Research And Implementation Of Feature Selection Algorithm Based On Risk Factor Of Cancer

Topic on this subject comes from the Ministry of health deployed (management)hospital clinical subjects focus projects <<Research and Application of PrognosticRisk Assessment and Early Warning in the Cancer>>. It is our Grid Computing andInternet Security Lab and Bethune first hospital of Jilin University’s interdisciplinarycooperation projects between the different areas. Project is aimed at using theadvanced computer and information technology, basic information about the use ofalready sick patients, genetic statistics, and daily habits, physiological conditions tobuild the mathematical modeling of such statistical data and make it useful, so as torealize on random test on the crowd at a specific tumor onset probability prediction ofpossibility. Through a series of programs to implement, we expect to an effectivewarning of high-risk people with cause of the disease and the purpose of relativeimprovement in our country’s high levels of tumor incidence of disease in developingcountries, and no breakthroughs in cancer early-warning technology developmentstatus.Repeated negotiations and cooperation on the basis of the clear core meaning andgoal of the project, by extensively researching medical history and development in theareas at home and abroad, I select the topic of research interest under the featuredimension reduction method based on risk factors for cancer research andimplementation of algorithm for feature selection.In the1st chapter, the paper states the background of the research and thesignificance on the subject, and gives you the full organizational structure. In the2ndchapter it describes in detail the concepts of medical data, both at home and abroaddevelopment situation and main problems in current, from the angle of methodologyoverviews the concept of feature dimension reduction, classification standards and implementation strategies, models and other background knowledge. It describesfeature dimension reduction two main branches: feature extraction and featureselection. Setting out a framework for feature selection method, the defining feature ofselecting four main steps: feature subset your build process, evaluation criteria, stopthe validation process criteria for feature subset. In the3rd chapter, it describes thealgorithm used by the relevant background knowledge. First, it presents the basicprinciples and basic steps in statistical hypothesis testing methods, comprehension andmastery in the use of hypothesis testing theory you need to be aware of problems.Further it gives the core ideas of t-test algorithm under hypothesis testing algorithm,introduces the t-test algorithms under different scenarios of different forms ofoperation and rationalization. For this problem to be solved I make the choice andimprovement. In the experimental part of the4th chapter, it completely introduces theprocess of experimental data collection, aggregation and consolidation effort, clearlyspeaks out experimental design theory and frame model, gives the algorithm sourcecode, and combined with artificial and computer-two different angles gives the featureselection algorithm for analysis of the results of the implementation and verification.Future work of the5th chapter will be further in-depth studying from theviewpoint of the following three areas:(1) transforming the problem of the specificareas to improve the of the t-test feature selection algorithm;(2) base on the existingproblem, analyzing the complex degrees of two kinds of angles on time complexityand space complexity of optimization algorithm;(3) to complete the process of featureselection based on cancer risk factors of adaptive way fitting for mathematicalmodeling, besides from the Angle of model we will study deeply.