Jan 2014 – Dec 2014
Mining and Analysis Social Media Data to Identify and Monitor Adverse Drug Reactions (ADRs)
Pharmacovigilance is the science of detection, assessment, understanding and prevention of adverse effects of medicine. Adverse Drug Reactions (ADRs) is defined by the World Health Organization (WHO) as “Unintended, harmful reaction suspected to have been caused by a drug under normal conditions of usage”
Main source of ADR detection are clinical trials, after which post-marketing surveillance is used to monitor performance and ADRs . Drugs go through rigorous clinical trials before they are released to the market. These clinical trials are however not very reliable because they focus mainly short term safety and efficacy, and the inclusion/ exclusion criteria of clinical trials makes it focused on a limited number of subjects.
Identifying ADRs play vital role as they could be life threatening and have a major cause of deaths in developed countries like USA and UK. Early detection of ADRS is therefore beneficial. Unfortunately, more than half of severe ADRs are not identified until 7yrs or more after drug has been FDA approved.
We used With Natural Language Processing (NLP) techniques to analyze this data by identifying some biomedical entities, establishing the relations between these entities and the events that are associated with them.
Our analysis shows that Social Media platforms especially Twitter are promising sources of real time data for pharmacovigilance because it is publicly available, there is great volume of data and it is frequently updated by users. This data however remains very noisy and needs lots of effort for cleaning up. Extracting relations between drugs and ADRs can improve pharmacovigilance as it will help in indexing, searching precise data and makes information tracing faster.