Research And Application Of Breast Cancer Prognosis Model Based On Improved Bayesian Network

In the information age,big data and artificial intelligence have greatly promoted the level of social medical care.From the slow blockage of SARS in 2003 to the rapid discovery of new coronary pneumonia virus infection routes and comprehensive treatment methods,information technology has been used in disease detection,diagnosis and treatment,disease model construction and Forecasting and other aspects have played an irreplaceable role.Cancer is a disease that humans must overcome in addition to infectious diseases.In recent years,the analysis of high-risk pathogenic factors for breast cancer has been a research hotspot.Combined with the massive clinical data accumulated in breast cancer,a new big data analysis algorithm is used to build a prognostic model to explore the high-risk pathogenic factors leading to breast cancer.This paper introduces Bayesian network for model construction,and uses genetic algorithm for model structure optimization and improvement,which can more accurately establish the optimal network directed graph in structure learning,so as to better explore the prognostic factors of breast cancer at high risk.The main research of this paper The work is as follows:First,the technical problems of the existing Bayesian network in the construction of breast cancer model are studied,and the improvement methods are sought.Based on the analysis of the advantages of the Logistic regression analysis model and genetic algorithm,the basic methods of Bayesian network model in structure learning and parameter learning are compared and analyzed,and several types of algorithm models that may be cited in the improvement of structure learning are discussed.Then the adaptive search scoring function genetic algorithm is introduced into the network topology graph construction part of Bayesian network model structure learning,and the Bayesian network breast cancer prognosis model based on genetic optimization is researched and proposed.Optimized and improved the traditional Bayesian network model algorithm to solve the shortcomings of the traditional model structure,by dynamically changing the model's population size,iterations and other key parameters to better fit the clinical data of breast cancer.In order to verify the above method,the paper used Rstudiosoftware to process the existing medical SEER database,screened out 28,000 breast cancer patient records,and compared the results of the algorithm model proposed in the paper with the common machine learning models in the existing literature methods.The performance of the method proposed in the paper is discussed.Finally,based on the breast cancer prognosis model method constructed in the paper,a breast cancer early warning system that can be used for clinical breast cancer pre-adjuvant detection was developed.According to the in-depth analysis of breast cancer early warning needs of related departments of the hospital,the software development back-end framework and database construction are optimized.Through the design of the front-end interactive interface of the early warning platform,clinicians are provided with informational methods and means for breast cancer adjuvant treatment.