Generative AI in Assisting Patients with Syringomyelia and Hydrocephalus: A Preliminary Comparison of Chatgpt and Gemini

Abstract

Patients have been relying on online resources for more information on their symptoms and diagnosis. Existing research has examined Generative AI (GenAI), mostly via ChatGPT as a way of providing more information or even comfort to patients. However, research is scarce on the appropriateness of GenAI to support and inform patients with rare conditions. These patients often find themselves in a difficult-to-navigate maze especially because they have never heard of their condition before the diagnosis and the information online can be overwhelming. In this pre-study, we demonstrate the potential of GenAI to provide support to patients concerning their diagnoses. We work with a patient who has two rare neurological conditions – syringomyelia and hydrocephalus. Utilizing a qualitative and quantitative methodology, including the Patient Education Materials Assessment Tool for Printable Materials (PEMAT-P) and the Flesch Kincaid Reading Ease (FRE) score, we analyzed the patient’s feedback to a series of prompts derived from their medical reports. The results demonstrate the potential of generative AI as a valuable support for patients, with preliminary findings indicating that Gemini excels in certain aspects such as understandability, actionability, readability, and is more trustworthy, making it an effective tool for patient education in this context.