• Posted January 21, 2026

AI Could Make At-Home Balance Exercises More Effective

AI might be able to help people undergoing balance training as part of their physical rehabilitation, a new study says.

Patients wearing just four sensors — on each thigh, the lower back and upper back — can get accurate real-time, AI-driven feedback on balance exercises they’re performing at home, researchers recently reported in the Journal of NeuroEngineering and Rehabilitation.

“Our machine learning model used data from wearable sensors to predict how physical therapists would rate patients' performance on balance exercises, providing a basis to make recommendations about the most appropriate set of exercises to perform next,” said senior researcher Kathleen Sienko, a professor of mechanical engineering at the University of Michigan.

“This type of AI-based support would be helpful in between appointments or after people complete their insurance-reimbursed sessions with a clinician," she said in a news release.

Balance training helps reduce the risk of falls among seniors and people with sensory or motor impairments, helping them maintain their independence, researchers said in background notes.

Typically, physical therapists observe patients during in-clinic sessions, continually assessing the amount of difficulty they’re having. They use these observations to choose appropriately challenging balance exercises.

But such observation isn’t available when patients do balance exercises at home, impeding their effectiveness.

To build an AI model that would help with this, researchers filmed participants performing standing balance exercises at various levels of difficulty, while wearing 13 sensors attached to their bodies with Velcro straps.

These videos were used to train an AI on evaluating balance difficulty, researchers said.

A group of physical therapists later watched these videos and judged how hard the participants were working for each balance exercise.

The AI model’s judgments jibed with the physical therapists’ evaluations with nearly 90% accuracy, researchers found.

In fact, the AI only needed four of the 13 sensors to maintain its accuracy, researchers said.

This AI could be particularly helpful for seniors living in places without easy access to physical therapy, such as rural parts of the country, researchers said.

"In some regions, access to physical therapists specializing in balance rehabilitation may not be possible," Sienko said.

"I was excited by the possibility of developing something that could expand access to services like balance training — not only for people in rural areas across the U.S. who may lack regular access to physical therapists, but also for individuals globally,” she added.

However, the researchers said more development and testing is needed before AI-assisted balance training will be widely available.

"It is very important to understand both the strengths and potential failure modes of machine learning in physical therapy, where people's well-being is directly at stake,” researcher Xun Huan said in a news release. He’s an associate professor of mechanical engineering at the University of Michigan.

“To protect patients, these systems should be validated on real-world data and used with therapist oversight so unexpected or risky suggestions can be caught before harm," Huan said.

More information

The Mayo Clinic has more on balance exercises.

SOURCE: University of Michigan, news release, Dec. 8, 2025