29-01-2025 | Attention Deficit Hyperactivity Disorder | Original Article
A cross-sectional study to assess patient information guides generated by ChatGPT vs. Google Gemini: autism, attention deficit hyperactivity disorder and post-traumatic stress disorder
Authors:
Daaniya Tariq, Aneesha Kaur, Pratik Jadhav, Karanbir Singh Thappar
Published in:
Journal of Public Health
Login to get access
Abstract
Introduction
Use of Artificial Intelligence tools such as ChatGPT and Google Gemini enables understanding the disease perspective for the patients with underlying pathogenesis and aids in management. The present study aims to compare ChatGPT and Google Gemini responses for writing patient information guides on Autism, Attention Deficit Hyperactivity Disorder (ADHD) and Post-Traumatic Stress Disorder (PTSD) among pediatric age groups.
Methodology
This cross-sectional study was conducted by generating patient education guides for three psychiatric disorders using ChatGPT and Google Gemini. The responses generated were compared for parameters such as grade level, ease score, word count and sentence count. Tools such as the Flesch–Kincaid calculator, Quillbot and modified DISCERN score were used to assess the ease, similarity and reliability of the responses, respectively. Analysis was done using RStudio v4.3.2.
Results
Responses generated by ChatGPT and Google Gemini show no significant difference in the word count (P = 0.1022), sentence count (P = 0.1289) and average word per sentence (P = 0.7629). A higher ease score was observed for Google Gemini (34.70) compared to ChatGPT (17.27), with no statistical significance. Grade level scores calculated for ChatGPT were 13.07 and 10.05 for Google Gemini. Modified DISCERN score predicted a higher reliability score for Google Gemini (4.67/5) versus ChatGPT (4.33/5).
Conclusions
Data collected using ChatGPT and Google Gemini reveals a higher ease score for Google Gemini, which signifies easy understanding of the text for the general population. The Modified DISCERN score was also higher for Google Gemini which indicates reliability of the data collected. The study has its own limitations as a greater variety of AI tools needs to be explored to enhance knowledge and understanding for further study and patient’s perspective.