Cure8 research brief
Why This Matters
AI chatbots are increasingly used for health information. If chatbot answers about IBD nutrition are inaccurate or hard to read, patients could be confused or misinformed. This study assesses those risks for ChatGPT-3.5.
Who Should Pay Attention
Patients with IBD seeking nutrition info, dietitians/nutritionists, gastroenterology clinicians, and researchers in AI-for-health education.
Study Snapshot
What To Know
This Scientific Reports article evaluated ChatGPT (v3.5) responses to 20 common nutrition questions about IBD gathered from Google. Readability was measured with standard scores (FRES, FKGL, SMOG) and found to be poor — responses required roughly a high-school reading level.
Nine dietitians/nutritionists with IBD expertise rated scientific accuracy and comprehensibility; ratings were generally favorable (around 4/5) but declined for more technical or complex questions.
The study suggests ChatGPT can produce fairly accurate and generally understandable nutrition answers for patients with IBD, but the text may be hard to read for some patients and performs worse on technical queries. The paper calls for further research to confirm and extend these findings.
Keep In Mind
Readability scores indicated responses were difficult to read (FRES 44.5; FKGL ~11.6; SMOG ~13.8). Expert ratings rated accuracy and comprehensibility ~4/5 but declined with question complexity. The study evaluated ChatGPT-3.5 only; results may differ for newer models.
Source Details
Review the original publication for the complete reporting, methods, and context.
This Cure8 brief is based on source text from the linked article. Cure8 is informational only and is not a substitute for professional medical advice, diagnosis, or treatment.