Developing and evaluating a remote, self-administered, AI-powered occupational therapy initial assessment: A crossover randomized controlled trial.

In 30 seconds

This crossover randomized controlled trial compared a remote, self-administered AI occupational therapy (OT) initial assessment with a human-administered assessment in 30 adults. Participants completed both formats in randomized order one to two weeks apart. AI usability was high (modified System Usability Scale 81.1), and ratings of whether goals were clear, personalized, relevant, attainable and motivating did not differ overall. However, perceived compassion and satisfaction favored the human format; 83% (25/30) preferred it. The findings suggest that AI can support assessment and goal setting, but participants endorsed the human interaction more strongly.

Key findings

• AI assessment usability was high, with a modified System Usability Scale score of 81.1.
• Perceived compassion favored human assessment: median 9.90 vs 7.53; Hodges-Lehmann difference 2.39, 95% CI 1.56-3.11; p < 0.001.
• Overall, 83% (25/30) of participants preferred the human-administered format.
• Agreement on goal clarity, personalization, relevance, attainability and motivation did not differ, although strong agreement favored the human format on four of five items.

Why it matters

Initial OT assessments serve both information-gathering and relationship-building functions. These results indicate that a remote AI format may achieve high usability and comparable overall goal acceptance, while perceived compassion, satisfaction and strength of endorsement remain advantages of human administration. Evaluating both task outcomes and the interpersonal experience is therefore important when considering AI in client-facing assessment.

What to keep in mind

The study included 30 adults and assessed user experience after two assessment formats; the abstract does not report evidence on clinical outcomes or longer-term effects.