Design And Implementation Of Tumor Early Warning System Based On Probability SVM

According to the global monitoring data of the Cancer Research Center ofWHO (World Health Organization), by2030the world will have26million newcancer cases and17million deaths cases, most of which will occur in low-incomedeveloping countries. Although the exact cause of the tumor is not very clear, but badhabits influence tumor that induced by cell gene mutation. If we can record the detailsof life of cancer patients, then we may construct mathematical model based onstatistical data so as to predict the probability of illness of a random population.According to the result of the probability of illness, warning the crowd with highprobability of illness to improve their living habits for preventing the tumor, which hasa very important practical significance in exploring the new direction of the tumorprevention.As a tool to solve the problem of machine learning by means of optimizationmethods, SVM is mainly used in the statistical classification and regression analysis.And it has a very wide range of applications. In the medical field, SVM is frequentlyused as an adjunct to diagnose the tumor, such as the classification of X-ray image ofthe breast cancer using SVM. The diagnosis of cancer in most cases depends on acomplex combination of clinical and histopathological data. Because of thiscomplexity, there exists a significant amount of interest among clinical professionalsand researchers regarding the efficient and accurate prediction of breast cancers. SoAustin H Chen et al developed a breast cancer prognosis predict system (BCPP) thatcan assist medical professionals in predicting breast cancer prognosis status based onthe clinical data of patients. A complete computer system can not only supplement theprocessing of data, improve the efficiency of data processing, the main thing is to reduce the human operator errors, and thus contribute to the accuracy of the diagnosisof tumor.This article is based on this thinking, through the probability SVM and featureselection with the factors of lifestyle, we design and implement the tumor earlywarning system.In order to design the tumor early warning system, we do the analysis offunctional requirements, and then get the logical framework of the system, in the endwe design and implement the training modules, the prediction module, the adjustmentmodule, as well as the feature selection function. In the experimental part of the paper,we process the experimental data firstly, and through the preliminary experiments, onthe one hand, we verify the availability of the basic functions of the the tumor earlywarning system, on the other hand, we verify the practicality of classification of SVMin the field of tumor warning. In other words, the experiments verify the feasibility ofpredicting whether the patient is suffering from a tumor, through the statistics oflifestyle of cancer patients, such as smoking, diet and inheritance. Although the type ofdata used is different from the BCPP’s, but the present system introduce a dataadjustment module, a function of posterior probability estimation and a featureselection module. Compared with BCPP, it obtains a good prediction accuracy on thesame results. After feature selection, although the accuracy of the forecast declines,since the convenience of feature selection, it can still be used as an option.