Detecting Adverse Drug Reactions Based On Deep Learning And Sentiment Analysis

Adverse drug reactions(ADRs)is an undesired harmful side effect occurred from a medication or drugs.ADRs have been one of the most important reasons which harm the public health,so the research on the detecting adverse drug reactions has been one of very important subject that can’t be ignored in medical field.Due to the complexity of research process,traditional medical methods such as clinical trials needs a lot of money and time consumption in ADRs mining.To tackle this problem,new methods for mining drug adverse reactions have been proposed in recent years.With the rapid development of Internet,spontaneous reporting systems and social media have become the main methods of adverse drug reactions detection.So this paper proposed the method of mining the adverse drug reactions from the spontaneous reporting systems and social media based on the deep learning and sentiment analysis.Towards the adverse drug event reports,we utilize the association rule mining to reconstruct the data from adverse drug event reports,and apply modified embedding models to calculate the relevance of the drug and adverse reactions to detect potential ADRs.We examine the effectiveness of methods by conducting experiments on two drugs.And in the research of ADRs detecting from social media,we automatically crawls the information published by users from the MedHelp Medical Forum and Twitter,then represented the text as embedding vectors,combined the features such sentiment features to classify the ADRs-related sentences through convolutional neural network.Experimental results shows that our method is very effective to detect ADRs for drugs.