Intelligent classification of clinically actionable Genetic Mutations based on clinical evidences

摘要

Cancer is the most worried ailment as the percentage of cancer patients is increasing in huge number. The early diagnosis and prognosis of a cancer type plays important role for the treatment and for clinical management of patients as well as its been important topic of research. For early detection, treatment and related recovery, examination of genes is important. Consequently, customized medicinal drug plays an essential component in treating the cancer. The personalized medicines are advised by means of studying the genetic profile of an individual with disorder. However, adopting personalized medicine in cancer treatment is happening slowly due to the big quantity of manual work is still required. Once sequenced, most cancers tumor can have hundreds of genetic mutations. However distinguishing the mutations that contribute to tumor growth from the neutral mutations is hard. The genetic mutations needs to be categorized to sort of cancer by means of considering clinical studies papers or medical observations. Currently this interpretation of genetic mutations is being completed manually. That is a very time-eating task wherein a medical specialist has to manually evaluate and classify every single genetic mutation primarily based on evidence from textual content-based medical literature. Therefore way to intelligently classify these clinically actionable genetic mutations primarily based on the clinical evidences remains a hassle to deal with. In this paper, to classify genetic mutations based on clinical evidences, the one hot encoding technique is used for extracting features from genes and their variations and tf-idf technique is used for extracting features from clinical text data. The classification is done using logistic regression. The results shows 64% classification accuracy. The accuracy is achieved using stratified K-Folds cross-validation.