Related Experiment Video
Updated: Jul 2, 2025

E-Patient Counseling Trial E-PACO: Computer Based Education versus Nurse Counseling for Patients to Prepare for Colonoscopy
Published on: August 1, 2019
ChatGPT vs. web search for patient questions: what does ChatGPT do better?
Sarek A Shen1, Carlos A Perez-Heydrich2, Deborah X Xie3
1Department of Otolaryngology-Head and Neck Surgery, Johns Hopkins School of Medicine, 601 North Caroline Street, Baltimore, MD, 21287, USA. sarek.shen@gmail.com.
ChatGPT provides more understandable medical information for diagnoses than web searches. Prompting improves ChatGPT readability for patients seeking health information online.
Area of Science:
- Medical Informatics
- Artificial Intelligence in Healthcare
- Patient Education
Background:
- Online patient information acquisition is evolving.
- Chat generative pretrained transformer (ChatGPT) offers new avenues for accessing medical data.
- Evaluating AI-generated health information against traditional sources is crucial.
Purpose of the Study:
- To compare the readability and appropriateness of ChatGPT responses versus traditional web searches for patient medical questions.
- To assess the accuracy and understandability of information provided by ChatGPT.
- To characterize the performance of ChatGPT across different question types (fact, policy, diagnosis).
Main Methods:
- Patient questions were sourced from online posts related to American Academy of Otolaryngology-Head and Neck Surgery guidelines.
- Questions were categorized into fact, policy, and diagnosis.
- ChatGPT and traditional web searches were used to answer questions.
- Readability (Flesch Reading Ease, Flesch-Kinkaid Grade Level) and understandability (PEMAT) were assessed.
- Accuracy was evaluated by blinded clinical experts.
Main Results:
- ChatGPT responses had lower average readability than web searches (FRE: 42.3 vs. 55.6).
- Understandability was comparable between ChatGPT and web search (PEMAT: 93.8% vs. 93.5%).
- ChatGPT outperformed web search for diagnosis-related questions (p < 0.01).
- Readability of ChatGPT responses improved with additional prompting (FRE 55.6).
Conclusions:
- ChatGPT is superior to web search for symptom-based diagnostic questions and equivalent for factual and policy information.
- Targeted prompting can enhance the readability of ChatGPT responses without compromising accuracy.
- Educating patients on the benefits and limitations of AI tools for medical information is essential.
Related Concept Videos
Methods of Documentation III: PIE
Patient-centered Care
Methods of Documentation II: POMR
Methods of Documentation VI: Case Management Model
For example, a patient with a chronic...
Assessment of the Gastrointestinal System II: Health Perception Pattern
Health Perception Patterns
Health perception patterns offer valuable insights into a patient's lifestyle habits and how they may impact their GI health. These patterns include:
Issues And Trends In Healthcare Delivery System
Cost Containment
Payment for healthcare services has historically promoted adoption of costly and often unnecessary or inefficient...

