Systematic review of high-sensitivity troponin based protocols for rule-out and rule-in acute myocardial infarction: a comparison of traditional and Generative Pre-trained Transformer (GPT)-assisted approaches

Introduction:Chest pain patients' risk is classified using a focused history and cardiovascular risk assessment, and clinical decision pathways are protocols used to define their risk.Natural language processing (NLP) is a field of computer science that enables computers to understand, interpret, and generate human language.ChatGPT is a new network architecture based on attention mechanisms.Objective: The objective of the study is to investigate the potential of ChatGPT for NLP by selecting articles within a systematic review and utilizing ChatGPT to assist in writing a paper.Methods: Our group conducted a systematic review of the literature to evaluate the diagnostic accuracy of troponin testing for early discharge of patients with suspected acute coronary syndrome.ChatGPT was tasked with selecting articles by screening their titles, abstracts, and full texts.The results obtained were compared with those of the original search.Reference documents were entered into the chatpdf.complatform under strict supervision.Results: The initial screening identified 3509 studies, of which three articles were selected for inclusion in the systematic review by ChatGTP: the HEART Pathway randomized trial, a comparison of the 2-h ADAPT versus HEART Pathway, and the RAPID-TnT.These three articles represented 60% of the five articles identified in the original review.Conclusion: The present paper shows the potential value of ChatGPT in aiding the article selection process for systematic reviews and assist in writing the paper.Nonetheless, further investigation and validation are required to confirm the effectiveness of this technology.