Ovarian Tumor Forecast And Analysis Research Based On Machine Learning

Since the 21 st century, with rapid development of information technology, the computer plays an increasingly important role in our life. With remarkable advances of hospital informationization, application of electronic medical records and data storage technology, hospital database has accumulated a large amount of data. These precious medical data for prediction to disease, diagnosis, treatment and medical science research are valuable. However, the majority of hospitals are just at the stage of the "add, delete, change, check" for data processing at present. Owing to lack of data integration and analysis technology, it is insurmountable obstacle for data supporting medical decisions and automatic acquisition of knowledge. On the other hand, in the face of the massive amount of data, traditional data analysis and processing methods have been unable to obtain the hidden information and the inherent connections between data. The problem is that now we have significant advancement over data collection, data storage technology, whereas how to utilize hard-won data process is our main obstacle we have to circumvent.In this paper, some theories of data mining are employed the on the collected sample medical data for the evaluation of ovarian tumor screening and preprocessing. After studying relevant data mining methods, four classifiers are selected for medical data mining(Support Vector Machine classifier, Naive Bayesian classifier, Nearest Neighbor classifier, Random Forest algorithm classifier). Support vector machine(SVM) is fit for small sample, nonlinear and high dimensional pattern recognition; meanwhile it can be applied to the function fitting and other machine learning problems. Naive bayes model requires only a few estimates of parameters and less sensitive to missing data, which has a solid mathematical foundation, stability classification effect. Nearest neighbor classifier prefer to the class field cross or overlap more stay points sample set. Random forest algorithm can produce high accuracy of classifier and suitable for processing a large number of input variables, simultaneously the learning process is fast. This paper designs an artificial neural network for building a data classification algorithm. Since it has self-learning ability, high speed, the ability to find the optimal solution, and association storage capabilities, it obtains remarkable effect. This paper engages the five types of algorithm to predict classification respectively and from the perspective of comparison to domestic and foreign research results on the accuracy of the analysis. Through the analysis of the experimental results of this paper, to extract the about the classification of ovarian tumor clinical data extraction rules, the paper achieve the purpose of for the early prediction of ovarian cancer, to aid in clinical diagnosis and improve the survival rates of patients with ovarian cancer.