Explainable AI for Enhanced Safety Signal Detection and Mitigation in Clinical Trials: Unveiling Insights from SDTM Data
Authors:Lasya Shree Sharma
Abstract-Clinical trials play a crucial role in ensuring the safety and efficacy of emerging drugs and treatments. However, the conventional statistical methods employed for analyzing adverse event (AE) data within Safety Domain Terminology Mapping (SDTM) datasets often lack transparency, posing challenges in interpretation and impeding targeted risk mitigation efforts. Addressing this issue, we propose a novel approach that involves harnessing Explainable AI (XAI) algorithms to discern key features and relationships relevant to specific safety signals within SDTM AE data. This paper delves into the potential transformative impact of employing XAI in conjunction with traditional safety analyses, thereby enhancing our comprehension of safety concerns and the overall effectiveness of risk management techniques. By leveraging XAI, we aim to not only uncover hidden patterns and correlations within the intricate web of AE data but also to provide a more interpretable framework for stakeholders involved in clinical trials. This innovative integration of XAI into safety analyses has the potential to significantly augment our ability to identify and understand safety signals, ultimately contributing to more informed decision-making in the realm of drug development and patient care.
