Intelligent thoracic surgery risk assessment and early warning system

Abstract

The invention discloses an intelligent thoracic surgery risk assessment and early warning system which is an advanced medical auxiliary tool integrating data processing, algorithm analysis and early warning prompt. The system collects individual data of patients, such as age, gender, ASA score, behavior condition, dyspnea score and the like, carries out accurate risk assessment by using an advanced machine learning algorithm, can monitor key indexes in the operation process in real time, and immediately sends out early warning information once abnormal conditions are found, so that the risk of the patients can be accurately assessed. Doctors and operation teams are reminded to take measures in time, and the operation risk is effectively reduced. In addition, the system can provide personalized risk assessment results according to actual conditions of patients, and help doctors to formulate more reasonable surgical plans and postoperative rehabilitation plans. The application of the system not only improves the accuracy and efficiency of risk assessment, but also optimizes the medical resource configuration, improves the satisfaction of patients and the medical quality, and provides a powerful guarantee for the safety and success rate of thoracic surgery.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical equipment, and in particular to a risk assessment and early warning system for thoracic surgery. BACKGROUND

[0002] With the rapid development of artificial intelligence (AI) technology, its application in the medical field is becoming increasingly widespread, especially in thoracic surgery, where AI technology has shown great potential. Currently, AI applications in thoracic surgery are mainly focused on the following aspects:

[0003] Image diagnosis: AI algorithms based on deep learning have been widely used in low-dose chest CT, which can accurately detect and classify lung nodules and predict the probability of malignancy, assisting doctors in rapid diagnosis and developing subsequent treatment strategies. AI can also predict potential risks during surgery by combining individual patient data.

[0004] Preoperative planning: AI provides strong support for preoperative planning in thoracic surgery through advanced three-dimensional reconstruction technology. This technology can convert CT or MRI image data into intuitive three-dimensional models, clearly showing the spatial relationship between tumors and surrounding anatomical structures (such as blood vessels, bronchi, and adjacent organs), allowing doctors to fully understand the complexity of the surgical area before surgery and develop the optimal surgical path.

[0005] Intraoperative assistance: During thoracic surgery, AI can analyze intraoperative data in real time to help doctors optimize their operation strategies. For example, in robot-assisted thoracoscopic surgery, AI algorithms adjust the mechanical arm's movement trajectory in real time to reduce errors and improve operation accuracy. Meanwhile, AI's real-time monitoring and analysis of patients' vital signs can help reduce the incidence of intraoperative complications.

[0006] Postoperative management: AI can efficiently integrate multi-modal data (including imaging, biomarkers, pathology, and intraoperative real-time monitoring data) to support individualized treatment. In addition, through the accumulation and training of postoperative data, the accuracy and reliability of AI systems will be further improved.

[0007] In the context of intelligent medicine, some hospitals have begun to apply AI technology to surgical risk assessment and early warning systems. These systems typically collect and mine diagnostic and treatment data in real time, use medical risk models to perform data analysis, achieve intelligent identification and prediction of medical risks, and provide key information warnings to help medical staff promptly grasp medical risk information and intervene in a timely manner.

[0008] Despite the significant progress made by AI technology in thoracic surgery risk assessment and early warning, the closest current technology still has some drawbacks:

[0009] Data standardization issues: Differences in data formats and standards among different medical institutions can cause difficulties for AI systems when processing and analyzing data. The lack of unified data standards and sharing platforms limits the widespread application and accuracy improvement of AI technology.

[0010] Privacy Protection Challenges: Medical data involves patient privacy, and how to protect patient privacy while using AI technology for risk assessment and early warning is a crucial issue. Currently, some AI systems still have security vulnerabilities in data processing and storage.

[0011] Technical limitations: While AI technology excels in some areas, it is still limited by factors such as algorithms and computing power. For example, the accuracy and reliability of AI systems may be affected when dealing with complex or rare cases.

[0012] In conclusion, although AI technology has shown great potential in risk assessment and early warning for thoracic surgery, the closest technologies to it still have drawbacks such as data standardization, privacy protection, ethical norms, technological limitations, and dependence and over-trust. Summary of the Invention

[0013] The purpose of this invention is to provide an intelligent thoracic surgery risk assessment and early warning system. This system collects patient medical history information, physiological parameters and other data, uses machine learning algorithms to build a risk assessment model, monitors the surgical process in real time and provides early warnings of potential risks.

[0014] To achieve the above objectives, the present invention adopts the following technical solution:

[0015] A smart thoracic surgery risk assessment and early warning system includes:

[0016] a) Data acquisition module: used to collect patients' medical history, physiological parameters and preoperative imaging data;

[0017] b) Data preprocessing module: Cleans and standardizes the data collected by the data acquisition module to eliminate outliers and ensure data consistency;

[0018] c) Risk assessment module: A risk assessment model is built using machine learning algorithms;

[0019] d) Early warning module: Set a risk threshold T. When the surgical risk score calculated by the risk assessment module exceeds the threshold T, an early warning signal is triggered.

[0020] e) Display module: Used to display risk assessment results, early warning information, and related patient data;

[0021] f) Model update module: Regularly collect new surgical case data and update the risk assessment model through incremental learning or retraining to improve the accuracy of assessment.

[0022] Furthermore, the risk assessment model calculates the surgical risk score based on the following formula:

[0023]

[0024] Where, x i w represents the i-th feature value after preprocessing. i The weight w represents the weight corresponding to the feature, b represents the bias term, and n represents the total number of features; i The bias term b is obtained by optimization using a supervised learning algorithm on the training dataset, wherein the supervised learning algorithm is logistic regression, random forest or support vector machine.

[0025] Furthermore, the machine learning algorithm uses cross-validation to evaluate model performance during the training phase and optimizes model parameters based on accuracy, recall, and F1 score metrics.

[0026] Furthermore, the risk assessment module further includes a deep learning submodule, which specifically performs the following steps:

[0027] Image data preprocessing: Preprocessing of preoperative imaging data, including image enhancement, noise reduction, and normalization.

[0028] Feature extraction: A convolutional neural network structure is used to automatically extract key features from the image through multiple convolutional layers, pooling layers and fully connected layers. Key features include tumor size, location and the clarity of the boundary with surrounding tissues.

[0029] Classification and scoring: The extracted features are input into the classifier and combined with non-image features to calculate the surgical risk score.

[0030] Furthermore, the early warning module also includes a time-sensitive early warning submodule, which is specifically composed of the following units:

[0031] Surgical stage identification unit: Identifies the current stage of the surgery by monitoring signals inside and outside the operating room in real time;

[0032] Dynamic threshold adjustment: The risk threshold T is dynamically adjusted according to different surgical stages. During critical steps or high-risk stages of surgery, the threshold is lowered to improve early warning sensitivity.

[0033] Optimized warning signals: Different warning signals are used according to the urgency of the warning, including sound, light color, and screen flashing, to intuitively and quickly notify doctors and surgical teams.

[0034] Furthermore, the display module also includes the following functional units:

[0035] Interactive data input unit: Provides a user interface that allows doctors and surgical teams to enter additional patient information at any time before or during surgery, including laboratory test results and intraoperative findings. The additional patient information will be automatically integrated into the risk assessment model.

[0036] Risk assessment results visualization unit: The risk assessment results are displayed intuitively in the form of charts and dashboards, including surgical risk scores and the contribution of each feature to the score, to help doctors and surgical teams quickly understand the sources of risk;

[0037] Warning History Record Unit: Records detailed information for each warning, including the time the warning was triggered, the reason, and the response measures taken, to facilitate subsequent analysis and improvement.

[0038] Furthermore, the system also includes a security verification module, which specifically performs the following steps:

[0039] Simulated surgical scenario construction: Based on historical surgical data and expert knowledge, a series of simulated surgical scenarios are constructed, including different surgical types, patient characteristics, and surgical risk levels;

[0040] Model performance testing: Run the risk assessment model in a simulated surgical scenario and record the model's prediction accuracy, stability, and other indicators;

[0041] Security Analysis Report: Generate a security analysis report that points out potential limitations of the model, false positive rate, false negative rate, and proposes improvement measures;

[0042] Model Iteration and Optimization: Based on the safety analysis report, the model is adjusted and optimized to improve its safety and reliability in actual surgery.

[0043] This invention proposes an intelligent thoracic surgery risk assessment and early warning system, which has the following beneficial effects:

[0044] I. Improve the accuracy and efficiency of risk assessment

[0045] The intelligent thoracic surgery risk assessment and early warning system integrates advanced algorithms and models to automatically and quickly process and analyze large amounts of medical data, including key information such as the patient's age, gender, ASA score, behavioral status, dyspnea score, timing of surgery, surgical method, pathological results, and comorbidities. This automated processing method not only significantly improves the accuracy and efficiency of risk assessment, but also reduces errors and uncertainties caused by human factors.

[0046] II. Achieving Personalized Risk Assessment

[0047] This system can perform personalized risk assessments based on individual patient data, such as age, gender, and medical history. This personalized assessment method is more in line with the patient's actual situation and can provide doctors with more accurate and reliable surgical risk prediction results, which helps doctors to develop more reasonable surgical plans and postoperative rehabilitation plans.

[0048] III. Provide early warning information to reduce surgical risks

[0049] The intelligent thoracic surgery risk assessment and early warning system can monitor the patient's vital signs and key indicators during the operation in real time. Once an abnormality is detected, the system will immediately issue an early warning message to remind the doctor and surgical team to take timely measures to intervene. This real-time early warning mechanism helps to reduce surgical risks and improve the safety and success rate of the operation.

[0050] IV. Optimize the allocation of medical resources

[0051] Through the intelligent thoracic surgery risk assessment and early warning system, hospitals can allocate medical resources more rationally, such as operating rooms and medical staff. For high-risk patients, hospitals can prioritize experienced doctors and advanced medical equipment to ensure the safety and effectiveness of the surgery. At the same time, the system can also help hospitals optimize surgical procedures, improve surgical efficiency, and reduce medical costs.

[0052] V. Improve patient satisfaction and medical quality

[0053] The application of an intelligent thoracic surgery risk assessment and early warning system enables doctors to conduct a more comprehensive and in-depth risk assessment of patients before surgery, thereby developing more reasonable surgical plans and postoperative rehabilitation plans. This not only helps to reduce surgical risks and improve the success rate of surgery, but also enhances patient satisfaction and the quality of medical care. At the same time, the system's real-time early warning and monitoring functions also help doctors to detect and deal with abnormal situations in a timely manner, ensuring patient safety and recovery.

[0054] In conclusion, the application of the intelligent thoracic surgery risk assessment and early warning system has significant beneficial effects, including improving the accuracy and efficiency of risk assessment, enabling personalized risk assessment, providing early warning information to reduce surgical risks, optimizing the allocation of medical resources, and improving patient satisfaction and medical quality. These beneficial effects not only help improve medical standards and service quality but also bring patients a safer and more efficient medical service experience. Attached Figure Description

[0055] Figure 1 This is a system structure diagram of an intelligent thoracic surgery risk assessment and early warning system according to the present invention;

[0056] Figure 2This is a system flowchart of an intelligent thoracic surgery risk assessment and early warning system according to the present invention. Detailed Implementation

[0057] I. Technical Solution Content

[0058] This invention provides an intelligent thoracic surgery risk assessment and early warning system. The system integrates multiple modules such as data acquisition, preprocessing, risk assessment, early warning, display, and model updating. It aims to achieve accurate assessment and real-time early warning of thoracic surgery risks through advanced machine learning algorithms and deep learning technology.

[0059] II. System Modules and Functions

[0060] 1. Data Acquisition Module

[0061] This module is responsible for collecting patients' medical history, physiological parameters (heart rate, blood pressure, blood oxygen saturation), and preoperative imaging data (CT and MRI images). This data is automatically acquired through the electronic medical record system and medical device interface to ensure the accuracy and completeness of the data.

[0062] The specific implementation process is as follows:

[0063] 1.1 Data Collection and Preparation

[0064] Determine the types of data that need to be collected, including medical history, physiological parameters, and preoperative imaging data;

[0065] Establish connections with hospital electronic medical record systems and medical equipment interfaces to ensure automatic data acquisition;

[0066] Design a data collection table, clearly defining the field names, data types, and data lengths.

[0067] 1.2 Collection of medical history information

[0068] The electronic medical record system automatically extracts patients' medical history information, including past surgical history, disease history, and allergy history.

[0069] For medical history information that cannot be automatically extracted, such as family history of genetic diseases, it is obtained through manual inquiry and recording.

[0070] 1.3 Collection of physiological parameters

[0071] Through the medical device interface, the patient's physiological parameters, including heart rate, blood pressure, and blood oxygen saturation, are collected in real time.

[0072] Physiological parameters should be updated in real time and measured multiple times before surgery to ensure the accuracy and stability of the data.

[0073] 1.4 Collection of Preoperative Imaging Data

[0074] The system automatically acquires the patient's preoperative imaging data, including CT and MRI images, through medical device interfaces or image storage and transmission systems (PACS).

[0075] Imaging data should be stored in DICOM file format to ensure data integrity and readability.

[0076] 1.5 Data Validation and Integration

[0077] The collected data is validated to check its integrity, accuracy, and consistency.

[0078] The verified data is integrated into the database to facilitate subsequent risk assessment and early warning analysis.

[0079] 1.6 Data Types

[0080] Medical history information

[0081] Data type: Text data;

[0082] Example fields: Name, Gender, Age, Past Surgical History, Medical History, Allergy History, Family History of Hereditary Diseases.

[0083] physiological parameters

[0084] Data type: Numeric data;

[0085] Example fields: Heart rate (beats / min), blood pressure (mmHg), blood oxygen saturation (%);

[0086] Physiological parameters are usually real-time dynamic data and need to be measured multiple times before surgery to obtain stable data values.

[0087] Preoperative imaging data

[0088] Data type: Image data;

[0089] Example fields: CT images, MRI images;

[0090] Imaging data is stored in DICOM file format, containing image pixel values, image size, and resolution information. This data is crucial for subsequent image processing and feature extraction.

[0091] In summary, the data acquisition module collects patients' medical history, physiological parameters, and preoperative imaging data through a combination of automatic acquisition and manual recording. After verification and integration, this data provides a solid foundation for subsequent risk assessment and early warning analysis.

[0092] 2. Data Preprocessing Module

[0093] The data preprocessing module cleans, standardizes, and normalizes the collected data to eliminate outliers, missing values, and noise, ensuring data quality. For imaging examination data, image enhancement and denoising preprocessing operations are also performed to improve the accuracy of feature extraction.

[0094] The specific implementation process is as follows:

[0095] 2.1 Data Cleaning

[0096] Identify outliers:

[0097] Statistical methods (3σ principle, box plots) are used to detect outliers in the data;

[0098] Analyze the identified outliers to determine whether they are genuine anomalies or data entry errors;

[0099] Based on the analysis results, a decision is made as to whether to delete outliers or replace them with appropriate values ​​(mean, median).

[0100] Handling missing values:

[0101] Check the data for missing values ​​and record the location and number of missing values;

[0102] Choose an appropriate imputation method based on the cause of the missing values ​​and the characteristics of the dataset, including mean imputation, median imputation, interpolation imputation, or imputation based on prediction of other features;

[0103] For missing values ​​that cannot be effectively filled, consider deleting the relevant data rows or columns.

[0104] Noise removal:

[0105] Smooth the data to reduce the impact of random noise;

[0106] For physiological parameter data, moving average or filtering algorithms can be used for noise reduction.

[0107] 2.2 Data Standardization and Normalization

[0108] Data standardization:

[0109] Transform the data into a standard normal distribution with a mean of 0 and a standard deviation of 1.

[0110] The standardized formula is:

[0111] Where x represents the original data, mean represents the mean, and σ represents the standard deviation.

[0112] Data normalization:

[0113] Scaling the data to a specified range (e.g., [0,1]);

[0114] The normalization formula is:

[0115] Where x is the original data, min is the minimum value, and max is the maximum value.

[0116] 2.3 Preprocessing of Imaging Examination Data

[0117] Image enhancement:

[0118] Enhancement processing is performed on imaging data such as CT and MRI to improve image contrast and clarity;

[0119] Common image enhancement methods include histogram equalization and contrast stretching.

[0120] Image denoising:

[0121] Median filtering and Gaussian filtering algorithms are used to denoise the image in order to reduce the impact of noise on feature extraction.

[0122] Choose an appropriate denoising algorithm based on the type and distribution of noise.

[0123] Image standardization:

[0124] The pixel values ​​of an image are standardized so that the pixel values ​​of different images have the same scale and distribution;

[0125] This helps with subsequent feature extraction and model training.

[0126] 2.4 Data Validation and Integration

[0127] Data validation:

[0128] The preprocessed data is validated to check its integrity, accuracy, and consistency.

[0129] Ensure that the preprocessing process does not introduce new errors or outliers.

[0130] Data integration:

[0131] The preprocessed data is integrated into the database to facilitate subsequent risk assessment and early warning analysis.

[0132] Ensure that the integrated data is in the correct format and is easy to access and analyze.

[0133] In summary, the data preprocessing module ensures data quality and accuracy through a series of cleaning, standardization, and normalization operations, as well as special preprocessing operations for imaging examination data. This provides a reliable data foundation for subsequent risk assessment and early warning analysis.

[0134] 3. Risk Assessment Module

[0135] The risk assessment module is the core of the system. It uses machine learning algorithms to build a risk assessment model. This model combines the patient's medical history, physiological parameters, and imaging data to calculate the surgical risk score using the following formula:

[0136]

[0137] Where, x i w represents the i-th feature value after preprocessing. i The weight w represents the weight corresponding to the feature, b represents the bias term, and n represents the total number of features; i The bias term b is obtained through supervised learning algorithms on the training dataset, including logistic regression, random forest, or support vector machine optimization.

[0138] In addition, the risk assessment module also includes a deep learning submodule for extracting key features from imaging examination data. This submodule adopts a convolutional neural network (CNN) structure, which automatically extracts features such as tumor size, location, and morphology in the image through multiple convolutional layers, pooling layers, and fully connected layers, and combines them with other non-imaging features to calculate the surgical risk score.

[0139] The risk assessment module uses machine learning algorithms to build a risk assessment model, the details of which are as follows:

[0140] Model Architecture

[0141] The risk assessment model is built on machine learning algorithms and mainly includes data preprocessing, feature selection, model training and validation, and risk score calculation.

[0142] Data preprocessing

[0143] Data collection: Collect individualized patient data from sources such as the Hospital Information System (HIS) and Electronic Medical Record System (EMR), including age, gender, ASA score, behavioral status, dyspnea score, timing of surgery, surgical procedure, pathology results, and comorbidities.

[0144] Data cleaning: Cleaning the collected data to remove duplicates, missing values, or outliers to ensure data quality and accuracy;

[0145] Data standardization: Standardize data of different dimensions so that each feature has a fair weight in the model.

[0146] Feature selection

[0147] Based on medical knowledge and experience, features closely related to surgical risk are selected from the preprocessed data and used as input variables for the model. These features should be able to comprehensively reflect factors such as the patient's physical condition, surgical difficulty, and postoperative recovery.

[0148] Model training and validation

[0149] Algorithm selection: Based on data characteristics and business needs, select appropriate machine learning algorithms for model training, such as logistic regression, random forest, and support vector machine;

[0150] Model training: The selected features are used as input variables, and surgical risk level or complication rate are used as output variables to train the model;

[0151] Model validation: The model is validated using cross-validation and hold-out methods to evaluate its accuracy and stability. At the same time, the model is adjusted and optimized based on the validation results.

[0152] Risk score calculation

[0153] Based on the trained model, the patient's individualized data is input into the model for calculation to obtain a surgical risk score. This score can reflect the risk level of the patient undergoing thoracic surgery and provide decision support for doctors and surgical teams.

[0154] In summary, the risk assessment model achieves accurate assessment of patients' surgical risks by integrating machine learning algorithms with data preprocessing and feature selection techniques.

[0155] The following is the specific structure of the model:

[0156] Model input layer

[0157] The input layer of the risk assessment model is responsible for receiving various types of data from surgical patients, including but not limited to:

[0158] Basic information: Patient's age, gender, height, and weight;

[0159] Medical history information: the patient's past medical history, surgical history, and drug allergy history;

[0160] Preoperative examination information: results of preoperative examinations including electrocardiogram, complete blood count, and urinalysis;

[0161] Intraoperative real-time data: vital signs monitoring data, blood loss, and operation time during the operation.

[0162] Feature Engineering Layer

[0163] The feature engineering layer preprocesses and extracts features from the input data to form feature vectors suitable for machine learning algorithms. The main tasks of this layer include:

[0164] Data cleaning: removing invalid, duplicate, and abnormal data;

[0165] Data standardization: Converting data of different dimensions into the same dimension for subsequent calculations;

[0166] Feature selection: Screening out features from the raw data that have a significant impact on risk assessment;

[0167] Feature transformation: Transform selected features to improve the predictive performance of the model.

[0168] Model training layer

[0169] The model training layer uses machine learning algorithms to train the feature vectors output by the feature engineering layer in order to build a risk assessment model. Commonly used machine learning algorithms include:

[0170] Supervised learning algorithms: Support Vector Machine, Random Forest, Neural Network. These algorithms require known output labels (i.e., surgical risk levels) for training.

[0171] Unsupervised learning algorithms: clustering analysis, these algorithms can discover hidden patterns in data without knowing the output labels;

[0172] During training, the algorithm iteratively optimizes the feature vectors to minimize prediction error and improve the model's generalization ability.

[0173] Model output layer

[0174] The model output layer is responsible for visualizing the results of the risk assessment model or outputting numerical or categorical labels that can be used for decision-making. The output results typically include:

[0175] Risk score: A quantifiable score that indicates the level of surgical risk;

[0176] Risk level: Classified according to risk score, low risk, medium risk, high risk;

[0177] Risk warning: When the risk score exceeds a certain threshold, an early warning mechanism is triggered to remind doctors and the surgical team to take appropriate preventive measures.

[0178] Model Evaluation and Optimization Layer

[0179] The model evaluation and optimization layer is responsible for evaluating and optimizing the performance of the risk assessment model. Evaluation metrics include accuracy, recall, and F1 score. Based on the evaluation results, the model can be tuned, such as adjusting algorithm parameters or increasing the number of features, to improve the model's predictive performance and stability.

[0180] In summary, the specific structure of the risk assessment model includes an input layer, a feature engineering layer, a model training layer, a model output layer, and a model evaluation and optimization layer. These layers work together to achieve accurate assessment and early warning of surgical patient risks.

[0181] The specific implementation process is as follows:

[0182] 3.1 Data Preparation

[0183] Data collection:

[0184] Collect patients' medical history, physiological parameters, and imaging examination data from hospital information systems (HIS), electronic medical record systems (EMR), and picture archiving and communication systems (PACS);

[0185] Ensure the integrity, accuracy, and consistency of the data.

[0186] Data preprocessing:

[0187] The collected data is cleaned, standardized, and normalized to eliminate outliers, missing values, and noise.

[0188] For imaging examination data, preprocessing operations such as image enhancement and denoising are performed to improve the accuracy of feature extraction.

[0189] 3.2 Feature Extraction

[0190] Non-image feature extraction:

[0191] Extract features related to surgical risk from medical history information and physiological parameters, such as age, gender, medical history, blood pressure, and heart rate;

[0192] These features will serve as one of the inputs to the risk assessment model.

[0193] Image feature extraction:

[0194] Key features are automatically extracted from imaging examination data using the convolutional neural network (CNN) structure in the deep learning submodule;

[0195] CNNs automatically learn and extract features such as tumor size, location, and morphology from images through multiple convolutional layers, pooling layers, and fully connected layers.

[0196] These imaging features, along with non-imaging features, will be used to calculate the surgical risk score.

[0197] 3.3 Model Construction and Optimization

[0198] Choosing a machine learning algorithm:

[0199] Choose the appropriate machine learning algorithm based on the characteristics of the data and the complexity of the problem, including logistic regression, random forest or support vector machine;

[0200] These algorithms will be used to build risk assessment models and calculate the weights and biases corresponding to the features.

[0201] Model training:

[0202] The model is trained using the preprocessed training dataset;

[0203] By optimizing weights and biases using supervised learning algorithms, the model can accurately predict surgical risk scores.

[0204] Model validation and tuning:

[0205] The trained model is validated using a validation dataset to evaluate its performance.

[0206] The model is then fine-tuned based on the validation results, including adjusting hyperparameters and adding features, to improve the model's accuracy and generalization ability.

[0207] 3.4 Surgical risk score calculation

[0208] Input feature values:

[0209] The preprocessed non-image features and image features are input into the risk assessment model.

[0210] Calculate the risk score:

[0211] According to the formula Calculate the surgical risk score;

[0212] in, w represents the i-th feature value after preprocessing (after normalization). i represents the weight corresponding to the feature, and b represents the bias term.

[0213] Output risk score:

[0214] The calculated surgical risk score is output to the system for doctors and patients to refer to.

[0215] 3.5 Interpretation and Application of Results

[0216] Interpretation of results:

[0217] Explain the surgical risk score, its significance, and its scope of application;

[0218] Based on their risk scores, patients are categorized into different risk levels, including low risk, medium risk, and high risk.

[0219] Clinical applications:

[0220] Doctors can develop personalized surgical plans and postoperative care programs based on surgical risk scores and the patient's specific condition.

[0221] The system can also trigger an early warning mechanism based on risk scores to remind doctors to pay attention to potential risk factors.

[0222] In summary, the risk assessment module achieves accurate assessment of patients' surgical risks through steps such as data preparation, feature extraction, model construction and optimization, surgical risk score calculation, and result interpretation and application. This helps doctors develop more reasonable surgical plans, improve surgical safety, and increase patient satisfaction.

[0223] 4. Early Warning Module

[0224] The early warning module sets a risk threshold T. When the surgical risk score calculated by the risk assessment module exceeds this threshold, an early warning signal is triggered. The early warning signal notifies the doctor and surgical team in the form of sound and flashing lights, reminding them to pay attention to potential risks. In addition, the early warning module also includes a time-sensitive early warning submodule, which dynamically adjusts the risk threshold T based on real-time information of the surgical progress to achieve more accurate early warning.

[0225] The specific implementation process is as follows:

[0226] 4.1 Setting the risk threshold T

[0227] Initial settings:

[0228] Based on historical surgical data and the performance of the risk assessment model, a reasonable risk threshold T is initially set.

[0229] This threshold T balances the accuracy and sensitivity of the warning, ensuring that the warning is triggered when the risk actually exists, while avoiding too many false alarms.

[0230] Dynamic adjustment:

[0231] As surgical data accumulates and risk assessment models are continuously optimized, the risk threshold T is dynamically adjusted.

[0232] By analyzing historical early warning data, surgical outcomes, and patient feedback, the risk threshold T is fine-tuned to improve the accuracy and practicality of early warnings.

[0233] 4.2 Triggering and Notification of Early Warning Signals

[0234] Risk score calculation:

[0235] The risk assessment module calculates a surgical risk score based on the patient's medical history, physiological parameters, and imaging data.

[0236] This score will serve as an important basis for determining whether a warning signal needs to be triggered.

[0237] Warning signal triggered:

[0238] When the surgical risk score exceeds the set risk threshold T, the warning module will trigger a warning signal;

[0239] Warning signals include various forms such as audible alarms and flashing lights, to attract the attention of doctors and surgical teams.

[0240] Notify the doctors and surgical team:

[0241] The early warning module sends early warning signals to doctors and surgical teams in real time through the system's internal communication mechanism;

[0242] Doctors and surgical teams can receive alerts through the system interface and mobile devices and take appropriate measures immediately.

[0243] 4.3 Implementation of the Time-Sensitive Early Warning Submodule

[0244] Real-time information on surgical progress:

[0245] The time-sensitive early warning submodule is connected to the monitoring equipment and sensors in the operating room to obtain real-time information on the surgical process.

[0246] This information includes the start time of the surgery, the stage of the surgery, and the patient's vital signs.

[0247] Dynamically adjust the risk threshold T:

[0248] Based on real-time information about the surgical progress, the time-sensitive early warning submodule dynamically adjusts the risk threshold T;

[0249] In critical stages of surgery or when a patient's vital signs become abnormal, the risk threshold T should be appropriately lowered to improve the sensitivity of early warning.

[0250] Conversely, during non-critical phases of surgery or when the patient's vital signs are stable, the risk threshold T should be appropriately increased to reduce the likelihood of false alarms.

[0251] Precise early warning:

[0252] By dynamically adjusting the risk threshold T, the time-sensitive early warning submodule can achieve more accurate early warnings;

[0253] This helps doctors and surgical teams to identify and address potential risks in a timely manner during surgery, thereby improving the safety and success rate of the procedure.

[0254] 4.4 Optimization and Iteration of the Early Warning Module

[0255] Data Analysis and Feedback:

[0256] The early warning module has data analysis capabilities and can perform statistical analysis on historical early warning data;

[0257] By analyzing the distribution, frequency, and accuracy of early warning data, the performance of the early warning module can be evaluated and potential problems can be identified.

[0258] User feedback collection:

[0259] As direct users of the early warning module, the feedback from doctors and surgical teams is crucial for optimizing the module.

[0260] Establish an effective feedback mechanism to encourage doctors and surgical teams to provide valuable opinions and suggestions when using the early warning module.

[0261] Iterative upgrades:

[0262] Based on data analysis results and user feedback, the early warning module has undergone iterative upgrades.

[0263] We continuously improve the accuracy and usability of the early warning module by optimizing algorithms, adding new features, or improving the user interface.

[0264] In summary, the early warning module achieves accurate early warning of surgical risks through steps such as setting a risk threshold T, triggering early warning signals, notifying doctors and surgical teams, and implementing time-sensitive early warning sub-modules. This helps doctors and surgical teams to identify and respond to potential risks in a timely manner during surgery, thereby improving the safety and success rate of the surgery.

[0265] 5. Display module

[0266] The display module provides a user-friendly interface for showcasing risk assessment results, warning information, and related patient data. This interface visually displays surgical risk scores and the contribution of each feature to the score in the form of charts and dashboards, helping doctors and surgical teams quickly understand the sources of risk. Simultaneously, the display module supports interactive data input, allowing doctors and surgical teams to enter additional patient information at any time before or during surgery to optimize risk assessment results. It also supports warning history recording, storing detailed information for each warning, including the time of triggering the warning, the reason, and the corresponding countermeasures, facilitating subsequent analysis and improvement.

[0267] The specific implementation process is as follows:

[0268] 5.1 Interface Design

[0269] User-friendliness:

[0270] Design a simple and clear user interface to ensure that doctors and surgical teams can quickly get started and accurately understand the information on the interface;

[0271] It uses intuitive icons, buttons, and a navigation bar to facilitate various operations for users.

[0272] Data visualization:

[0273] Visualization tools such as charts and dashboards are used to present information such as surgical risk scores and the contribution of each feature to the score in an intuitive way.

[0274] For example, bar charts can be used to show the contribution of different features to the surgical risk score, and dashboards can be used to show the real-time changes in the surgical risk score.

[0275] Interactivity:

[0276] Design interactive data input functionality to allow doctors and surgical teams to enter additional patient information at any time before or during surgery;

[0277] Provide clear input prompts and error feedback to ensure users can enter data correctly and optimize risk assessment results;

[0278] in,

[0279] The user interface includes a data input module for receiving patient information from doctors and the surgical team. This module has the following functions:

[0280] Multiple input methods: Supports multiple input methods such as text input, numeric input, and selection boxes to meet the input needs of different types of information;

[0281] Real-time validation: Perform real-time validation on user-input data to ensure its accuracy and integrity. For example, for numerical data, a reasonable range of values ​​can be set; for text data, specific formatting requirements can be set.

[0282] Automatic saving: Once the user finishes entering and submitting the data, it will be automatically saved to the system without the need for manual saving.

[0283] When a user enters additional patient information, the system needs to automatically integrate this information into the risk assessment model and update the risk assessment results. The specific implementation steps are as follows:

[0284] Data parsing and transformation: The system first parses the user-input data and converts it into a format that the risk assessment model can recognize;

[0285] Information integration: The parsed data is integrated into the patient's basic information database and merged with the original patient information;

[0286] Risk assessment model update: Based on the integrated patient information, the risk assessment model will be recalculated to obtain updated risk assessment results;

[0287] Results Display: The updated risk assessment results are displayed on the user interface for doctors and surgical teams to refer to.

[0288] 5.2 Data Display

[0289] Risk assessment results presentation:

[0290] The interface displays the surgical risk score calculated by the risk assessment module in real time.

[0291] It provides a detailed explanation and range of surgical risk scores to help doctors and surgical teams quickly understand the level of risk.

[0292] Feature contribution display:

[0293] It demonstrates the contribution of each feature to the surgical risk score, helping doctors and surgical teams identify potential risk factors;

[0294] Different colors or icons are used to represent the risk level or contribution level of different characteristics.

[0295] Patient data display:

[0296] It displays the patient's medical history, physiological parameters, and imaging data, providing comprehensive patient information support for doctors and surgical teams;

[0297] Ensure the accuracy and privacy protection of patient data, and avoid leaking sensitive information.

[0298] 5.3 Early Warning Information Display

[0299] Warning signal display:

[0300] When the surgical risk score calculated by the risk assessment module exceeds the set risk threshold, an early warning signal is triggered and displayed on the interface;

[0301] The warning signal is presented in the form of sound and flashing lights, along with clear text prompts and warning level classifications.

[0302] Warning history records:

[0303] Record detailed information for each warning, including the time the warning was triggered, the reason for it, and the response measures taken.

[0304] It provides functions for querying and exporting warning history records, making it convenient for doctors and surgical teams to conduct subsequent analysis and improvements.

[0305] 5.4 Interactive Data Input Function

[0306] Data input interface:

[0307] The system features a clear data input interface that allows doctors and surgical teams to enter additional patient information at any time before or during surgery.

[0308] Provide necessary data validation and error message functions to ensure the accuracy and completeness of input data.

[0309] Data update and synchronization:

[0310] Real-time updates and synchronization of patient information ensure that the risk assessment module can perform calculations and issue warnings based on the latest data;

[0311] Provides data update prompts and confirmation functions to avoid data inconsistencies caused by accidental operations.

[0312] 5.5 System Optimization and Iteration

[0313] User feedback collection:

[0314] Establish an effective user feedback mechanism to encourage doctors and surgical teams to provide valuable opinions and suggestions when using the display module;

[0315] Regularly analyze and summarize user feedback to provide guidance for system optimization and iteration.

[0316] System performance monitoring:

[0317] Monitor the operating performance and stability of the display module, and promptly identify and address potential problems;

[0318] Based on system performance monitoring results and user feedback, the display module is continuously optimized and upgraded.

[0319] In summary, the display module provides doctors and surgical teams with comprehensive risk assessment and early warning support through intuitive data visualization, interactive data input functions, and early warning history. This helps them quickly understand the sources of risk, optimize risk assessment results, and take corresponding countermeasures, thereby improving the safety and success rate of surgery.

[0320] 6. Model Update Module

[0321] The model update module is responsible for regularly collecting new surgical case data and updating the risk assessment model through incremental learning or retraining. This module also includes a safety verification submodule, which is used to verify the safety and reliability of the model by simulating surgical scenarios before deploying a new model or updating an existing model. The safety verification submodule generates a safety analysis report, pointing out potential limitations, false positive rates, false negative rates, and other issues of the model, and proposes improvement measures.

[0322] The specific implementation process is as follows:

[0323] 6.1 Data Collection and Preprocessing

[0324] Collect new data regularly:

[0325] We regularly collect new surgical case data from sources such as hospital information systems and surgical record systems.

[0326] Ensure that the collected data includes key information such as the patient's medical history, surgical procedure, and surgical outcome.

[0327] Data preprocessing:

[0328] The newly collected data undergoes preprocessing steps such as cleaning, deduplication, and format conversion to ensure data quality and consistency.

[0329] Extract features related to risk assessment, such as patient age, type of surgery, and operation time, and construct feature vectors.

[0330] 6.2 Model Update

[0331] Incremental learning:

[0332] If the amount of newly collected data is small, or if you want to maintain the stability of the model, you can use incremental learning to update the model.

[0333] New data is added to the existing training dataset, and the model is fine-tuned using an incremental learning algorithm to adapt to the new data distribution.

[0334] Retraining:

[0335] If the amount of newly collected data is large, or if you want the model to learn new features or patterns, you can consider retraining the model.

[0336] The model was retrained using a new training dataset, and its performance was re-evaluated.

[0337] 6.3 Security Verification

[0338] Simulated surgical scenario:

[0339] The safety verification submodule uses simulated surgical scenarios to verify the safety and reliability of new or updated models;

[0340] The simulation scenarios should cover different types of surgeries and patients with different risk levels to ensure the comprehensiveness of the validation.

[0341] The specific methods are as follows:

[0342] 6.3.1 Constructing a database of simulated surgical scenarios

[0343] Surgical data collection: Collect a large amount of thoracic surgery data from major hospitals and medical institutions, including surgery type, patient information, surgical procedure records, and postoperative complications;

[0344] Classification and labeling: The collected surgical data were classified and labeled according to surgical type (lobectomy, radical resection of lung cancer) and risk level (low risk, medium risk, high risk);

[0345] Constructing simulated scenarios: Based on classified and labeled data, multiple simulated surgical scenarios are constructed using computer simulation technology. Each scenario includes a specific surgical type, patient information, and surgical procedure.

[0346] 6.3.2 Simulated Surgical Procedure and Risk Assessment

[0347] Input simulation data: Input data from simulated surgical scenarios into the intelligent thoracic surgery risk assessment and early warning system;

[0348] Running the risk assessment model: The system runs the risk assessment model based on the input data, calculates the surgical risk score, and generates corresponding risk warning information;

[0349] Record assessment results: Record the risk assessment results and early warning information for subsequent analysis and comparison.

[0350] 6.3.3 Verify the safety and reliability of the model

[0351] Comparing simulation results with actual situations: The risk assessment results of the simulated surgical scenario are compared with the actual situation (postoperative complication rate, surgical success rate) to analyze the accuracy and reliability of the model;

[0352] Adjusting model parameters: Based on the comparison results, the parameters of the risk assessment model are adjusted and optimized to improve the accuracy and reliability of the model;

[0353] Repeated validation: The adjusted model is applied again to simulated surgical scenarios for validation to ensure that the safety and reliability of the model are fully verified.

[0354] By employing the above methods, it can be ensured that the intelligent thoracic surgery risk assessment and early warning system has undergone comprehensive and rigorous simulation of surgical scenarios before being applied to actual surgeries, thereby improving the system's safety and reliability.

[0355] Security Analysis Report:

[0356] After simulating a surgical scenario, a safety analysis report is generated, pointing out potential limitations, false positive rates, and false negative rates of the model.

[0357] The analysis report should record in detail all aspects of the simulation process, including the model's predictions, the actual surgical results, and the differences between the two.

[0358] Improvement measures:

[0359] Based on the issues and recommendations in the security analysis report, the model was optimized and improved.

[0360] For example, adjusting model parameters, adding new features, or improving feature extraction methods can enhance the model's accuracy and reliability.

[0361] 6.4 Model Deployment and Monitoring

[0362] Model Deployment:

[0363] After security verification and confirmation that the model performance meets the requirements, the new or updated model will be deployed to the actual production environment.

[0364] Ensure a smooth and orderly model deployment process to avoid unnecessary impact on existing systems.

[0365] Model monitoring:

[0366] Continuously monitor the deployed model and track its performance in real-world applications;

[0367] If a decline in model performance or other problems are found, timely measures should be taken to adjust and optimize the model.

[0368] 6.5 Continuous Iteration and Optimization

[0369] Collect feedback:

[0370] Establish an effective feedback mechanism to collect opinions and suggestions from doctors and surgical teams regarding the use of the model;

[0371] Regularly analyze and summarize feedback to guide the continuous iteration and optimization of the model.

[0372] Technology Update:

[0373] Pay attention to the latest technological advancements and research findings in the field of artificial intelligence, and promptly apply new technologies to the models;

[0374] Through continuous technological updates and iterations, we improve the accuracy and reliability of the model, providing patients with better surgical risk assessment and early warning services.

[0375] In summary, the model update module ensures that the model can adapt to the ever-changing surgical environment and patient needs by regularly collecting new data, updating the risk assessment model, conducting safety verification, and continuously iterating and optimizing. This helps to improve the safety and success rate of surgery and reduce medical risks.

[0376] III. System Implementation Results

[0377] 1. Accurate risk assessment

[0378] Reduce the incidence of surgical complications:

[0379] The system utilizes advanced algorithms and models to comprehensively consider various factors such as the patient's medical history, physiological parameters, type and difficulty of surgery, and conducts accurate risk assessment.

[0380] By identifying potentially high-risk patients and surgical procedures in advance, doctors can take targeted preventative measures to effectively reduce the incidence of surgical complications such as infection, bleeding, and organ damage.

[0381] Improve surgical safety:

[0382] The system provides real-time risk assessment results and early warning information before, during, and after surgery, helping doctors to promptly identify and handle abnormal situations.

[0383] This helps doctors make more informed decisions during surgery, ensuring the safety of the procedure and reducing surgical risks and uncertainties.

[0384] 2. Real-time early warning and response

[0385] Real-time alert function:

[0386] The system can monitor the patient's physiological parameters and surgical process data in real time, and immediately trigger an early warning signal once an abnormality is detected;

[0387] The warning signals are displayed intuitively on the interface, along with detailed warning information and suggested countermeasures to help doctors react quickly.

[0388] Improve emergency response speed:

[0389] Because the system provides real-time early warning information and response suggestions, doctors can identify and handle emergencies more quickly;

[0390] This helps reduce delays and errors during surgery, improves emergency response speed, and enhances overall medical quality.

[0391] 3. Improve patient satisfaction

[0392] Reduce patient anxiety:

[0393] The system provides detailed risk assessments and early warning information, enabling patients and their families to better understand the risks and potential complications of surgery.

[0394] This helps reduce patients' anxiety and fear, and increases their confidence and satisfaction with the surgery.

[0395] Optimize postoperative care:

[0396] The system continues to provide risk assessment and early warning services after surgery, helping doctors develop more personalized postoperative care plans;

[0397] This helps reduce the occurrence of postoperative complications and improves the speed of patient recovery and satisfaction.

[0398] 4. Improve the work efficiency of doctors and surgical teams

[0399] Interactive interface:

[0400] The system provides a user-friendly interactive interface, allowing doctors and surgical teams to easily view risk assessment results, warning information, and related patient data;

[0401] This helps doctors understand patients' conditions and surgical risks more quickly, improving the efficiency and accuracy of decision-making.

[0402] Model update mechanism:

[0403] The system regularly collects new surgical case data and updates the risk assessment model through incremental learning or retraining.

[0404] This ensures that the model can adapt to changing surgical environments and patient needs, improving the accuracy and reliability of risk assessment;

[0405] At the same time, the model update mechanism also reduces the workload of doctors and surgical teams in manually updating the model, improving their work efficiency.

[0406] 5. Provides strong support for risk management in thoracic surgery.

[0407] Comprehensive risk management:

[0408] The system provides comprehensive support for risk management in thoracic surgery through multiple aspects, including risk assessment, early warning, interactive interface, and model update mechanism.

[0409] This helps hospitals and surgical teams better identify, assess, and control surgical risks, thereby improving the quality and safety of medical care.

[0410] Promote the improvement of medical standards:

[0411] The implementation and application of the system have promoted the improvement and progress of medical standards;

[0412] By continuously optimizing and improving risk assessment models and early warning mechanisms, doctors and surgical teams can continuously learn and master new medical technologies and methods, thereby improving the overall level of medical care.

[0413] In conclusion, the intelligent thoracic surgery risk assessment and early warning system has achieved significant results in practical applications. It has not only reduced the incidence of surgical complications, improved surgical safety and patient satisfaction, but also increased the work efficiency of doctors and surgical teams, providing strong support for risk management in thoracic surgery. This will contribute to the continued development and progress of the field of thoracic surgery.

[0414] Specific application examples:

[0415] I. System Overview

[0416] The intelligent thoracic surgery risk assessment and early warning system collects and analyzes individualized patient data and uses advanced machine learning algorithms to provide accurate risk assessment results and early warning information for thoracic surgery. The system can automatically identify the patient's risk factors and give corresponding risk scores, helping doctors and surgical teams make more informed decisions before surgery.

[0417] II. Specific Applications

[0418] 1. Patient Information

[0419] Suppose there is a patient about to undergo thoracic surgery, whose pre-treatment characteristics and related information are as follows:

[0420] Feature 1 (Age): 60 years old (standardized value x1 = 0.6, assuming the corresponding value for 60 years old in the standardization process is 0.6);

[0421] Feature 2 (Gender): Male (converted to a numerical feature, x2 = 1, assuming male is 1 and female is 0);

[0422] Feature 3 (ASA score): Level III (converted to a numerical feature, x3 = 2, assuming Level I is 0, Level II is 1, and Level III is 2);

[0423] Feature 4 (behavioral status): Requires partial assistance (converted to a numerical feature, x4 = 2, assuming complete self-care is 0, requiring partial assistance is 2, and complete dependence on others is 4);

[0424] Feature 5 (Dyspnea score): 2 points (x5 = 2);

[0425] Feature 6 (Surgical Timing): Elective surgery (converted to a numerical feature, x6 = 0, assuming emergency surgery is 1 and elective surgery is 0);

[0426] Feature 7 (surgical method): Lobectomy (converted to a numerical feature, x7 = 1, assuming other surgical methods are 0 and lobectomy is 1);

[0427] Feature 8 (pathological result): Malignant (converted to a numerical feature, x8 = 1, assuming benign is 0 and malignant is 1);

[0428] Feature 9 (number of comorbidities): 2 (x9 = 2).

[0429] 2. Weights and Bias Terms

[0430] Suppose that the following weights w have been obtained through a supervised learning algorithm (such as logistic regression) on the training dataset. i And the value of the bias term b:

[0431] w1=0.3, w2=0.1, w3=0.2, w4=0.15, w5=0.1, w6=-0.05, w7=0.05, w8=0.2, w9=0.15;

[0432] b = -0.5.

[0433] 3. Surgical risk score calculation

[0434] According to the surgical risk scoring formula:

[0435]

[0436] We substitute the above eigenvalues ​​and weights into the formula to calculate:

[0437] Risk Score=(0.3*0.6)+(0.1*1)+(0.2*2)+(0.15*2)+(0.1*2)+(-0.05*0)+(0.05*1)+(0.2*1)+(0.15*2)-0.5

[0438] Risk Score=0.18+0.1+0.4+0.3+0.2+0+0.05+0.2+0.3-0.5

[0439] Risk Score = 1.83 - 0.5

[0440] Risk Score = 1.33

[0441] 4. Risk Interpretation and Early Warning

[0442] The calculated surgical risk score can be compared with a preset risk threshold. Assuming the risk threshold is set to 1.0, the patient's surgical risk score is 1.33, which exceeds the risk threshold, so the system will trigger an alert.

[0443] The warning information may include: the patient has a high surgical risk, and it is recommended that the doctor and surgical team conduct a more detailed assessment and preparation before the operation to ensure the safety and success rate of the operation; at the same time, the system can also provide specific risk factor prompts, such as advanced age, high ASA score, malignant pathology results, etc., to help doctors and surgical teams more accurately identify and manage risk factors.

[0444] III. Summary

[0445] Through the specific application examples above, we can see that the intelligent thoracic surgery risk assessment and early warning system can utilize advanced machine learning algorithms and personalized data to provide accurate risk assessment results and early warning information for thoracic surgery. This not only helps doctors and surgical teams make more informed decisions before surgery but also improves the safety and success rate of surgery, providing better medical services for patients.

[0446] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. An intelligent thoracic surgery risk assessment and early warning system, characterized in that, include: a) Data acquisition module: used to collect patients' medical history, physiological parameters and preoperative imaging data; b) Data preprocessing module: Cleans and standardizes the data collected by the data acquisition module to eliminate outliers and ensure data consistency; c) Risk assessment module: A risk assessment model is built using machine learning algorithms; d) Early warning module: Set a risk threshold T. When the surgical risk score calculated by the risk assessment module exceeds the threshold T, an early warning signal is triggered. e) Display module: Used to display risk assessment results, early warning information, and related patient data; f) Model update module: Regularly collect new surgical case data and update the risk assessment model through incremental learning or retraining to improve the accuracy of assessment.

2. The intelligent thoracic surgery risk assessment and early warning system according to claim 1, characterized in that: The risk assessment model calculates the surgical risk score based on the following formula: Where, x i w represents the i-th feature value after preprocessing. i The weight w represents the weight corresponding to the feature, b represents the bias term, and n represents the total number of features; i The bias term b is obtained by optimization using a supervised learning algorithm on the training dataset, wherein the supervised learning algorithm is logistic regression, random forest or support vector machine.

3. The intelligent thoracic surgery risk assessment and early warning system according to claim 1, characterized in that: The machine learning algorithm uses cross-validation to evaluate model performance during the training phase and optimizes model parameters based on accuracy, recall, and F1 score metrics.

4. The intelligent thoracic surgery risk assessment and early warning system according to claim 1, characterized in that: The risk assessment module further includes a deep learning sub-module, which specifically performs the following steps: Image data preprocessing: Preprocessing of preoperative imaging data, including image enhancement, noise reduction, and normalization. Feature extraction: A convolutional neural network structure is used to automatically extract key features from the image through multiple convolutional layers, pooling layers and fully connected layers. Key features include tumor size, location and the clarity of the boundary with surrounding tissues. Classification and scoring: The extracted features are input into the classifier and combined with non-image features to calculate the surgical risk score.

5. The intelligent thoracic surgery risk assessment and early warning system according to claim 1, characterized in that: The early warning module also includes a time-sensitive early warning submodule, which is specifically composed of the following units: Surgical stage identification unit: Identifies the current stage of the surgery by monitoring signals inside and outside the operating room in real time; Dynamic threshold adjustment: The risk threshold T is dynamically adjusted according to different surgical stages. During critical steps or high-risk stages of surgery, the threshold is lowered to improve early warning sensitivity. Optimized warning signals: Different warning signals are used according to the urgency of the warning, including sound, light color, and screen flashing, to intuitively and quickly notify doctors and surgical teams.

6. The intelligent thoracic surgery risk assessment and early warning system according to any one of claims 1 to 5, characterized in that: The display module also includes the following functional units: Interactive data input unit: Provides a user interface that allows doctors and surgical teams to enter additional patient information at any time before or during surgery, including laboratory test results and intraoperative findings. The additional patient information will be automatically integrated into the risk assessment model. Risk assessment results visualization unit: The risk assessment results are displayed intuitively in the form of charts and dashboards, including surgical risk scores and the contribution of each feature to the score, to help doctors and surgical teams quickly understand the sources of risk; Warning History Record Unit: Records detailed information for each warning, including the time the warning was triggered, the reason, and the response measures taken, to facilitate subsequent analysis and improvement.

7. The intelligent thoracic surgery risk assessment and early warning system according to claim 6, characterized in that: The system also includes a security verification module, which specifically performs the following steps: Simulated surgical scenario construction: Based on historical surgical data and expert knowledge, a series of simulated surgical scenarios are constructed, including different surgical types, patient characteristics, and surgical risk levels; Model performance testing: Run the risk assessment model in a simulated surgical scenario and record the model's prediction accuracy, stability, and other indicators; Security Analysis Report: Generate a security analysis report that points out potential limitations of the model, false positive rate, false negative rate, and proposes improvement measures; Model Iteration and Optimization: Based on the safety analysis report, the model is adjusted and optimized to improve its safety and reliability in actual surgery.

Images