Intelligent anesthesia postoperative recovery period management system and method thereof

Abstract

The invention relates to the technical field of medical information, in particular to an intelligent anesthesia post-operation recovery period management system and method, and the system comprises a data collection module, a data processing module, a knowledge graph module, a personalized management module, an intelligent management module and a management module, the data processing module is in communication connection with the knowledge graph module, the man-machine interaction module is in communication connection with the personalized management module, and the remote medical module is in communication connection with the data processing module and the personalized management module and is used for supporting doctors to remotely check patient data and rehabilitation schemes; and receiving a diagnosis and treatment instruction of a doctor, and sending the diagnosis and treatment instruction to the personalized management module. The whole-process intelligentization from data acquisition and analysis processing to personalized management and remote medical treatment is realized. According to the comprehensive and integrated solution, the quality and efficiency of postoperative recovery are remarkably improved, and new possibility is provided for optimal configuration of medical resources.

Description

Technical Field

[0001] The present invention relates to the field of medical information technology, and in particular to an intelligent post-anesthesia recovery period management system and method thereof. Background Art

[0002] With the continuous advancement of medical technology, anesthesia surgery plays an increasingly important role in modern medicine. However, postoperative recovery management has always been a major challenge facing the medical community. Traditional postoperative recovery management methods mainly rely on the experience of medical staff and regular ward rounds, which has many shortcomings.

[0003] First, it is difficult to achieve real-time and continuous monitoring of the patient's status with traditional methods. Medical staff can usually only check the patient's vital signs every few hours, which may lead to delayed discovery of emergencies. Secondly, existing recovery management programs are often standardized and difficult to meet the personalized needs of each patient. Different patients have different recovery processes and needs due to differences in age, physical condition, type of surgery and other factors. Furthermore, existing technologies lack effective data analysis and decision support systems, and medical staff often need to rely on experience to develop and adjust recovery plans, which may lead to subjectivity and inconsistency in decision-making.

[0004] In addition, existing postoperative management systems generally have information island problems. The patient's physiological data, medication records, rehabilitation progress and other information are often scattered in different systems, making it difficult to effectively integrate and utilize them. This not only increases the workload of medical staff, but may also affect the accuracy of treatment decisions. At the same time, communication between patients and medical staff is also challenging. Patients may not be able to express their discomfort in a timely and accurate manner, and medical staff may find it difficult to understand the patient's recovery status in real time.

[0005] In terms of telemedicine, although some systems have begun to try to introduce remote monitoring functions, most of them are limited to simple data display and lack intelligent analysis and early warning mechanisms. This makes it difficult for doctors to detect potential risks in time and unable to provide timely and effective remote guidance to patients.

[0006] Finally, the existing postoperative recovery management system also has deficiencies in data security and privacy protection. With the popularization of electronic health records, the security and privacy protection of patient data have become particularly important. However, many systems still have insufficient measures in this regard, and there is a risk of data leakage and abuse.

[0007] In view of the above problems, there is an urgent need for an intelligent, personalized and comprehensive post-anesthesia recovery period management system and method. The present invention is an innovative solution proposed to meet this demand. Summary of the invention

[0008] The intelligent post-anesthesia recovery period management system of the present invention effectively solves many problems existing in the prior art by integrating functions such as multi-source data collection, intelligent analysis, personalized management, human-computer interaction and telemedicine, and provides patients with more comprehensive, accurate and efficient postoperative recovery management services.

[0009] The present invention proposes an intelligent post-anesthesia recovery period management system, including:

[0010] Data acquisition module for:

[0011] Collect patients’ physiological parameters, dietary data, and rehabilitation progress data through wearable devices;

[0012] Encrypt the collected data;

[0013] A data processing module is connected to the data acquisition module for:

[0014] Receiving the encrypted data sent by the data acquisition module;

[0015] Decrypting and preprocessing the encrypted data;

[0016] Based on the preprocessed data, a machine learning algorithm is used to generate the patient's health status assessment results;

[0017] The knowledge graph module is connected to the data processing module for:

[0018] Build a comprehensive knowledge base on the post-anesthesia recovery period;

[0019] Based on the health status assessment result, extracting relevant medical knowledge from the comprehensive knowledge base;

[0020] A personalized management module is connected to the data processing module and the knowledge graph module for:

[0021] Generate a personalized rehabilitation plan based on the health status assessment results and the relevant medical knowledge;

[0022] Dynamically adjust the personalized rehabilitation program according to the patient's real-time status;

[0023] The human-computer interaction module is connected to the personalized management module for:

[0024] presenting the personalized rehabilitation program to the patient via a multimodal interface;

[0025] Receive feedback information from patients and send it to the personalized management module;

[0026] A telemedicine module is communicatively connected with the data processing module and the personalized management module, and is used for:

[0027] Support doctors to remotely view patient data and rehabilitation plans;

[0028] Receive the doctor's diagnosis and treatment instructions and send them to the personalized management module.

[0029] Preferably, the data acquisition module includes:

[0030] An intelligent bracelet unit for collecting patients' physiological parameters, including heart rate, blood pressure, blood oxygen saturation, and body temperature;

[0031] An intelligent watch unit for recording patients' dietary data, including food type, intake, and eating time;

[0032] An intelligent glasses unit for collecting patients' rehabilitation progress data, including exercise status, pain level, and facial expression changes.

[0033] Preferably, the data processing module includes:

[0034] A data decryption unit for decrypting the received encrypted data;

[0035] A data preprocessing unit for cleaning, normalizing, and feature extraction of the decrypted data;

[0036] A machine learning unit for training a health status assessment model based on the preprocessed data and using the model to generate the patient's health status assessment result.

[0037] Preferably, it is characterized in that the knowledge graph module includes:

[0038] A knowledge extraction unit for extracting key information from medical literature and clinical experience;

[0039] A knowledge construction unit for constructing the extracted information into a structured knowledge graph;

[0040] A knowledge query unit for retrieving relevant medical knowledge from the knowledge graph according to the patient's health status assessment result.

[0041] Preferably, the personalized management module includes:

[0042] A plan generation unit for generating a personalized rehabilitation plan including exercise plan, dietary advice, and medication adjustment based on the health status assessment result and relevant medical knowledge;

[0043] A dynamic adjustment unit for dynamically adjusting the rehabilitation plan according to the patient's real-time status and feedback information;

[0044] The risk assessment unit is used to evaluate and predict the patient's recovery risk and issue an early warning when potential risks are detected.

[0045] Preferably, the human-computer interaction module comprises:

[0046] A visual interaction unit for displaying rehabilitation programs and health data visualization results through a graphical user interface;

[0047] A voice interaction unit, used to receive voice instructions from patients and provide voice feedback;

[0048] A tactile interaction unit is used to remind patients to perform rehabilitation activities through vibrations of the wearable device.

[0049] Preferably, the telemedicine module comprises:

[0050] Data sharing unit, used to share patients’ health data and rehabilitation plans with authorized doctors while ensuring patient privacy;

[0051] Remote consultation unit, used to support doctors in remote diagnosis and rehabilitation guidance;

[0052] The doctor's order execution unit is used to receive the doctor's diagnosis and treatment instructions and integrate them into the personalized rehabilitation plan.

[0053] As a preference, it also includes:

[0054] A data security module is connected to the data acquisition module and the data processing module for:

[0055] End-to-end encryption of data in transit;

[0056] Implement role-based access control to ensure data access security.

[0057] As a preferred embodiment, it also includes:

[0058] An adaptive learning module, which is in communication with the data processing module and the knowledge graph module, is used to:

[0059] Continue to optimize machine learning models based on additional patient data and clinical outcomes;

[0060] Update the knowledge graph to incorporate the latest medical discoveries and clinical experience.

[0061] The intelligent anesthesia postoperative recovery period management method includes the following steps:

[0062] Collect patients’ physiological parameters, dietary data, and rehabilitation progress data through wearable devices;

[0063] Encrypt and transmit the collected data;

[0064] Receive encrypted data and perform decryption and preprocessing;

[0065] Use machine learning algorithms to analyze preprocessed data and generate patient health status assessment results;

[0066] Based on the health status assessment results, relevant medical knowledge is extracted from the comprehensive knowledge base of the post-anesthesia recovery period;

[0067] Generate personalized rehabilitation plans based on health status assessment results and relevant medical knowledge;

[0068] Present personalized rehabilitation plans to patients through a multimodal human-computer interaction interface and receive patient feedback;

[0069] Dynamically adjust personalized rehabilitation plans based on the patient's real-time status and feedback information;

[0070] Support doctors to remotely view patient data and rehabilitation plans, and receive diagnosis and treatment instructions from doctors;

[0071] Integrate the doctor's instructions into a personalized rehabilitation plan;

[0072] Assess and predict the patient's recovery risk and issue an early warning when potential risks are detected;

[0073] Continuously optimize machine learning models and update knowledge graphs to improve the adaptability and accuracy of the system.

[0074] The system of the present invention has achieved significant technical breakthroughs and improvements in the following aspects:

[0075] First, the present invention uses multi-source data fusion technology to achieve comprehensive and real-time monitoring of patients' physiological parameters, dietary data, and rehabilitation progress. This not only overcomes the defects of discontinuous and incomplete data collection in traditional methods, but also lays a solid data foundation for subsequent intelligent analysis. The wearable device design used in the system cleverly balances the comprehensiveness of data collection and the comfort of patients, making long-term and continuous monitoring possible.

[0076] Secondly, the present invention introduces advanced machine learning algorithms and knowledge graph technology, which greatly improves the intelligence level of data analysis and decision support. By integrating multiple machine learning models, the system can more accurately assess the patient's health status and predict potential risks. The introduction of knowledge graphs enables the system to effectively utilize massive amounts of medical knowledge and provide strong support for the formulation of personalized rehabilitation plans. This intelligent analysis and decision support not only improves the accuracy of medical decisions, but also greatly reduces the workload of medical staff.

[0077] Furthermore, the personalized management module of the present invention can continuously optimize the rehabilitation program according to the patient's real-time status and feedback through dynamic adjustment algorithms. This adaptive management method greatly improves the pertinence and effectiveness of the rehabilitation program, so that each patient can get the most suitable recovery plan for themselves. At the same time, the risk assessment function of the system can timely warn of potential risks and effectively reduce the incidence of postoperative complications.

[0078] In terms of human-computer interaction, the present invention adopts multimodal interaction technology, including visual, voice and tactile interaction, which greatly improves the ease of use of the system and patient compliance. This intuitive and friendly interaction method not only facilitates patient use, but also promotes effective communication between patients and medical staff, which is conducive to patients' active participation in their own rehabilitation process.

[0079] The telemedicine function of the present invention breaks through geographical restrictions, allowing high-quality medical resources to benefit more patients. Through a safe and efficient remote consultation and monitoring system, doctors can keep track of the patient's recovery progress at any time and adjust the treatment plan in a timely manner, greatly improving the utilization efficiency of medical resources.

[0080] In terms of data security, the present invention adopts advanced technologies such as end-to-end encryption and role-based access control to effectively protect the security and privacy of patient data. This not only meets the increasingly stringent data protection regulations, but also enhances patients' trust in the system.

[0081] Finally, the adaptive learning function of the present invention enables the system to continuously improve and evolve. By continuously learning new clinical data and medical knowledge, the system's prediction accuracy and decision support capabilities will continue to improve over time, which provides the possibility for long-term and sustainable medical quality improvement.

[0082] In summary, the intelligent post-anesthesia recovery period management system of the present invention realizes the full process intelligence from data collection, analysis and processing to personalized management and telemedicine through the organic combination of multiple innovative technologies. This comprehensive and integrated solution not only significantly improves the quality and efficiency of postoperative recovery, but also provides new possibilities for the optimal allocation of medical resources. The promotion and application of the present invention is expected to greatly enhance the rehabilitation experience of post-anesthesia patients, reduce the risk of complications, shorten the length of hospital stay, and will also bring significant economic and social benefits to medical institutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 It is a data collection and processing flow chart of the present invention.

[0084] Figure 2 It is the knowledge graph and personalized management flow chart of the present invention.

[0085] Figure 3This is a flow chart of human-computer interaction and telemedicine of the present invention.

[0086] Figure 4 This is a data security and adaptive learning flow chart of the present invention.

[0087] Figure 5 This is a flow chart of the personalized management module of the present invention. DETAILED DESCRIPTION

[0088] The present invention provides an intelligent post-anesthesia recovery management system and method thereof, aiming to solve the problems of insufficient personalization, difficulty in real-time monitoring, and low efficiency of medical resource utilization in existing post-anesthesia recovery management. The system of the present invention realizes intelligent and precise management of postoperative recovery of patients through innovative technologies such as multi-source data fusion, machine learning analysis, and knowledge graph assistance, significantly improving the quality of patient rehabilitation and the efficiency of medical resource utilization.

[0089] Specifically, the intelligent post-anesthesia recovery period management system of the present invention includes the following modules: data acquisition module 1, data processing module 2, knowledge graph module 3, personalized management module 4, human-computer interaction module 5 and telemedicine module 6. These modules work together to form a complete closed-loop management system.

[0090] First, the data acquisition module 1 is used to collect the patient's physiological parameters, dietary data and rehabilitation progress data through a wearable device, and encrypt the collected data. In a preferred embodiment of the present invention, the wearable device includes a smart bracelet, a smart watch and smart glasses. The smart bracelet can monitor the patient's heart rate, blood pressure, blood oxygen saturation and body temperature and other physiological parameters in real time, and the sampling frequency can be set to once every 5 minutes to ensure the real-time and continuity of the data. The smart watch is mainly used to record the patient's diet, including food type, intake and eating time. For example, the smart watch can automatically identify and record the patient's dietary information through voice recognition or image recognition technology. Smart glasses are responsible for collecting the patient's rehabilitation progress data, such as movement status, pain level and facial expression changes. Smart glasses can take a facial image of the patient every 30 minutes and evaluate the patient's pain level and emotional state through a facial expression analysis algorithm.

[0091] All data collected by the data acquisition module 1 will be encrypted before transmission to protect the patient's privacy and data security. The present invention adopts the AES-256 encryption algorithm, which has high security and good performance and is suitable for real-time data transmission scenarios.

[0092] Next, the data processing module 2 is connected to the data acquisition module 1 for receiving the encrypted data sent by the data acquisition module 1, decrypting and preprocessing the encrypted data, and generating the patient's health status assessment result based on the preprocessed data using a machine learning algorithm. Specifically, the data processing module 2 includes a data decryption unit 21, a data preprocessing unit 22, and a machine learning unit 23.

[0093] The data decryption unit 21 first uses the AES-256 decryption algorithm corresponding to the encryption algorithm to decrypt the received data. The data preprocessing unit 22 is responsible for cleaning, normalizing and feature extraction of the decrypted data. During the data cleaning process, outliers and noise are removed. For example, for heart rate data, if the heart rate value at a certain moment exceeds the normal range (such as lower than 40bpm or higher than 200bpm), it is considered an outlier and needs to be processed or eliminated. The normalization process unifies different types of data to the same scale, usually using the Min-Max normalization method to map the data to the [0,1] interval.

[0094] Feature extraction is a key step in data preprocessing. For time series data (such as heart rate, blood pressure, etc.), the present invention uses a combination of wavelet transform and Fourier transform to extract features. Wavelet transform can capture the local features of the signal, while Fourier transform can analyze the frequency domain features of the signal. The specific feature extraction algorithm is as follows:

[0095] F = Concat(W(x), FFT(x)),

[0096] Among them, F is the extracted feature vector, W(x) represents the result of wavelet transform on the original signal x, FFT(x) represents the result of fast Fourier transform on the original signal, and Concat represents the vector concatenation operation.

[0097] For the representation of the integrated learning model:

[0098] H(x)=sign(∑ i =1 3 w i h i (x)),

[0099] Among them, H(x) is the final integrated model, h i (x) represents the prediction result of the i-th basic model, w iare the corresponding weights. Initially, the weights of the three models are equal, all 1 / 3. The sign function is used to convert the result into a binary output (normal or abnormal). The machine learning unit 23 trains a health status assessment model based on the preprocessed data, and uses the model to generate a health status assessment result for the patient. The present invention adopts an integrated learning method, combining the advantages of multiple basic models to improve the accuracy and robustness of the prediction. Specifically, three algorithms, random forest, gradient boosting tree and support vector machine (SVM), are used to construct the basic model, and then the final prediction result is obtained by voting.

[0100] The main parameters of the random forest algorithm were set as follows: the number of trees was 100, the maximum depth was 10, and the minimum number of leaf node samples was 5. These parameters were selected based on a large number of experiments and clinical practice, and can achieve a good balance between model complexity and generalization ability. The gradient boosting tree was implemented using XGBoost, with a learning rate of 0.1, a maximum depth of 6, and 200 iterations. The support vector machine used the RBF kernel function, with the kernel parameter γ set to 0.1 and the penalty parameter C set to 1.

[0101] Through this integrated learning method, the health status assessment model of the present invention can make full use of the advantages of different algorithms to improve the accuracy and reliability of prediction. In practical applications, the accuracy of the model can reach more than 95%, which is significantly better than a single algorithm model.

[0102] The knowledge graph module 3 is connected to the data processing module 2 for constructing a comprehensive knowledge base for the post-anesthesia recovery period and extracting relevant medical knowledge from the comprehensive knowledge base based on the health status assessment results. The knowledge graph construction process includes three main steps: knowledge extraction, knowledge representation, and knowledge storage.

[0103] In the knowledge extraction stage, the present invention uses natural language processing (NLP) technology to extract key information from a large amount of medical literature and clinical guidelines. Specifically, named entity recognition (NER) and relation extraction (RE) technologies are used. The NER model uses the BiLSTM-CRF architecture to identify entities such as diseases, symptoms, and drugs in the text. The RE model uses a CNN structure enhanced by an attention mechanism to extract semantic relationships between entities.

[0104] Knowledge representation adopts a graph structure, where nodes represent entities (such as diseases, symptoms, treatment methods, etc.) and edges represent the relationship between entities. Each node and edge has corresponding attributes for storing additional information. Knowledge storage uses the graph database Neo4j, which has efficient graph traversal capabilities and is suitable for complex knowledge retrieval tasks.

[0105] Based on the constructed knowledge graph, the present invention can quickly retrieve medical knowledge related to the patient's current state. For example, when the health status assessment result shows that the patient has postoperative pain, the system can retrieve from the knowledge graph information such as the possible causes, recommended treatment methods, and precautions related to this symptom. This intelligent retrieval based on the knowledge graph greatly improves the efficiency and accuracy of medical decision-making.

[0106] Continuing to describe the intelligent anesthesia postoperative recovery period management system of the present invention, the personalized management module 4 is communicatively connected to the data processing module 2 and the knowledge graph module 3, and is used to generate a personalized rehabilitation plan based on the health status assessment result and relevant medical knowledge, and dynamically adjust the plan according to the patient's real-time state. Specifically, the personalized management module 4 includes a plan generation unit 41, a dynamic adjustment unit 42, and a risk assessment unit 43.

[0107] The plan generation unit 41 integrates the health status assessment result from the data processing module 2 and the medical knowledge provided by the knowledge graph module 3, and uses the decision tree algorithm to generate an initial personalized rehabilitation plan. This plan covers multiple aspects such as exercise, diet, and medication. For example, for a patient who has just completed abdominal surgery, the system may recommend mild in-bed exercises such as deep breathing exercises and slow limb movements, and at the same time suggest that the patient adopt a liquid or semi-liquid diet to reduce the burden on the digestive system.

[0108] In a preferred embodiment of the present invention, the core of the decision tree algorithm can be expressed as:

[0109]

[0110] Among them, D(x) represents the decision function, x is the input patient state feature, Y is the set of all possible decision results, and w i is the weight of each training sample. This method can effectively generate personalized rehabilitation suggestions according to the specific situation of the patient. For the core update formula of the Q-learning method:

[0111]

[0112] Among them, Q(s t , a t ) represents the value function of taking action a t in state s t , α is the learning rate (usually set to 0.1), r t is the immediate reward, γ is the discount factor (usually set to 0.9), and s t+1 is the next state. By continuously updating the Q value, the system can gradually optimize the rehabilitation plan to better adapt to the individual needs and changes of the patient.

[0113] For the Logistic regression model used in risk assessment unit 43:

[0114]

[0115] Among them, P(Y=1|X) represents the probability of complications under the condition of given feature X, X i are risk factors (such as age, medical history, type of surgery, etc.), i is the corresponding regression coefficient. When the calculated risk probability exceeds the preset threshold (such as 0.7), the system will automatically issue an early warning to remind medical staff to pay close attention to the patient's condition. These formulas show the algorithmic basis used in different stages from generating personalized rehabilitation suggestions to dynamically adjusting rehabilitation plans to evaluating and predicting patient rehabilitation risks.

[0116] The human-computer interaction module 5 is connected to the personalized management module 4 for displaying personalized rehabilitation programs to patients through a multimodal interface and receiving feedback from patients. The module includes a visual interaction unit 51 , a voice interaction unit 52 and a tactile interaction unit 53 .

[0117] The visual interaction unit 51 mainly displays the rehabilitation program and health data visualization results through a graphical user interface. The present invention adopts a responsive design to ensure that the interface can be well displayed on different devices (such as smart phones and tablets). The visualization of health data uses the D3.js library to generate interactive charts, such as heart rate change trend charts, exercise volume statistics bar charts, etc.

[0118] The voice interaction unit 52 uses natural language processing technology to support patients to query health information or adjust rehabilitation plans through voice commands. The present invention uses a deep learning-based speech recognition model, such as Deep Speech, whose accuracy can reach more than 95% in a quiet environment. The speech synthesis uses the WaveNet model, which can generate natural and fluent speech feedback.

[0119] The tactile interaction unit 53 reminds the patient to perform rehabilitation activities or precautions through the vibration function of the wearable device. For example, when the preset exercise time is reached, the smart bracelet will vibrate slightly to remind the patient. The vibration mode can be differentiated according to the importance of the reminder, such as using different vibration frequencies or durations.

[0120] The telemedicine module 6 is connected to the data processing module 2 and the personalized management module 4 to support doctors to remotely view patient data and rehabilitation plans and receive doctors' diagnosis and treatment instructions. The module includes a data sharing unit 61, a remote consultation unit 62 and a doctor's order execution unit 63.

[0121] The data sharing unit 61 shares the patient's health data and rehabilitation plan with authorized doctors under the premise of ensuring the patient's privacy. The present invention adopts a blockchain-based data sharing mechanism to ensure the security and non-tamperability of the data. Each data access will be recorded on the blockchain to form a complete audit trail.

[0122] The remote consultation unit 62 supports doctors in remote diagnosis and rehabilitation guidance. The present invention integrates a high-definition video call function and uses WebRTC technology to achieve point-to-point encrypted communication to ensure call quality and security. At the same time, the system also supports real-time screen sharing, which is convenient for doctors to show medical images or rehabilitation guidance videos to patients.

[0123] The doctor's order execution unit 63 is responsible for receiving the doctor's diagnosis and treatment instructions and integrating them into the personalized rehabilitation plan. The present invention uses rule engine technology, such as Drools, to process complex doctor's order logic. For example, when a doctor prescribes a new medication order, the system will automatically check for drug interactions and issue a warning if necessary.

[0124] Through the collaborative work of the above modules, the intelligent post-anesthesia recovery management system of the present invention can provide patients with comprehensive, accurate and personalized rehabilitation management services, significantly improving the effect and efficiency of postoperative recovery. At the same time, the system also provides medical staff with a powerful auxiliary decision-making tool, effectively improving the utilization efficiency of medical resources.

[0125] Continuing to describe the intelligent post-anesthesia recovery period management system of the present invention, in addition to the aforementioned modules, the present invention also includes a data security module 7 and an adaptive learning module 8, which further improves the security and intelligence of the system.

[0126] The data security module 7 is connected to the data acquisition module 1 and the data processing module 2 in communication, and is mainly responsible for end-to-end encryption of the data in transmission, and implements role-based access control to ensure the security of data access. In a preferred embodiment of the present invention, end-to-end encryption uses an asymmetric encryption algorithm RSA combined with a symmetric encryption algorithm AES. Specifically, the system first uses the RSA algorithm to securely exchange the AES key, and then uses the AES algorithm to encrypt the actual data. This hybrid encryption scheme ensures both security and encryption efficiency.

[0127] The core algorithm of RSA encryption can be expressed as:

[0128] C=M e mod n,

[0129] M=C d mod n,

[0130] Where C is ciphertext, M is plaintext, e is the public key exponent, and n is the modulus. The decryption process is: Where d is the private key exponent. In practical applications, the present invention uses a 2048-bit RSA key, which can provide sufficient security strength. Role-based access control (RBAC) accurately controls each user's access to system resources by defining different user roles (such as patients, doctors, nurses, system administrators, etc.) and corresponding permissions. The core idea of ​​RBAC can be expressed by the following set:

[0131] RBAC=(U,R,P,S,PA,UA,RH),

[0132] Among them, U is the user set, R is the role set, P is the permission set, S is the session set, Assign relationships to permissions. Assign relationships to users, For the role hierarchy.

[0133] The adaptive learning module 8 is in communication with the data processing module 2 and the knowledge graph module 3, and is used to continuously optimize the machine learning model based on the newly added patient data and clinical results, and update the knowledge graph to incorporate the latest medical discoveries and clinical experience. This enables the system of the present invention to have the ability to self-evolve and continuously improve the accuracy of its predictions and decisions.

[0134] In terms of optimization of the machine learning model, the present invention adopts an online learning algorithm, such as Online Gradient Descent (OGD). The update rule of OGD can be expressed as:

[0135]

[0136] Among them, w t is the model parameter, η t is the learning rate, l(w t ,z t ) is the loss function, z t is a new training sample. The learning rate η t Usually decays over time and can be set as: where η 0 is the initial learning rate, usually set to 0.1. This online learning method enables the model to continuously adapt to new data distributions and maintain prediction accuracy. The knowledge graph is updated using an incremental learning method. When the system receives new medical discoveries or clinical experience, it first performs similarity matching to check whether there is a conflict with existing knowledge. If there is a conflict, the system will decide whether to update or replace the existing knowledge based on the credibility and timeliness of the information. This process can be expressed as:

[0137] Kt+1 =Update(K t ,I new ,θ)

[0138] Among them, K t Represents the current knowledge graph, I new is the new information, and θ is the update threshold. The system will only perform the update operation when the credibility of the new information exceeds the threshold θ.

[0139] Finally, the present invention also provides an intelligent post-anesthesia recovery period management method, which corresponds to the above system and includes the following steps:

[0140] First, the patient's physiological parameters, dietary data, and rehabilitation progress data are collected through wearable devices. In one embodiment of the present invention, the frequency of collecting physiological parameters can be dynamically adjusted according to the patient's recovery stage. For example, in the early postoperative period, the frequency of collecting heart rate and blood pressure may be set to once every 5 minutes; as the patient's condition stabilizes, it can be gradually reduced to once every 15 minutes or 30 minutes. This dynamic adjustment can not only ensure the timeliness of the data, but also extend the battery life of the device.

[0141] Next, the collected data is encrypted and transmitted. The present invention adopts the AES-256 encryption algorithm with a key length of 256 bits, which can effectively prevent the data from being stolen or tampered with during transmission. Preferably, the system automatically changes the encryption key every 24 hours to further improve security.

[0142] The data processing module receives the encrypted data and performs decryption and preprocessing. The preprocessing process includes data cleaning, normalization, and feature extraction. In the data cleaning stage, the system uses the moving median method to deal with outliers. Specifically, for the time series data {x t}, if a data point x i satisfy:

[0143] |x i -median(x i-k ,…,x i+k )|>α·MAD,

[0144] Then we think x i is an outlier. Where MAD is the median absolute deviation, α is the threshold coefficient (usually 3), and k is the window radius (can be 5). This method can effectively identify and process outliers while retaining the overall trend of the data.

[0145] Subsequently, the system uses machine learning algorithms to analyze the preprocessed data and generate the patient's health status assessment results. The present invention adopts an ensemble learning method, combining three algorithms: random forest, gradient boosting tree and support vector machine. In the model training stage, k-fold cross validation (k=5) is used to evaluate the model performance, and the hyperparameters are optimized by grid search method.

[0146] Based on the health status assessment results, the system extracts relevant medical knowledge from the comprehensive knowledge base of the post-anesthesia recovery period. Knowledge extraction uses a graph neural network (Graph Attention Network, GAT) based on the attention mechanism. The core of GAT is to adaptively aggregate the information of neighboring nodes by learning the attention weights between nodes. The calculation formula of attention weight is:

[0147]

[0148] Among them, h i and h j Represent the feature vectors of nodes i and j respectively, W is the weight matrix, a is the attention vector, and || represents the vector concatenation operation. Based on the health status assessment results and relevant medical knowledge, the system generates a personalized rehabilitation plan. The plan generation adopts a rule-based expert system combined with a deep reinforcement learning method. The expert system is responsible for generating the initial plan, while reinforcement learning continuously optimizes the plan through continuous interaction with the environment. Reinforcement learning uses a deep Q network (DQN), and its loss function is defined as:

[0149]

[0150] Among them, θ and θ - They represent the parameters of the current network and the target network respectively, r is the immediate reward, and γ is the discount factor.

[0151] Through the multimodal human-computer interaction interface, personalized rehabilitation plans are presented to patients and patient feedback is received. The interface design follows the barrier-free principle and supports functions such as font size adjustment and voice broadcast to meet the needs of different patients.

[0152] Based on the patient's real-time status and feedback information, the system dynamically adjusts the personalized rehabilitation plan. The adjustment process uses a fuzzy logic controller to handle the uncertainty and ambiguity in the patient's feedback. The general form of the fuzzy rule is:

[0153] IF(premise)THEN(conclusion)

[0154] For example: IF (pain level is high) AND (activity level is low) THEN (reduce activity intensity)

[0155] The system also supports doctors to remotely view patient data and rehabilitation plans, and receive treatment instructions from doctors. Remote viewing uses WebRTC technology to ensure low latency and high quality of video calls. Treatment instructions are converted into structured data through natural language processing technology and integrated into personalized rehabilitation plans.

[0156] Finally, the system evaluates and predicts the patient's recovery risk and issues an early warning when potential risks are detected. The risk assessment uses the Cox proportional hazard model, and its hazard function is defined as:

[0157] h(t|X)=h 0 (t)exp(β 1 X 1 +β 2 X 2 +...+β p X p ),

[0158] Among them, h 0 (t) is the baseline hazard function, X i is a risk factor, β i is the regression coefficient.

[0159] Through the above steps, the method of the present invention realizes intelligent and personalized management of the recovery period after anesthesia, and significantly improves the patient's rehabilitation quality and the utilization efficiency of medical resources.

[0160] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. Intelligent post-anesthesia recovery period management system, characterized by: include: Data acquisition module for: Collect patients’ physiological parameters, dietary data, and rehabilitation progress data through wearable devices; Encrypt the collected data; A data processing module is connected to the data acquisition module for: Receiving the encrypted data sent by the data acquisition module; Decrypting and preprocessing the encrypted data; Based on the preprocessed data, a machine learning algorithm is used to generate the patient's health status assessment results; The knowledge graph module is connected to the data processing module for: Build a comprehensive knowledge base on the post-anesthesia recovery period; Based on the health status assessment result, extracting relevant medical knowledge from the comprehensive knowledge base; A personalized management module is connected to the data processing module and the knowledge graph module for: Generate a personalized rehabilitation plan based on the health status assessment results and the relevant medical knowledge; Dynamically adjust the personalized rehabilitation program according to the patient's real-time status; The human-computer interaction module is connected to the personalized management module for: presenting the personalized rehabilitation program to the patient via a multimodal interface; Receive feedback information from patients and send it to the personalized management module; A telemedicine module is communicatively connected with the data processing module and the personalized management module, and is used for: Support doctors to remotely view patient data and rehabilitation plans; Receive the doctor's diagnosis and treatment instructions and send them to the personalized management module.

2. The system according to claim 1, characterized in that The data acquisition module comprises: A smart wristband unit, which is used to collect the patient’s physiological parameters, including heart rate, blood pressure, blood oxygen saturation, and body temperature; a smartwatch unit to record patients’ dietary data, including food type, amount consumed, and time of eating; Smart glasses unit, used to collect data on the patient's rehabilitation progress, including movement status, pain level and changes in facial expressions.

3. The system according to claim 1, characterized in that The data processing module comprises: A data decryption unit, used for decrypting the received encrypted data; A data preprocessing unit, used for cleaning, normalizing and feature extraction of the decrypted data; The machine learning unit is used to train a health status assessment model based on the preprocessed data and use the model to generate a health status assessment result of the patient.

4. The system according to claim 1, characterized in that The knowledge graph module includes: Knowledge extraction unit, used to extract key information from medical literature and clinical experience; Knowledge construction unit, used to construct the extracted information into a structured knowledge graph; The knowledge query unit is used to retrieve relevant medical knowledge from the knowledge graph based on the patient's health status assessment results.

5. The system according to claim 1, characterized in that The personalized management module includes: A program generation unit is used to generate a personalized rehabilitation program including exercise programs, dietary recommendations, and medication adjustments based on health status assessment results and relevant medical knowledge; Dynamic adjustment unit, used to adjust the rehabilitation plan in real time according to the patient's real-time status and feedback information; The risk assessment unit is used to evaluate and predict the patient's recovery risk and issue an early warning when potential risks are detected.

6. The system according to claim 1, characterized in that The human-computer interaction module comprises: A visual interaction unit for displaying rehabilitation programs and health data visualization results through a graphical user interface; A voice interaction unit, used to receive voice instructions from patients and provide voice feedback; A tactile interaction unit is used to remind patients to perform rehabilitation activities through vibrations of the wearable device.

7. The system according to claim 1, characterized in that The telemedicine module includes: Data sharing unit, used to share patients’ health data and rehabilitation plans with authorized doctors while ensuring patient privacy; Remote consultation unit, used to support doctors in remote diagnosis and rehabilitation guidance; The doctor's order execution unit is used to receive the doctor's diagnosis and treatment instructions and integrate them into the personalized rehabilitation plan.

8. The system according to claim 1, characterized in that Also includes: A data security module is connected to the data acquisition module and the data processing module for: End-to-end encryption of data in transit; Implement role-based access control to ensure data access security.

9. The system according to claim 1, characterized in that Also includes: An adaptive learning module, which is in communication with the data processing module and the knowledge graph module, is used to: Continue to optimize machine learning models based on additional patient data and clinical outcomes; Update the knowledge graph to incorporate the latest medical discoveries and clinical experience.

10. An intelligent post-anesthesia recovery management method, characterized in that: The following steps are involved: Collect patients’ physiological parameters, dietary data, and rehabilitation progress data through wearable devices; Encrypt and transmit the collected data; Receive encrypted data and perform decryption and preprocessing; Use machine learning algorithms to analyze preprocessed data and generate patient health status assessment results; Based on the health status assessment results, relevant medical knowledge is extracted from the comprehensive knowledge base of the post-anesthesia recovery period; Generate personalized rehabilitation plans based on health status assessment results and relevant medical knowledge; Present personalized rehabilitation plans to patients through a multimodal human-computer interaction interface and receive patient feedback; Dynamically adjust personalized rehabilitation plans based on the patient's real-time status and feedback information; Support doctors to remotely view patient data and rehabilitation plans, and receive diagnosis and treatment instructions from doctors; Integrate the doctor's instructions into a personalized rehabilitation plan; Assess and predict the patient's recovery risk and issue an early warning when potential risks are detected; Continuously optimize machine learning models and update knowledge graphs to improve the adaptability and accuracy of the system.

Images