Mount Sinai Researchers Find Simple AI Can Forecast Sitting in Women With Pelvic Pain

An AI reminder to move sounds useful—unless sitting is what makes pain manageable. Mount Sinai researchers found that wearable data could help predict prolonged sitting among women with chronic pelvic pain. Their study supports an early-warning system, not yet a treatment: whether acting on those predictions helps patients remains untested.

The findings appeared in npj Women’s Health. In an October 7 HealthDay story, senior researcher Ipek Ensari described the aim as anticipating inactivity rather than offering generic advice afterward. Ensari, an assistant professor of artificial intelligence and human health at Mount Sinai’s Icahn School of Medicine, wants prompts that fit naturally into a person’s day.

From Fitbit readings to a forecast

The study included 134 women with chronic pelvic pain, primarily associated with endometriosis, and 61 healthy volunteers. Participants wore Fitbit devices for up to 90 days. The devices collected minute-by-minute readings on activity, heart rate and sleep, giving researchers a record of daily routines rather than a single snapshot of movement.

Researchers used about 10 days of data from each participant to train personalized models that forecast activity one hour ahead. They then narrowed the predictions to 15-minute periods of sedentary behavior during waking hours. The proposed use was to identify opportunities for short movement breaks—what the researchers called an “exercise snack”—before a long stretch of sitting.

A small model, a proposed coach

The envisioned tool would deliver a few meaningful prompts each day, while limiting notifications that people might learn to ignore. Its intended role is closer to a personalized coach than a stream of generic reminders: use the forecast to choose when to suggest standing or taking a short walk.

When sitting is a choice, not an oversight

Writing on October 7, Pain News Network’s Crystal Lindell questioned how prompts would handle necessary sitting, including work, meals or times when pain makes movement difficult. She urged researchers to ask patients whether they would want and use the technology. That criticism concerns the proposed intervention, not the model’s ability to forecast activity.

Ensari said prospective clinical trials—studies that follow people while testing an intervention—are needed to determine whether personalized prompts reduce sedentary time, improve symptoms or enhance quality of life. Those are the next tests for the proposed coach, beyond predicting when someone will sit.