Abstract
AbstractA large amount of patient harm is preventable with timely and accurate detection of adverse drug events (ADEs). The ability to identify these events has been a challenge due to the passiveness of the reporting system and limited clinical trial data. Electronic health records (EHRs) gathered during routine clinical care as well as from mobile devices, contain real-time, real-world heath data, offering a potentially more proactive approach to monitor and detect safety signals. The advancement of machine learning (ML) and natural language processing (NLP) methods have provided new opportunities to identify drug safety signals in the largely unstructured free-text EHR data. Here, we surveyed pressing drug safety issues and the evolving ML and NLP methods that developed to identify them. As indicated in the literatures, the AI-aided system, if correctly implemented, would enhance patient safety by detecting the adverse events, improving the error detection, patient stratification, and drug management. Nevertheless, future work is still needed to address the many limitations in the EHR-based ML pharmacovigilance systems.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call