Optimised Selection of Stroke Biomarker Based on Svm and Information Theory

• Main Authors: Wang Xiang, Shi Wei, Wang Xiao-Cui, Wang Tao
• Format: Article
• Language: English
• Published: EDP Sciences 2017-01-01
• Series: ITM Web of Conferences
• Online Access: https://doi.org/10.1051/itmconf/20171205013

With the development of molecular biology and gene-engineering technology, gene diagnosis has been an emerging approach for modern life sciences. Biological marker, recognized as the hot topic in the molecular and gene fields, has important values in early diagnosis, malignant tumor stage, treatment and therapeutic efficacy evaluation. So far, the researcher has not found any effective way to predict and distinguish different type of stroke. In this paper, we aim to optimize stroke biomarker and figure out effective stroke detection index based on SVM (support vector machine) and information theory. Through mutual information analysis and principal component analysis to complete the selection of biomarkers and then we use SVM to verify our model. According to the testing data of patients provided by Xuanwu Hospital, we explore the significant markers of the stroke through data analysis. Our model can predict stroke well. Then discuss the effects of each biomarker on the incidence of stroke.