Diagnosis by Prompt, Optimizing Hospital Visits

Abstract

Recently, AI has made significant advances in the medical field. Notably, OpenAI’s GPT-4 successfully passed the Japanese medical licensing exam, demonstrating its diagnostic capabilities and showing its potential application in healthcare settings. Currently, methods to optimise patient hospital visits have become vital. Optimisation benefits patients because they can avoid unnecessary medical expenses for non-urgent symptoms, while hospitals, many of which face a shortage of doctors, can better utilise medical resources. Against this backdrop, an efficient and effective system is needed for both patients and healthcare providers. Our project leveraged the latest large language model and prompt engineering techniques to develop a system which diagnoses and advises patients on whether they need to visit a hospital. By using detailed symptom information provided by users, the AI offers appropriate medical advice, helping patients evaluate their symptoms confidently and reducing the burden on healthcare facilities. We expect this system to serve as a bridge between patients and medical institutions.