Evaluating Artificial Intelligence in the Support of Aneurysmal Subarachnoid Hemorrhage Management: A Comparative Analysis of Contemporary Clinical Guidelines, Expert-Identified Controversial Questions, and Three Major AI Models

Abstract Background Aneurysmal Subarachnoid Hemorrhage (aSAH) remains a significant medical challenge. Controversies in managing aSAH, such as optimal blood pressure levels and antifibrinolytic therapy, persist despite advancements in treatment. The recently published guidelines from the NICE, the NCS, and the AHA/ASA show divergence in several key management aspects. This study aims to explore the processing and analysis capabilities of Artificial Intelligence (AI) models in handling controversial aSAH management issues. Methods Twelve controversial questions regarding aSAH management were presented to three artificial intelligence (AI) models: ChatGPT-3.5, ChatGPT-4, and Bard. Questions covered areas like blood pressure management, timing for aneurysm securing procedures, the use of intravenous Nimodipine, handling Nimodipine-induced hypotension, and the effectiveness of transcranial sonography in monitoring vasospasm-induced delayed cerebral injury (DCI). Results AI models’ responses were generally aligned with AHA/ASA guidelines and expert opinions on blood pressure management before and after aneurysm securing. However, significant gaps were noted in their knowledge, especially regarding the role of intravenous Nimodipine and its hypotensive effects. The use of transcranial sonography for monitoring DCI induced by vasospasm was another area where the models showed limited understanding, with only ChatGPT-4 suggesting integration with other imaging techniques and clinical assessment. Conclusions AI models demonstrate potential in assisting with complex medical decision-making in aSAH management. However, their current capabilities highlight the need for ongoing updates and integration with real-world clinical expertise. AI should be viewed as a complementary tool to human judgment. Future developments in AI should focus on enhancing its accuracy and relevance to current medical practices.