
[0049] Mapping the ciphertext data to bits corresponding to the header to obtain intermediate data;
[0050] Encoding the intermediate data to obtain data to be transmitted;
[0051] The data to be transmitted is transmitted to a designated terminal device.
[0052] A drug follow-up management device based on ERAS, the drug follow-up management device based on ERAS comprising:
[0053] a collection unit, configured to collect, in response to a drug follow-up management instruction for a target user, target physiological data of the target user during the perioperative period, target situational data within a specified period, and target clinical data of the target user during the perioperative period;
[0054] A segmentation unit, used to perform time series segmentation on the target physiological data and the target situational data using a time point segmentation strategy to obtain continuous and time-series segmented data;
[0055] An extraction unit, configured to extract features from the target physiological data using a first convolutional neural network model to obtain target physiological features;
[0056] The extraction unit is further used to extract features from the time series segmentation data using a long short-term memory network model to obtain target time series segmentation features;
[0057] The extraction unit is further used to extract features from the target clinical data using a second convolutional neural network model to obtain target clinical features;
[0058] A fusion unit, used for fusing the target physiological feature, the target time series segmentation feature and the target clinical feature using a feature fusion model to obtain a target fusion feature;
[0059] An input unit, used for inputting the target fusion feature into a drug label prediction model trained based on a network model with a predictive function to obtain a label prediction result;
[0060] A generating unit is used to generate a drug follow-up management report for the target user according to the tag prediction result.
[0061] A computer device, comprising:
[0062] a memory storing at least one instruction; and
[0063] A processor executes the instructions stored in the memory to implement the ERAS-based drug follow-up management method.
[0064] A computer-readable storage medium stores at least one instruction, and the at least one instruction is executed by a processor in a computer device to implement the ERAS-based drug follow-up management method.
[0065] It can be seen from the above technical solutions that the present invention can collect the target physiological data of the target user during the perioperative period, the target situational data within a specified time period, and the target clinical data during the perioperative period, so as to comprehensively collect data and improve the accuracy of subsequent analysis; the target physiological data and the target situational data are segmented in time series using a time point segmentation strategy to highlight the temporal nature of the data; the convolutional neural network model and the long short-term memory network model are used for feature extraction respectively, and the extracted features are fused, and the fused features are input into the drug label prediction model for label prediction, so as to combine artificial intelligence means and multidimensional data to accurately predict labels, and automatically generate drug follow-up management reports, which not only improves the efficiency of report generation, but also effectively reduces the dependence on manpower. BRIEF DESCRIPTION OF THE DRAWINGS
[0066] Figure 1 It is a flow chart of a preferred embodiment of the drug follow-up management method based on ERAS of the present invention.
[0067] Figure 2 It is a functional module diagram of a preferred embodiment of the drug follow-up management device based on ERAS of the present invention.
[0068] Figure 3 It is a structural schematic diagram of a computer device of a preferred embodiment of the present invention for implementing the ERAS-based drug follow-up management method. DETAILED DESCRIPTION
[0069] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.
[0070] like Figure 1 , which is a flow chart of a preferred embodiment of the drug follow-up management method based on ERAS of the present invention. According to different requirements, the order of the steps in the flow chart can be changed, and some steps can be omitted.
[0071] The ERAS-based drug follow-up management method is applied to one or more computer devices, which are devices that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and their hardware includes but is not limited to microprocessors, application specific integrated circuits (ASICs), programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.
[0072] The computer device may be any electronic product that can perform human-computer interaction with a user, such as a personal computer, a tablet computer, a smart phone, a personal digital assistant (PDA), a game console, an interactive network television (IPTV), a smart wearable device, etc.
[0073] The computer device may also include a network device and / or a user device, wherein the network device includes, but is not limited to, a single network server, a server group consisting of multiple network servers, or a cloud consisting of a large number of hosts or network servers based on cloud computing.
[0074] The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), as well as big data and artificial intelligence platforms.
[0075] Among them, Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.
[0076] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.
[0077] The network where the computer device is located includes but is not limited to the Internet, a wide area network, a metropolitan area network, a local area network, a virtual private network (VPN), etc.
[0078] The nursing staff, clinical staff, and health professionals mentioned in this technical plan refer to professionals whose current positions are in health technical work among all employees paid by health institutions, including Chinese medicine practitioners, Western medicine practitioners, senior doctors of integrated Chinese and Western medicine, nurses, Chinese medicine pharmacists, Western pharmacists, laboratory technicians, other technicians, Chinese medicine practitioners, Western medicine practitioners, nurses, midwives, Chinese medicine pharmacists, Western pharmacists, laboratory technicians, other technicians, other Chinese medicine practitioners, nurses, Chinese medicine pharmacists, Western pharmacists, laboratory technicians, other technicians, other Chinese medicine practitioners, nurses, Chinese medicine pharmacists, Western pharmacists, laboratory technicians, other primary health technical personnel.
[0079] S10, in response to a drug follow-up management instruction for a target user, collecting target physiological data of the target user during the perioperative period, target situational data within a specified period, and target clinical data of the target user during the perioperative period.
[0080] In this embodiment, the target users may include postoperative patients.
[0081] In this embodiment, the medication follow-up management instruction may be automatically triggered after the surgical plan is confirmed.
[0082] In this embodiment, the acquisition of the target physiological data of the target user during the perioperative period, the target situational data within a specified period, and the target clinical data during the perioperative period includes:
[0083] Collecting physiological data of the target user within a first preset time period before the operation, during the operation, and within a second preset time period after the operation as the target physiological data;
[0084] Collecting the target user's lifestyle, disease situation and activity data during hospitalization within the specified time period as the target situation data; wherein the target situation data includes one or more of the following combinations: preoperative rehabilitation data, smoking data, preoperative bowel preparation data, catheter management data, intraoperative body temperature data, early postoperative feeding data, early postoperative activity and physical therapy data, and behavioral data;
[0085] The data used to construct the patient's health portrait are collected within the first preset time period before the operation and the second preset time period during and after the operation as the target clinical data.
[0086] For example, the first preset time period may be one week, the second preset time period may be two weeks, three weeks, one month, etc., and the designated time period may be the perioperative period.
[0087] The target physiological data may include, but is not limited to, a combination of one or more of the following data: heart rate, respiratory rate, body temperature, and other information related to physiological conditions. Specifically, data such as heart rate, respiratory rate, body temperature, ECG (electrocardiogram), PPG (Post-Prandial Glycaemia), heart sounds, and blood oxygen saturation can be input through implanted devices, manually input, or continuously recorded through smart portable devices (such as smart phones, etc.). In addition, the target physiological data may also include data recorded occasionally, such as weight and blood pressure measured at irregular intervals.
[0088] The target physiological data includes not only data related to physiological conditions, but also physiological parameters such as central physiological signals and peripheral physiological signals. The central physiological signals may include: electroencephalographic signals, magnetic brain signals, functional magnetic resonance imaging, etc.; the peripheral physiological signals may include: electromyographic signals, electrocardiographic signals, electrooculographic signals, galvanic skin response, respiratory signals, skin temperature signals, volume pulse waves, gastric electrical signals, blood pressure, heart sound signals, blood flow, oxygen partial pressure and carbon dioxide partial pressure, etc.
[0089] The target physiological data can also be collected in combination with wearable sensors, target physiological data monitoring devices and medical devices and / or environmental sensors, such as temperature sensors, altitude sensors, air quality sensors, and accelerometers. These sensors can be embedded in the living environment of the target user or integrated into portable devices such as smart phones and tablet computers.
[0090] The target context data may include information about actions, locations, activities, attributes and / or other similar situations related or associated with the patient during the detection time period.
[0091] Among them, the preoperative rehabilitation data includes preoperative exercise, which is an important part of preoperative education. It is through preoperative rehabilitation exercise that the patient's body functions are regulated, the patient's physiological functions are enhanced, and the patient can smoothly cope with the strong stress event of surgery, which is conducive to postoperative rehabilitation. There is no specific recommendation on the time, intensity and structure of preoperative exercise and the choice of patient's exercise type in the medical field, but it can generally include climbing stairs, deep breathing, walking machine (appropriately increasing the speed within an acceptable range), etc.
[0092] The smoking data: Smoking also has an important impact on patients' postoperative recovery and drug treatment methods. It is one of the high-risk factors for postoperative complications and readmission. It is generally believed that reducing smoking before surgery can reduce postoperative complications and improve the quality of patient recovery.
[0093] The preoperative intestinal preparation data: The general postoperative rehabilitation process is to fast for 12 hours and 6 hours before surgery, in order to reduce the risk of reflux aspiration during anesthesia induction. However, long-term fasting and abstinence on the night before surgery and the day of surgery will cause severe discomfort to patients. Hunger and thirst not only increase the secretion of insulin antagonists such as glucocorticoids and catecholamines, but may also aggravate the patient's postoperative insulin resistance. In the case of normal gastrointestinal function and no obstruction, solid food can be emptied after 6 hours, and liquid food can be emptied after 2 hours. Therefore, some medical guidelines recommend that patients take an appropriate amount of water 2 hours before surgery, stop taking solid food 6 hours before surgery, and still eat normally on the night before surgery. In large abdominal surgeries, it is recommended to take liquid carbohydrates before surgery so that the body's preoperative carbohydrate level reaches the metabolic state after eating, so as to reduce surgery-related insulin resistance and reduce the risk of related complications, and improve patient comfort, such as taking 200-400 ml of 10% glucose orally 2 hours before surgery. However, different types of liquid carbohydrates have different effects on gastric emptying, and there may be other complex beverages that are more suitable for use before surgery. For example, drinking 250-500 ml of soy milk on the morning of surgery can not only properly supplement carbohydrates, but also consume much less cholesterol and fat than milk.
[0094] Catheter management data: In traditional surgical models, urinary output is measured by indwelling urinary catheters during and after surgery to assess the amount of fluid replacement during and after surgery and to prevent postoperative urinary difficulties caused by anesthesia. However, the indwelling time of the urinary catheter has a variety of effects on the patient's postoperative recovery. For example, the urinary catheter will significantly affect the patient's postoperative comfort and easily cause restlessness during the anesthesia recovery period; or it may cause inconvenience to the patient's early postoperative bed activities, increase the risk of urinary tract infection, or cause urinary retention.
[0095] Intraoperative temperature data: When the body temperature drops during anesthesia and surgery, the patient's body temperature is prone to large fluctuations due to long-term exposure to the low temperature in the operating room, the inhibition of temperature regulation caused by anesthesia, a large amount of low-temperature fluid replacement, and evaporative losses caused by exposure of the surgical area to dry air. Low temperature may impair the body's normal coagulation function, increase the risk of bleeding, affect the body's white blood cell function, and increase the risk of cardiovascular complications. In addition, intraoperative low temperature may bring about changes in pharmacokinetic ability and affect postoperative anesthesia recovery.
[0096] The data on early postoperative feeding: In the traditional surgical model, it is believed that surgical patients should drink water 6 hours after surgery and eat after anal gas is discharged. In fact, most patients can tolerate eating before anal gas is discharged. On the one hand, long-term fasting after surgery brings extremely uncomfortable experience to patients, which may reduce compliance. On the other hand, eating may inhibit gastrointestinal motility. Studies have shown that early postoperative feeding is beneficial to the recovery of gastrointestinal function, protects intestinal mucosa, reduces the risk of infection, shortens hospital stay and reduces hospitalization costs, and it is safe and feasible to drink water within 6 hours after general anesthesia. Therefore, in the ERAS (Enhanced recovery after surgery) pathway, after the patient wakes up from anesthesia more thoroughly, he can drink a small amount of water within 6 hours after surgery, and can resume normal diet according to the patient's condition on the first morning after surgery. For complex cases or patients after major surgery, a step-by-step approach can be adopted to gradually adjust the diet to avoid situations where the patient cannot tolerate it.
[0097] Early postoperative mobilization and physical therapy data: Early postoperative ambulation is one of the most important links in the ERAS pathway, which can directly and effectively reduce complications related to immobilization and shorten hospital stay. In traditional concepts, due to anesthesia and surgical trauma, the first time a patient gets out of bed after surgery is usually 24 hours after surgery. Bed rest and immobilization not only reduce physical adaptability and weaken muscle strength, but also increase the incidence of atelectasis and pneumonia, and increase the risk of VTE and prolong hospital stay. Therefore, early ambulation of patients is of great significance to the implementation of ERAS, but due to large individual differences and different surgeries, there is currently no uniform standard for the specific time to get out of bed. In clinical practice, the time for getting out of bed should be formulated according to the specific situation of the patient, and patients should be encouraged to get out of bed as soon as possible while ensuring safety. The main factors affecting patients' early ambulation after surgery include orthostatic intolerance, limited activity (catheter, chest drainage tube, continuous fluid infusion), pain and poor compliance. Different from the traditional model and concept, each link in the ERAS pathway, such as avoiding fasting and drinking, early postoperative feeding, prevention and treatment of postoperative nausea and vomiting, unconventional indwelling catheterization, active strategy for chest tube removal, use of digital water seal bottle, restrictive fluid infusion, multimodal analgesia, etc., can effectively promote early ambulation after surgery. Therefore, education should be strengthened for patients and their families, and the importance of early activity should be repeatedly emphasized to change patients' inherent ideas and encourage them to cooperate actively.
[0098] Behavioral data: can be obtained from different sources, such as phone call records, emails, social media information interactions, and physical activities actively reported by patients. In addition, information related to patient behavior can be obtained by monitoring web browsing history, business activities, clickstream data, etc. A comprehensive understanding of the patient's behavioral data, including social interactions, daily activities, and Internet usage habits, can help to better understand the patient's lifestyle and behavior patterns. This in-depth understanding can be used to adjust personalized treatment plans, thereby improving treatment effectiveness.
[0099] Among them, the target clinical data may include administrative and demographic information, diagnosis, treatment, prescription drugs, over-the-counter drugs, laboratory test results, etc. These data can be collected through a variety of channels, including electronic medical record systems, laboratory information management systems, clinical databases, etc. These clinical data are clinical notes written by health professionals, digital nurse records, or data reported by subjects and / or health professionals through follow-up, questionnaires, general surveys, health event reports, clinical event reports, etc. These data provide detailed information about the patient's health status and help to more comprehensively evaluate the patient's condition and treatment effect. Comprehensive analysis of this information can provide the medical team with a comprehensive patient health portrait, so as to better formulate personalized treatment plans and health management strategies.
[0100] Through the above embodiments, multi-dimensional data of the target physiological data, target situational data and target clinical data of the target user can be collected to improve the accuracy of subsequent label prediction.
[0101] S11, performing time series segmentation on the target physiological data and the target situational data using a time point segmentation strategy to obtain continuous and time-series segmented data.
[0102] For example, the time series segmentation may be performed according to a specified time interval, and statistical analysis may be performed on the data in each segment to obtain the time series segmentation data.
[0103] S12, using the first convolutional neural network model to extract features of the target physiological data to obtain target physiological features.
[0104] S13, using a long short-term memory network model to extract features from the time series segmentation data to obtain target time series segmentation features.
[0105] S14, using a second convolutional neural network model to extract features of the target clinical data to obtain target clinical features.
[0106] S15, using a feature fusion model to fuse the target physiological feature, the target time series segmentation feature and the target clinical feature to obtain a target fusion feature.
[0107] In this embodiment, before the feature fusion model is used to perform feature fusion on the target physiological feature, the target time series segmentation feature and the target clinical feature, the method further includes:
[0108] Acquire a deep neural network, and initialize network parameters of the deep neural network to obtain an initial model;
[0109] Acquire training samples; wherein the training samples include a first sample set constructed by extracting features from physiological data using the first convolutional neural network model, a second sample set constructed by extracting features from the physiological data and contextual data after time series segmentation using the short-term memory network model, and a third sample set constructed by extracting features from clinical data using the second convolutional neural network model;
[0110] Continuously updating the network parameters of the initial model using the first sample set, the second sample set, and the third sample set, and obtaining the reconstruction error of the current round in each round of updating;
[0111] When it is detected that the reconstruction error is less than a preset threshold, the updating is stopped, and the currently obtained network parameters are determined as the final network parameters of the initial model to obtain the feature fusion model;
[0112] After initializing the network parameters of the deep neural network, the initialization network parameters of the initial model obtained include:
[0113]
[0114] Among them, Tr CNN-LSTM0 represents the initialization network parameters of the initial model, R represents the number of neurons in the input layer of the initial model, and W CNN-LSTM1 represents the weight from the input layer to the first hidden layer of the initial model, b CNN-LSTM1 represents the bias of the first hidden layer, W CNN-LSTM2 represents the weight from the first hidden layer to the second hidden layer of the initial model, b CNN-LSTM2 represents the bias of the second hidden layer, W CNN-LSTM3 represents the weight from the second hidden layer to the third hidden layer of the initial model, b CNN-LSTM3 represents the bias of the third hidden layer, h CNN-LSTM1 、h CNN-LSTM2 、h CNN-LSTM3respectively represent the number of neurons in the first hidden layer, the second hidden layer, and the third hidden layer; rand represents a random number function that randomly generates 0 to 1, and zeros represents a zero function;
[0115] Among them, the reconstruction error of each round obtained during each update includes:
[0116] L1=L θ1 +L θ2 +L θ3 ;
[0117] Among them, L1 represents the reconstruction error of the current round obtained during each round of update; L θ1 represents the reconstruction error of the network layer relative to each first sample obtained when each first sample in the first sample set is input into the current round model to update the network parameters in each round of update, L θ2 represents the reconstruction error of the network layer relative to each second sample obtained when each second sample in the second sample set is input into the current round model to update the network parameters in each round of updating; L θ3 It represents the reconstruction error of the network layer relative to each third sample obtained when each third sample in the third sample set is input into the current round model to update the network parameters in each round of updating.
[0118] The following formula is used to calculate L θ1 :
[0119]
[0120] Uc CNN1 =σ(W′ CNN-LSTM1 *σ(W′ CNN-LSTM2 *σ(W′ CNN-LSTM3 *X CNN +b′ CNN-LSTM3 ))+
[0121] b′ CNN-LSTM2 ))+b′ CNN-LSTM1 );
[0122] Where n represents the total number of neurons, X CNN represents any of the first samples, Uc CNN1 Indicates that X CNN The network layer pair X obtained when input to the current round model to update the network parameters CNN The reconstruction value, W′ CNN-LSTM1 represents the weight transferred from the first hidden layer to the output layer of the current round model, W′ CNN-LSTM2 represents the weight transferred from the second hidden layer to the output layer, W′ CNN-LSTM3 represents the weight transferred from the third hidden layer to the output layer, b′CNN-LSTM1 represents the bias of the output layer corresponding to the first hidden layer, b′ CNN-LSTM2 represents the bias of the output layer corresponding to the second hidden layer, b′ CNN-LSTM3 represents the bias of the output layer corresponding to the third hidden layer, and σ represents the activation function.
[0123] In the above embodiment, by reasonably configuring the reconstruction error and model parameters, an accurate feature fusion model can be trained to achieve accurate fusion of features in various dimensions.
[0124] S16, inputting the target fusion feature into a drug label prediction model trained based on a network model with predictive function to obtain a label prediction result.
[0125] In this embodiment, the label prediction results may include a therapeutic utility simulation index, an adverse reaction simulation index, and a drug-related problem risk simulation index.
[0126] The therapeutic utility simulation index is used to reflect the estimated value of the effectiveness of the drug in meeting the treatment expectations. The therapeutic utility index is a real-time measure of the effectiveness of the drug, taking into account the expected physiological effects, and represents the extent to which the treatment can meet expectations. The therapeutic utility index can be standardized within the range of 0 to 10. The larger the therapeutic utility index, the greater the positive impact or effect of the treatment.
[0127] Wherein, the adverse reaction simulation index is used to reflect the estimated value of the adverse reaction of the treatment. The adverse reaction index includes an estimate or measurement of one or more adverse reactions in the treatment, such as a measurement of the severity of one or more side effects or other adverse results. These adverse reactions may be known or unknown, and the measurement methods include but are not limited to real-time physiological data, patient-reported symptoms, questionnaires or laboratory reports. Therefore, the adverse reaction index can be continuous or sporadic. The adverse reaction index can also be quantified to vary in the range of 0 to 10, where the larger the adverse reaction index, the more severe the adverse reaction.
[0128] Among them, the drug-related problem risk simulation index is used to reflect the estimated risk of drug-related problems. Specifically, many drugs are known to have an impact on the individual's physiology and / or lifestyle. For example, beta-blockers have been confirmed to reduce the heart rate of patients with heart failure, while insulin can reduce the patient's overall blood sugar level, and analgesics with sedative effects may also reduce the patient's movement. The effects of these drugs may not only change physiological data, but also have an impact on situational data. Therefore, it is necessary to identify the risks of drug-related problems in order to configure drug use strategies more scientifically and reasonably.
[0129] In this embodiment, the label prediction results can also be used as feedback data for the training of the feature fusion model and the drug label prediction model. At the same time, the supplementary annotations of medical staff based on the label prediction results can also be fed back to the feature fusion model and the drug label prediction model for training, so as to continuously improve the model and improve the accuracy of the model.
[0130] In this embodiment, a machine learning algorithm can also be used to identify the relationship between the segmented data and the labels in the training samples, that is, the changes between the observed data and the previous or expected data, to feed back to the model and further improve the accuracy of the model.
[0131] In this embodiment, not only are individualized labels predicted by collecting data from different resources (such as physiological data and situational data from sensors, drug treatment plans, electronic medical records, etc.), as well as known positive and negative effects of treatments (such as symptoms reflected in physiological data, symptoms reflected in situational data, symptoms self-reported by patients, etc.), but clinical caregivers can also further optimize labels through annotations, and update the model in this cycle to ensure the accuracy of the model.
[0132] S17, generating a drug follow-up management report for the target user according to the tag prediction result.
[0133] In this embodiment, generating the drug follow-up management report for the target user according to the tag prediction result includes:
[0134] Obtaining a therapeutic utility simulation index, an adverse reaction simulation index, and a drug-related problem risk simulation index from the label prediction results;
[0135] Obtain clinical guidelines from a configuration database;
[0136] Matching the therapeutic utility simulation index, the adverse reaction simulation index, and the drug-related problem risk simulation index in the clinical guidelines to obtain an initial drug recommendation list;
[0137] Obtaining annotation information of the designated medical staff on the initial drug recommendation list, and obtaining the medication risk of each drug in the initial drug recommendation list;
[0138] Adding the annotation information to the initial drug recommendation list, and color-marking each drug in the initial drug recommendation list according to the medication risk of each drug, to obtain a target drug recommendation list;
[0139] Obtaining a medication follow-up strategy for each drug in the target drug recommendation list;
[0140] The drug follow-up management report is generated according to the target drug recommendation list and the medication follow-up strategy.
[0141] The configuration database may include various clinical guidelines, which may be uploaded and updated by medical staff to ensure the real-time availability of data.
[0142] Among them, the risks of medication may include being prohibited for use by pregnant women.
[0143] Among them, the color marking can be used to distinguish different medication risks.
[0144] Among them, the medication follow-up strategy can be used to remind patients to take medications and drug dosages in a timely manner, so as to assist in more scientific and convenient medication.
[0145] In this embodiment, after generating the drug follow-up management report according to the target drug recommendation list and the medication follow-up strategy, the method further includes:
[0146] When it is detected that there is a change in the data of a drug in the target drug recommendation list, an early warning prompt is issued;
[0147] The change data include: dosage, side effects, and indications.
[0148] Through the above embodiments, timely warning can be given when changes occur in the drugs, thus avoiding medication errors.
[0149] In this embodiment, after generating the drug follow-up management report for the target user according to the tag prediction result, the method further includes:
[0150] Generating a follow-up strategy for the target user according to the drug follow-up management report;
[0151] According to the follow-up strategy, obtain the follow-up information of the target user, the terminal information of the terminal device used by the target user to upload the follow-up information, and the business data encapsulation type corresponding to the follow-up strategy as data to be processed;
[0152] Encrypting the data to be processed to obtain ciphertext data;
[0153] Mapping the ciphertext data to bits corresponding to the header to obtain intermediate data;
[0154] Encoding the intermediate data to obtain data to be transmitted;
[0155] The data to be transmitted is transmitted to a designated terminal device.
[0156] Among them, any encryption algorithm can be used for encryption processing.
[0157] The designated terminal device may be a terminal device of a medical institution to follow up the follow-up data.
[0158] Through the above embodiments, the security of the data transmission process can be ensured, thereby protecting user privacy.
[0159] In this embodiment, educational materials may also be generated based on historical label prediction results and drug follow-up management reports for use in preoperative publicity.
[0160] In this embodiment, through the ERAS-based drug follow-up management, the patient is helped to maintain his health status or body balance during the period of taking medication and undergoing surgery, accelerate his recovery cycle, reduce hospitalization time, and ultimately convert into economic benefits, and monitor and manage the patient on this basis. At the same time, medical staff can quickly intervene and follow up based on the drug follow-up management report, which improves decision-making efficiency.
[0161] It can be seen from the above technical solutions that the present invention can collect the target physiological data of the target user during the perioperative period, the target situational data within a specified time period, and the target clinical data during the perioperative period, so as to comprehensively collect data and improve the accuracy of subsequent analysis; the target physiological data and the target situational data are segmented in time series using a time point segmentation strategy to highlight the temporal nature of the data; the convolutional neural network model and the long short-term memory network model are used for feature extraction respectively, and the extracted features are fused, and the fused features are input into the drug label prediction model for label prediction, so as to combine artificial intelligence means and multidimensional data to accurately predict labels, and automatically generate drug follow-up management reports, which not only improves the efficiency of report generation, but also effectively reduces the dependence on manpower.
[0162] like Figure 2 , which is a functional module diagram of a preferred embodiment of the drug follow-up management device based on ERAS of the present invention. The drug follow-up management device based on ERAS 11 includes a collection unit 110, a segmentation unit 111, an extraction unit 112, a fusion unit 113, an input unit 114, and a generation unit 115. The module / unit referred to in the present invention refers to a series of computer program segments that can be executed by a processor and can complete fixed functions, which are stored in a memory. In this embodiment, the functions of each module / unit will be described in detail in subsequent embodiments.
[0163] The collection unit 110 is used to collect the target physiological data of the target user during the perioperative period, the target situational data within a specified period, and the target clinical data of the target user during the perioperative period in response to the drug follow-up management instruction for the target user;
[0164] The segmentation unit 111 is used to perform time series segmentation on the target physiological data and the target situational data using a time point segmentation strategy to obtain continuous and time-series segmented data;
[0165] The extraction unit 112 is used to extract features from the target physiological data using a first convolutional neural network model to obtain target physiological features;
[0166] The extraction unit 112 is further used to extract features from the time series segmentation data using a long short-term memory network model to obtain target time series segmentation features;
[0167] The extraction unit 112 is further used to extract features from the target clinical data using a second convolutional neural network model to obtain target clinical features;
[0168] The fusion unit 113 is used to use a feature fusion model to perform feature fusion on the target physiological feature, the target time series segmentation feature and the target clinical feature to obtain a target fusion feature;
[0169] The input unit 114 is used to input the target fusion feature into a drug label prediction model trained based on a network model with a predictive function to obtain a label prediction result;
[0170] The generating unit 115 is used to generate a drug follow-up management report for the target user according to the tag prediction result.
[0171] It can be seen from the above technical solutions that the present invention can collect the target physiological data of the target user during the perioperative period, the target situational data within a specified time period, and the target clinical data during the perioperative period, so as to comprehensively collect data and improve the accuracy of subsequent analysis; the target physiological data and the target situational data are segmented in time series using a time point segmentation strategy to highlight the temporal nature of the data; the convolutional neural network model and the long short-term memory network model are used for feature extraction respectively, and the extracted features are fused, and the fused features are input into the drug label prediction model for label prediction, so as to combine artificial intelligence means and multidimensional data to accurately predict labels, and automatically generate drug follow-up management reports, which not only improves the efficiency of report generation, but also effectively reduces the dependence on manpower.
[0172] like Figure 3 , which is a schematic diagram of the structure of a computer device of a preferred embodiment of the present invention for implementing the ERAS-based drug follow-up management method.
[0173] The computer device 1 may include a memory 12 , a processor 13 , and a bus, and may also include a computer program stored in the memory 12 and executable on the processor 13 , such as an ERAS-based drug follow-up management program.
[0174] Those skilled in the art will appreciate that the schematic diagram is merely an example of the computer device 1 and does not constitute a limitation on the computer device 1. The computer device 1 may be a bus-type structure or a star-type structure. The computer device 1 may also include more or less other hardware or software than shown in the diagram, or a different arrangement of components. For example, the computer device 1 may also include input and output devices, network access devices, etc.
[0175] It should be noted that the computer device 1 is only an example, and other existing or future electronic products that are suitable for the present invention should also be included in the protection scope of the present invention and included here by reference.
[0176] Among them, the memory 12 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (for example: SD or DX memory, etc.), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 12 can be an internal storage unit of the computer device 1, such as a mobile hard disk of the computer device 1. In other embodiments, the memory 12 can also be an external storage device of the computer device 1, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the computer device 1. Further, the memory 12 can also include both an internal storage unit of the computer device 1 and an external storage device. The memory 12 can not only be used to store application software and various types of data installed in the computer device 1, such as the code of the drug follow-up management program based on ERAS, but also can be used to temporarily store data that has been output or is to be output.
[0177] In some embodiments, the processor 13 may be composed of an integrated circuit, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and combinations of various control chips. The processor 13 is the control core (Control Unit) of the computer device 1, and uses various interfaces and lines to connect various components of the entire computer device 1, and executes or executes programs or modules stored in the memory 12 (for example, executing an ERAS-based drug follow-up management program, etc.), and calls data stored in the memory 12 to execute various functions of the computer device 1 and process data.
[0178] The processor 13 executes the operating system of the computer device 1 and various installed applications. The processor 13 executes the applications to implement the steps in the above-mentioned embodiments of the ERAS-based drug follow-up management method, for example Figure 1 Steps shown.
[0179] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory 12 and executed by the processor 13 to complete the present invention. The one or more modules / units may be a series of computer-readable instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program in the computer device 1. For example, the computer program may be divided into a collection unit 110, a segmentation unit 111, an extraction unit 112, a fusion unit 113, an input unit 114, and a generation unit 115.
[0180] The above-mentioned integrated unit implemented in the form of a software function module can be stored in a computer-readable storage medium. The above-mentioned software function module is stored in a storage medium, and includes a number of instructions for enabling a computer device (which can be a personal computer, a computer device, or a network device, etc.) or a processor to execute the part of the ERAS-based drug follow-up management method described in various embodiments of the present invention.
[0181] If the module / unit integrated in the computer device 1 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 this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware devices through a computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of each of the above-mentioned method embodiments can be implemented.
[0182] The computer program includes computer program code, which may be in source code form, object code form, executable file or some intermediate form, etc. The computer readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory, etc.
[0183] Furthermore, the computer-readable storage medium may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function, etc.; the data storage area may store data created according to the use of the blockchain node, etc.
[0184] The blockchain referred to in this invention is a new application mode of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.
[0185] The bus can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 The bus is represented by only one straight line, but it does not mean that there is only one bus or one type of bus. The bus is configured to realize the connection and communication between the memory 12 and at least one processor 13, etc.
[0186] Although not shown, the computer device 1 may also include a power source (such as a battery) for supplying power to each component. Preferably, the power source may be logically connected to the at least one processor 13 through a power management device, so that the power management device can realize functions such as charging management, discharging management, and power consumption management. The power source may also include any components such as one or more DC or AC power sources, recharging devices, power failure detection circuits, power converters or inverters, power status indicators, etc. The computer device 1 may also include a variety of sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.
[0187] Furthermore, the computer device 1 may also include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is usually used to establish a communication connection between the computer device 1 and other computer devices.
[0188] Optionally, the computer device 1 may further include a user interface, which may be a display, an input unit (such as a keyboard), or a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touch device. The display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the computer device 1 and to display a visual user interface.
[0189] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.
[0190] Figure 3 Only the computer device 1 having components 12-13 is shown, and it can be understood by those skilled in the art that Figure 3 The structure shown does not constitute a limitation on the computer device 1, and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.
[0191] Combination Figure 1 The memory 12 in the computer device 1 stores a plurality of instructions to implement a drug follow-up management method based on ERAS, and the processor 13 can execute the plurality of instructions to implement:
[0192] In response to a drug follow-up management instruction for a target user, collecting target physiological data of the target user during the perioperative period, target situational data within a specified period, and target clinical data during the perioperative period;
[0193] Performing time series segmentation on the target physiological data and the target situational data using a time point segmentation strategy to obtain continuous and time-series segmented data;
[0194] Using a first convolutional neural network model to extract features from the target physiological data to obtain target physiological features;
[0195] Using a long short-term memory network model to extract features from the time series segmentation data, to obtain target time series segmentation features;
[0196] Using a second convolutional neural network model to extract features from the target clinical data to obtain target clinical features;
[0197] Using a feature fusion model to fuse the target physiological feature, the target time series segmentation feature and the target clinical feature to obtain a target fusion feature;
[0198] Inputting the target fusion feature into a drug label prediction model trained based on a network model with a predictive function to obtain a label prediction result;
[0199] A drug follow-up management report for the target user is generated according to the tag prediction result.
[0200] Specifically, the specific implementation method of the processor 13 for the above instructions can refer to Figure 1 The description of the relevant steps in the corresponding embodiments will not be repeated here.
[0201] It should be noted that the data involved in this case were all obtained legally.
[0202] In the several embodiments provided by the present invention, 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 only illustrative, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.
[0203] The present invention can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present invention can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present invention can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.
[0204] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0205] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.
[0206] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.
[0207] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any attached figure mark in the claims should not be regarded as limiting the claims involved.
[0208] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the present invention can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.
[0209] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.
Claims
1. A drug follow-up management method based on ERAS, characterized in that: The ERAS-based drug follow-up management method includes: In response to a drug follow-up management instruction for a target user, collecting target physiological data of the target user during the perioperative period, target situational data within a specified period, and target clinical data during the perioperative period; Performing time series segmentation on the target physiological data and the target situational data using a time point segmentation strategy to obtain continuous and time-series segmented data; Using a first convolutional neural network model to extract features from the target physiological data to obtain target physiological features; Using a long short-term memory network model to extract features from the time series segmentation data, to obtain target time series segmentation features; Using a second convolutional neural network model to extract features from the target clinical data to obtain target clinical features; Using a feature fusion model to fuse the target physiological feature, the target time series segmentation feature and the target clinical feature to obtain a target fusion feature; Inputting the target fusion feature into a drug label prediction model trained based on a network model with a predictive function to obtain a label prediction result; A drug follow-up management report for the target user is generated according to the tag prediction result.
2. The ERAS-based drug follow-up management method according to claim 1, characterized in that: The collecting of the target physiological data of the target user during the perioperative period, the target situational data within a specified period, and the target clinical data during the perioperative period includes: Collecting physiological data of the target user within a first preset time period before the operation, during the operation, and within a second preset time period after the operation as the target physiological data; Collecting the target user's lifestyle, disease situation and activity data during hospitalization within the specified time period as the target situation data; wherein the target situation data includes one or more of the following combinations: preoperative rehabilitation data, smoking data, preoperative bowel preparation data, catheter management data, intraoperative body temperature data, early postoperative feeding data, early postoperative activity and physical therapy data, and behavioral data; The data used to construct the patient's health portrait are collected within the first preset time period before the operation and the second preset time period during and after the operation as the target clinical data.
3. The ERAS-based drug follow-up management method according to claim 1, characterized in that: Before using the feature fusion model to perform feature fusion on the target physiological feature, the target time series segmentation feature and the target clinical feature, the method further includes: Acquire a deep neural network, and initialize network parameters of the deep neural network to obtain an initial model; Acquire training samples; wherein the training samples include a first sample set constructed by extracting features from physiological data using the first convolutional neural network model, a second sample set constructed by extracting features from the physiological data and contextual data after time series segmentation using the short-term memory network model, and a third sample set constructed by extracting features from clinical data using the second convolutional neural network model; Continuously updating the network parameters of the initial model using the first sample set, the second sample set, and the third sample set, and obtaining the reconstruction error of the current round in each round of updating; When it is detected that the reconstruction error is less than a preset threshold, the updating is stopped, and the currently obtained network parameters are determined as the final network parameters of the initial model to obtain the feature fusion model; After initializing the network parameters of the deep neural network, the initialization network parameters of the initial model obtained include: Among them, Tr CNN-LSTM0 represents the initialization network parameters of the initial model, R represents the number of neurons in the input layer of the initial model, and W CNN-LSTM1 represents the weight from the input layer to the first hidden layer of the initial model, b CNN-LSTM1 represents the bias of the first hidden layer, W CNN-LSTM2 represents the weight from the first hidden layer to the second hidden layer of the initial model, b CNN-LSTM2 represents the bias of the second hidden layer, W CNN-LSTM3 represents the weight from the second hidden layer to the third hidden layer of the initial model, b CNN-LSTM3 represents the bias of the third hidden layer, h CNN-LSTM1 、h CNN-LSTM2 、h CNN-LSTM3 respectively represent the number of neurons in the first hidden layer, the second hidden layer, and the third hidden layer; rand represents a random number function that randomly generates 0 to 1, and zeros represents a zero function; Among them, the reconstruction error of each round obtained during each update includes: L1=L θ1 +L θ2 +L θ3 ; Among them, L1 represents the reconstruction error of the current round obtained during each round of update; L θ1 represents the reconstruction error of the network layer relative to each first sample obtained when each first sample in the first sample set is input into the current round model to update the network parameters in each round of update, L θ2 represents the reconstruction error of the network layer relative to each second sample obtained when each second sample in the second sample set is input into the current round model to update the network parameters in each round of updating; L θ3 It represents the reconstruction error of the network layer relative to each third sample obtained when each third sample in the third sample set is input into the current round model to update the network parameters in each round of updating.
4. The ERAS-based drug follow-up management method according to claim 3, characterized in that: The method further comprises: Use the following formula to calculate L θ1 : Uc CNN1 =σ(W′ CNN-LSTM1 *σ(W′ CNN-LSTM2 *σ(W′ CNN-LSTM3 *X CNN +b′ CNN-LSTM3 ))+b′ CNN-LSTM2 ))+b′ CNN-LSTM1 ); Where n represents the total number of neurons, X CNN represents any of the first samples, Uc CNN1 Indicates that X CNN The network layer pair X obtained when input to the current round model to update the network parameters CNN The reconstruction value, W′ CNN-LSTM1 represents the weight transferred from the first hidden layer to the output layer of the current round model, W′ CNN-LSTM2 represents the weight transferred from the second hidden layer to the output layer, W′ CNN-LSTM3 represents the weight transferred from the third hidden layer to the output layer, b′ CNN-LSTM1 represents the bias of the output layer corresponding to the first hidden layer, b′ CNN-LSTM2 represents the bias of the output layer corresponding to the second hidden layer, b′ CNN-LSTM3 represents the bias of the output layer corresponding to the third hidden layer, and σ represents the activation function.
5. The ERAS-based drug follow-up management method according to claim 1, characterized in that: Generating the drug follow-up management report for the target user according to the tag prediction result includes: Obtaining a therapeutic utility simulation index, an adverse reaction simulation index, and a drug-related problem risk simulation index from the label prediction results; Obtain clinical guidelines from a configuration database; Matching the therapeutic utility simulation index, the adverse reaction simulation index, and the drug-related problem risk simulation index in the clinical guidelines to obtain an initial drug recommendation list; Obtaining annotation information of the designated medical staff on the initial drug recommendation list, and obtaining the medication risk of each drug in the initial drug recommendation list; Adding the annotation information to the initial drug recommendation list, and color-marking each drug in the initial drug recommendation list according to the medication risk of each drug, to obtain a target drug recommendation list; Obtaining a medication follow-up strategy for each drug in the target drug recommendation list; The drug follow-up management report is generated according to the target drug recommendation list and the medication follow-up strategy.
6. The ERAS-based drug follow-up management method according to claim 5, characterized in that: After generating the drug follow-up management report according to the target drug recommendation list and the medication follow-up strategy, the method further includes: When it is detected that there is a change in the data of a drug in the target drug recommendation list, an early warning prompt is issued; The change data include: dosage, side effects, and indications.
7. The ERAS-based drug follow-up management method according to claim 1, characterized in that: After generating the drug follow-up management report for the target user according to the tag prediction result, the method further includes: Generating a follow-up strategy for the target user according to the drug follow-up management report; According to the follow-up strategy, obtain the follow-up information of the target user, the terminal information of the terminal device used by the target user to upload the follow-up information, and the business data encapsulation type corresponding to the follow-up strategy as data to be processed; Encrypting the data to be processed to obtain ciphertext data; Mapping the ciphertext data to bits corresponding to the header to obtain intermediate data; Encoding the intermediate data to obtain data to be transmitted; The data to be transmitted is transmitted to a designated terminal device.
8. A drug follow-up management device based on ERAS, characterized in that: The ERAS-based drug follow-up management device includes: a collection unit, configured to collect, in response to a drug follow-up management instruction for a target user, target physiological data of the target user during the perioperative period, target situational data within a specified period, and target clinical data of the target user during the perioperative period; A segmentation unit, used to perform time series segmentation on the target physiological data and the target situational data using a time point segmentation strategy to obtain continuous and time-series segmented data; An extraction unit, configured to extract features from the target physiological data using a first convolutional neural network model to obtain target physiological features; The extraction unit is further used to extract features from the time series segmentation data using a long short-term memory network model to obtain target time series segmentation features; The extraction unit is further used to extract features from the target clinical data using a second convolutional neural network model to obtain target clinical features; A fusion unit, used for fusing the target physiological feature, the target time series segmentation feature and the target clinical feature using a feature fusion model to obtain a target fusion feature; An input unit, used for inputting the target fusion feature into a drug label prediction model trained based on a network model with a predictive function to obtain a label prediction result; A generating unit is used to generate a drug follow-up management report for the target user according to the tag prediction result.
9. A computer device, characterized in that: The computer device comprises: a memory storing at least one instruction; and A processor executes the instructions stored in the memory to implement the ERAS-based drug follow-up management method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores at least one instruction, and the at least one instruction is executed by a processor in a computer device to implement the ERAS-based drug follow-up management method as described in any one of claims 1 to 7.
Citation Information
Cited By
Eye medicine application management system with intelligent reminding function
CN120853993A
