A REVIEW OF EXPLAINABLE AI APPLICATIONS IN PHARMACOVIGILANCE FOR IMPROVED PATIENT SAFETY

Abstract

An alternative paradigm to classical AI's "black box" approach, explainable artificial intelligence (XAI) has recently garnered a lot of attention for its potential usefulness. This study aims to identify pharmacovigilance studies that have utilized XAI. By helping pharmacovigilance teams assess patient conditions like diabetic retinopathy and chronic diseases, as well as increase the speed and accuracy of signal detection, AI can solve major safety concerns. Therefore, experts and clinicians must continually evaluate the possible advantages and disadvantages of AI in pharmacovigilance as technology advances if it is to have the greatest possible effect on patient safety. Applying XAI in pharmacovigilance was incredibly challenging, as shown by the study's many obstacles. The fields of patient safety and pharmacovigilance make extensive use of AI for data collection on adverse pharmacological responses, analysis of medication interactions, and impact prediction; however, XAI is hardly employed in these fields.