Utility of Generative Artificial Intelligence for Patient Care Counseling for Mandibular Fractures

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Patient Education

Background:

• Patients increasingly use AI for medical information.
• AI-generated content on mandible fractures lacks assessment for accuracy and readability.
• This study addresses the gap in understanding AI's utility for patient queries on mandible fractures.

Purpose of the Study:

• To evaluate the readability and accuracy of information on mandible fractures generated by ChatGPT versions 3.5 and 4o.
• To determine if AI-generated medical information is suitable for patient consumption.

Main Methods:

• Twenty patient-centric questions on mandible fractures were generated using specialized tools.
• Questions were input into ChatGPT 3.5 and 4o.
• Readability was assessed using Flesch Kincaid metrics; accuracy was evaluated by a medical expert on a Likert scale.

Main Results:

• No significant differences in readability or accuracy were found between ChatGPT 3.5 and 4o.
• Readability scores were higher than recommended for patient materials.
• Accuracy was low, with many responses containing inaccuracies or omissions, rendering them inappropriate for patient use.

Conclusions:

• ChatGPT generates responses that are too complex and contain inaccuracies regarding mandible fractures.
• Patients and healthcare providers should exercise caution and be aware of the limitations of AI in providing medical information.