Generative Artificial Intelligence Chatbots for Parental Guidance in Pediatric Orthopaedics: A Pilot Feasibility...

Area of Science:

• Orthopaedic Surgery
• Artificial Intelligence in Medicine
• Medical Education

Background:

• Parents of children with pediatric orthopaedic conditions lack reliable post-visit resources.
• Existing online resources vary in quality and general AI models have safety concerns.
• A specialized, supervised generative AI chatbot offers a potentially safer alternative.

Purpose of the Study:

• To develop and evaluate a purpose-built generative AI chatbot for pediatric orthopaedic parental education.
• To assess the chatbot's accuracy, safety, usability, and clinical acceptability.

Main Methods:

• A generative AI chatbot was trained on pediatric orthopaedic educational materials.
• The chatbot was refined by pediatric orthopaedic surgeons and evaluated by multidisciplinary clinicians.
• Evaluation used a 12-item Likert scale survey assessing key performance domains.

Main Results:

• The chatbot achieved high mean scores (over 4.0/5.0) in 11 of 12 evaluated domains.
• Clarity of explanation was unanimously endorsed; protocol alignment scored highest.
• Citation accuracy was the lowest-rated domain (4.00 ± 0.77); no hallucinations were detected.

Conclusions:

• A purpose-built generative AI chatbot demonstrated high clinical accuracy and acceptability.
• The chatbot effectively avoided unprompted treatment recommendations, supporting its use.
• Pilot findings suggest feasibility as a parental education adjunct in pediatric orthopaedics.