Analysis of molecular level biomedical event trigger extraction using automated deep learning detection model

Abstract

Molecular events extraction is a predominant field of molecular biology affect the development of a phenotype over drug reactions or certain diseases. Machine or deep learning or meta-heuristic methods are often utilized for the training task for event trigger extraction in case of biomedical data, however, most of these approaches are rarely employed due to its difficulty on text annotation. newlineThe present study is designed as a task of allocating word labels in the biomedical event dataset. A rich collection of features is fed into the deep learning classification algorithm to train and improve the classification accuracy from an annotated sentence. These events are identified accurately using the ruleset generated by a machine learning algorithm or reinforcement learning algorithm. Further the output from the classifier is supported by these ruleset generations to enhance the detection using feedback sent to the ruleset. Finally, the classifier is used to extract the entire event trigger from the dataset after training. newlineIn the first part, gene ontology with RBNN (Radial Belief Neural Network) is used for Biomedical Event Extraction. In conjunction with the labelled data classifier for identification of lung molecular event causes, RBNN uses RBM (Restricted Boltzmann Machines) for the process of feature extraction and BPNN (Back Propagation Neural Network) for classification. newlineIn the second part, RNN with PSO (Particle Swarm Optimization) is used to extract the biomedical event triggers, where the classifier gathers data from the data sets and extracts the rich features. newline