Generative Artificial Intelligence in Patient Education: ChatGPT Takes on Hypertension Questions.

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Cardiovascular Health

Background:

• Uncontrolled hypertension is a major risk factor for mortality and severe medical conditions.
• Patients increasingly seek health information online, facing challenges in discerning reliable sources.
• Generative artificial intelligence (GeAI) offers new avenues for health information but raises concerns about accuracy and misinformation.

Purpose of the Study:

• To evaluate the accuracy of AI-generated information on hypertension.
• To assess the reproducibility of responses from generative artificial intelligence (GeAI).
• To explore the potential of GeAI in improving patient health literacy regarding hypertension.

Main Methods:

• A qualitative study assessed ChatGPT's accuracy and reproducibility on 100 hypertension-related questions.
• Questions were compiled by internal medicine residents and reviewed by a hypertension specialist.
• Responses were evaluated by an experienced physician for clinical appropriateness and consistency.

Main Results:

• ChatGPT provided appropriate responses in 92.5% of cases.
• Inappropriate responses were identified in 7.5% of the queries.
• ChatGPT achieved a reproducibility score of 93% for consistent information delivery.

Conclusions:

• ChatGPT exhibits significant accuracy and reliability in answering common hypertension queries.
• GeAI tools show promise as valuable resources for patient education on hypertension.
• Ongoing research is crucial to validate the trustworthiness and clinical utility of GeAI in medicine.