Risk prediction method and system for AI preoperative simulation operation

Abstract

The invention relates to the technical field of preoperative risk prediction, in particular to a risk prediction method and system for an AI preoperative simulation operation. The method comprises the following steps: acquiring multi-modal medical data of a patient, and performing quantitative analysis to obtain structural characteristics of a pathological part of the patient, a genetic marker of the patient and phenotypic characteristics of the patient; carrying out multi-modal deep feature fusion based on the structural features of the pathological part of the patient, the genetic marker of the patient and the phenotypic features, and constructing a personalized patient digital model; a patient operation log is obtained, multi-stage operation step analysis and deep risk factor prediction are carried out, and patient personalized risk factors are generated; self-adaptive operation path planning and step-by-step sudden execution prediction are carried out based on patient operation logs, and risk prediction points of different operation paths are obtained. According to the invention, through efficient and accurate preoperative risk prediction, the safety of the operation is visually displayed.

Claims

1. A risk prediction method for AI preoperative simulation surgery, characterized in that: The following steps are involved: Step S1: Acquire multimodal medical data of the patient and perform quantitative analysis to obtain the structural characteristics of the patient's pathological site, the patient's genetic markers, and phenotypic characteristics; Step S2: Perform multimodal deep feature fusion based on the patient's pathological site structural characteristics, patient genetic markers, and phenotypic characteristics to construct a personalized patient digital model; Step S3: Obtain the patient's surgical log, perform multi-stage surgical step analysis and deep risk factor prediction, and generate personalized risk factors for the patient; Step S4: performing adaptive surgical path planning and step-by-step execution prediction based on the patient's surgical log to obtain risk prediction points for different surgical paths; Step S5: Based on the risk prediction points of different surgical pathways and the patient's personalized risk factors, multiple rounds of scenario deductions are conducted, and the effectiveness of emergency measures is evaluated to obtain the effectiveness evaluation value of emergency measures for each emergency situation; Step S6: Optimize the emergency plan decision based on the effective evaluation value of the emergency measures for each emergency situation, and then conduct a preoperative rehearsal evaluation to obtain a preoperative rehearsal risk assessment result.

2. The risk prediction method for AI preoperative simulation surgery according to claim 1, characterized in that: The specific steps of step S1 are: Acquiring multimodal medical data of the patient, wherein the multimodal medical data of the patient includes medical images, genomic information, clinical history data, and vital signs data of the patient; Perform image detail enhancement processing on patient medical images and extract details to enhance medical images; Perform anatomical structure convolution analysis on detail-enhanced medical images to generate anatomical structure convolution features; Perform pathological quantitative analysis on the convolution features of the anatomical structure to obtain the structural characteristics of the patient's pathological site; Deep graph neural network mining of genomic information is performed to extract patient genetic markers and phenotypic characteristics.

3. The risk prediction method for AI preoperative simulation surgery according to claim 2, characterized in that: The specific steps of performing deep graph neural network mining on genomic information to extract genetic markers and phenotypic characteristics of patients are as follows: Perform quality control comparison of genomic information based on a preset reference genome table to identify biased genomic information; Analyze the variation position, type and conservation information of the deviated genomic information to obtain the gene variation characteristics; Identifying the variation type of the gene variation signature; Perform gene variation annotation marking according to the variation type to obtain a variation annotation genome table; Perform graph neural network modeling on the variant annotation genome table to obtain the variant genome graph structure; Deep graph neural network mining is performed on the variant genome graph structure, and forward propagation analysis is performed to extract patient genetic markers and phenotypic characteristics.

4. The risk prediction method for AI preoperative simulation surgery according to claim 1, characterized in that: The specific steps of step S2 are: Identify and mark abnormal noise points in clinical history data and vital signs data; Adaptive filtering and noise reduction processing is performed based on abnormal noise points to obtain noise-reduced and optimized clinical history data and noise-reduced and optimized vital sign data; Performing natural language semantic recognition on the noise-reduced and optimized clinical history data to extract historical symptom characteristics of the patient; Perform real-time status analysis on noise-reduced and optimized vital sign data to obtain the patient's real-time vital status; Based on the patient's real-time vital status, historical symptom characteristics, structural characteristics of the patient's pathological site, genetic markers and phenotypic characteristics, multimodal deep feature fusion is performed to construct a personalized patient digital model.

5. The risk prediction method for AI preoperative simulation surgery according to claim 1, characterized in that: The specific steps of step S3 are: Obtain the patient's surgical log; identify the patient's surgical type and surgical difficulty value based on the patient's surgical log; Predict the operation duration based on the patient's operation log to obtain the predicted operation duration value; Perform multi-stage surgical step analysis based on the patient's surgery type and surgical difficulty value to obtain the surgical execution steps for multiple stages; Conduct in-depth risk factor prediction on the surgical execution steps and surgical duration prediction values ​​at multiple stages to obtain multiple risk prediction factors; Perform end-to-end iterative learning on multiple risk predictors to generate patient-personalized risk factors.

6. The risk prediction method for AI preoperative simulation surgery according to claim 1, characterized in that: The specific steps of step S4 are: Identify pending surgical targets based on the patient's surgical log; Conduct deep learning of anatomical structure features on personalized patient digital models to identify topological correlation features between anatomical structures; Adaptive surgical path planning is performed based on the topological correlation features between the surgical target and the anatomical structure to generate multiple surgical planning paths; Based on multiple surgical planning paths, the surgical execution steps of multiple stages are predicted step by step to obtain risk prediction points for different surgical paths.

7. The risk prediction method for AI preoperative simulation surgery according to claim 1, characterized in that: The specific steps of step S5 are: Conduct multiple rounds of scenario simulations based on risk prediction points of different surgical pathways and individual patient risk factors to generate simulation data for different surgical pathways. Perform multi-condition risk prediction on simulation data of different surgical pathways to obtain multi-condition risk prediction data; Perform multi-scenario emergency probability prediction on multi-condition risk prediction data to obtain the emergency probability of different surgical pathways; The effectiveness of emergency measures is evaluated according to the probability of emergencies in different surgical pathways, and the effectiveness evaluation value of emergency measures for each emergency situation is obtained.

8. The risk prediction method for AI preoperative simulation surgery according to claim 1, characterized in that: The specific steps of step S6 are: Comprehensive surgical safety risk estimation is performed based on the effective evaluation value of emergency measures for each emergency situation to obtain the surgical safety risk estimation value; Optimize emergency plan decisions based on surgical safety risk estimates and risk prediction points of different surgical pathways, and generate multi-condition emergency optimization strategies; Conduct preoperative rehearsal evaluation on multi-condition emergency optimization strategies and obtain preoperative rehearsal risk assessment results.

9. An AI pre-operative simulation surgery risk prediction system, characterized in that: The risk prediction method for performing AI preoperative simulation surgery according to claim 1 comprises: The quantitative analysis module is used to obtain multimodal medical data of patients and perform quantitative analysis to obtain the structural characteristics of the patient's pathological site, genetic markers and phenotypic characteristics; A deep feature fusion module is used to perform multimodal deep feature fusion based on the structural characteristics of the patient's pathological site, genetic markers, and phenotypic characteristics to build a personalized patient digital model; The risk factor prediction module is used to obtain patient surgical logs, perform multi-stage surgical step analysis and in-depth risk factor prediction, and generate personalized risk factors for patients; The risk prediction module is used to perform adaptive surgical path planning and step-by-step execution prediction based on the patient's surgical log to obtain risk prediction points for different surgical paths; The multi-round scenario deduction module is used to conduct multiple rounds of scenario deduction based on the risk prediction points of different surgical pathways and the patient's personalized risk factors, and to evaluate the effectiveness of emergency measures to obtain the effectiveness evaluation value of emergency measures for each emergency situation; The preoperative risk assessment module is used to optimize emergency plan decisions based on the effective evaluation value of emergency measures for each emergency situation, and then conduct preoperative rehearsal evaluation to obtain preoperative rehearsal risk assessment results.

Images