Abstract
Toxic epidermal necrolysis (TEN) is a severe and potentially fatal adverse drug reaction. This case report presents a 19-year-old male with pulmonary tuberculosis undergoing anti-tubercular therapy who developed TEN. The patient had multiple comorbidities including type 1 diabetes mellitus and multisystem atrophy. ChatGPT was utilized alongside conventional methods to assess causality. While conventional scoring systems estimated mortality at 58.3% (SCORTEN) and 12.3% (ABCD-10), ChatGPT yielded divergent scores. Causality assessment using WHO-Uppsala Monitoring Centre (UMC) and Naranjo's scale indicated rifampicin and isoniazid as probable causative agents. However, ChatGPT provided ambiguous results. The study underscores the potential of AI in pharmacovigilance but emphasizes caution due to discrepancies observed. Collaborative utilization of artificial intelligence (AI) with clinical judgment is advocated to enhance diagnostic accuracy and treatment decisions in adverse drug reactions. This case highlights the importance of integrating AI into drug safety systems while acknowledging its limitations to ensure optimal patient care.