Artificial intelligence in occupational therapy documentation: Chatbot vs. Occupational Therapists.

Area of Science:

• Medical Documentation
• Artificial Intelligence in Healthcare
• Occupational Therapy

Background:

• AI language models show potential for medical documentation generation.
• Effectiveness of AI in occupational therapy (OT) documentation, particularly quality and empathy, is understudied.

Purpose of the Study:

• Compare quality and empathy of OT documentation generated by human therapists versus ChatGPT-3.5.
• Evaluate AI's role in clinical documentation for occupational therapy.

Main Methods:

• Standardized OT case scenarios used for human and AI documentation generation.
• Evaluated documentation using Likert scales for quality (completeness, correctness, concordance) and empathy (cognitive, affective, behavioral).
• Analyzed inter-rater reliability and correlations between quality and empathy.

Main Results:

• AI-generated documentation received significantly higher ratings for quality and empathy (p < 0.001).
• Human-generated documentation showed stronger quality-empathy correlations and higher inter-rater reliability.
• AI outputs may offer higher perceived quality but exhibit greater variability in interpretation.

Conclusions:

• AI tools can potentially reduce therapist documentation burden with high-quality, empathetic notes.
• Human-authored documentation currently offers greater consistency among evaluators.
• Further AI development is needed to improve contextual sensitivity and evaluator reliability in clinical settings.