Generative AI robot therapist helps stroke patients recover

Johannes Lachner of Purdue University and Noah Geiger of MIT built a robotic rehabilitation system that uses generative AI and real-time force feedback to learn from human therapists. It gives patients just the right amount of assistance during exercises such as reaching and lifting arms, aiming to extend therapists' reach rather than replace them. Next comes therapist-specific models and a long-term clinical study.

• Strokes leave 5 million people with long-term impairments each year
• The system learns from therapists and doses assistance during exercises
• Next phase is a long-term clinical study with the same patients
• Possible uses include contact-rich industrial and collaborative settings