An Efficient multi-class SVM and Bayesian network based biomedical document ranking and classification framework using Gene-disease and ICD drug discovery databases

Abstract

Biomedical document feature extraction and ranking play an essential role in the real-time document key phrase extraction and ranking. International classification of disease (ICD-10) is a list of medical related terms such as disease symptoms, abnormal discovery and disease signs. In most of the conventional methods, finding, extraction and ranking of biomedical disease patterns with the gene terms help to rank the phrase or document. However, the contextual disease patterns of these methods areindependent of gene entities, disease entities and drug discovery codes for document ranking and summarization.Conventional word embedding models such as gain ratio, entropy,Glove, chi-square and probabilistic measures are used to find the essential key terms and its relationships using static gene disease databases.The main objective of the proposed work is to optimize the word embedding model along with the key-phrase ranking and classification. Most of the biomedical applications use pre-trained gene-disease database with limited number of gene names for keyphrase ranking and classification process. In this work, an integrated gene-disease database and ICD drug database codes are used to train the model using the optimized SVM classification model and Bayesian estimation model.