Risk Assessment And Time Frame Prediction Of Surgical Complication

The occurrence of surgical complications can affect the quality of medical services and threaten the patient safety.Using scientific methods to assess the risk of surgical complications and predict their time frames are of great significance for the rational arrangement of patients’ postoperative care and monitoring,thus improving the quality of medical services and ensuring the patient safety.At present,the researches on the risk analysis of surgical complications are mostly based on statistical methods,which is lack of comprehensive risk factor identification.Other researches using reliability analysis haven’t take the state transition of some events into consideration.In addition,the researches on the time frames of postoperative complications mainly depend on the judgment of doctors,which is influenced by multiple factors such as accumulation of experience and skills.As a result,it is difficult to guarantee the accuracy of judgment.To overcome these problems,some scholars have conducted quantitative analysis based on historical data and achieved time frames of some surgical complications.However,since the patients are not classified according to their heterogeneity,the outcomes of time frames are generally oriented to the overall patients,not individuals,which might lead to bias when applying to individual patients.Therefore,this thesis applies FTA to model and analyze the risk of complications,and utilizes the probabilistic neural network to establish the prediction model for computing the time frames of surgical complications.Specifically,in the section of the modeling and analysis of the risk of complications,firstly,the risk factors that affect the specific complications has been identified based on the multi-criteria decision-making method.Secondly,based on the diseases and operation types,Fisher discriminant model of patient heterogeneity has been constructed to classify the patients.Thirdly,the fault tree model for the specific surgical complication has been established,and the probabilities of basic events with state transitions have been adjusted through Markov method.Finally,the more accurate probabilities of complications can be calculated in different categories of patients.In the section of time frame prediction of complications,firstly,the sample data of patient heterogeneity with surgical complications has been preprocessed to get the eigenvector.Secondly,the probabilistic neural network topology has been built according to the sample data,so as to find the mapping between the patient heterogeneity and time frame of complication.Thirdly,through the sample training and effectiveness testing,the accuracy of the prediction model has been proved.This thesis applies the proposed model to analyze the risk of incision infection in laparoscopic acute appendectomy and predict the time frame of its occurrence.The result shows that the model can be applied to personalized assessment on risk analysis and time frame prediction according to the patient heterogeneity.Compared with the previous work,the proposed model in this thesis is more pertinent and practical,which verifies the feasibility and effectiveness of the model.