5 Key Takeaways

• 1

  A multilingual, voice-enabled chatbot for retinal detachment education integrates clinician-approved knowledge with large language models.

• 2

  The chatbot uses a retrieval-augmented generation framework to provide personalized, conversational guidance while reducing hallucination risks.

• 3

  Accessibility features include speech-to-text input and multilingual text-to-speech output, addressing barriers for patients with visual impairments.

• 4

  GPT-4o outperformed other leading language models in the study, demonstrating superior consistency and alignment with reference answers.

• 5

  The chatbot is a research prototype that requires further clinical validation and real-world studies to assess usability and impact.