Testing the ability of ChatGPT to categorise urgent and non-urgent patient conditions : who ya gonna call?

This research explores the feasibility of utilising ChatGPT to categorise patient conditions as urgent and non-urgent. The primary objective is to assess the ChatGPT model's capacity to aid in the automation and digitalisation of healthcare processes, thereby alleviating the workload on healthcare professionals. The study employed a unique approach by presenting patient cases to the GPT and categorising the conditions based on urgency. In collaboration with an experienced hospital doctor, a set of questions was prepared and presented to a medical expert, along with the GPT model. Subsequently, the medical expert was consulted to assign urgency modalities for the same cases. The generated categorisations and the expert-assigned modalities were compared to evaluate the model's accuracy. The outcomes of this research have significant implications for healthcare management. Implementing AI to support triage processes and decisions could streamline patient care, ensuring appropriate and timely treatment allocation. By delegating specific tasks to AI, healthcare employees could focus on providing direct medical attention, leading to enhanced efficiency and improved patient outcomes. However, the results indicate that there is still uncertainty in using ChatGPT to provide medical advice. Ultimately, this study contributes to the broader exploration of AI's potential in healthcare decision-making, promoting the integration of advanced technologies to optimise medical services and enhance patient experiences.