Deliberate Problem-solving with a Large Language Model as a Brainstorm Aid Using a Checklist for Prompt Generation

Large language models (LLMs) use autoregression to generate text in response to queries. Crafting an appropriate prompt to elicit the desired response from these generative artificial intelligence (AI) models to solve a clinical problem can be a challenge to clinicians who may be unfamiliar with this technology. The use of checklists to generate carefully worded queries can leverage the potential of LLMs as a brainstorming aid for medical problem-solving. Systematically using different prompts to generate the most appropriate differential diagnoses for selected clinical case scenarios, a potential checklist for prompt generation has been created and is reported here.