AI Analyzes Surgical Movements to Improve Prostatectomy Outcomes

Assessing how specific intraoperative movements affect recovery is difficult, and manual performance review is time-consuming. To help address this challenge, investigators have developed an AI system that analyzes surgical technique during robot-assisted prostate cancer surgery to predict recovery and highlight movements linked to better functional outcomes.

The AI  system, called Frame-to-Outcome (F2O), was created at Cedars-Sinai Medical Center through a collaboration among the Department of Urology, the Department of Computational Biomedicine, and the Center for Artificial Intelligence Research and Education (CAIRE). The tool focuses on the nerve-sparing phase of robot-assisted prostatectomy, when surgeons work to preserve nerves responsible for sexual function. It is designed to identify technical patterns associated with better recovery and to inform opportunities for technique refinement.

F2O analyzes operative video to detect and interpret “surgical gestures,” defined as the specific sequences and characteristics of a surgeon’s movements. It replaces labor-intensive human review by automatically extracting patterns from the nerve-sparing portion of the procedure. The system then predicts patient recovery, including the likelihood of regaining sexual function, based on those movement patterns.

In development and testing, F2O’s performance aligned with expert human reviewers while greatly reducing the time needed to assess surgical technique. Researchers trained the model using videos annotated by human analysts from 294 surgeries performed by 23 surgeons across four international centers. They then tested the system on an additional 29 surgeries, where its predictions closely matched expert assessments of patient outcomes. The approach is intended to provide objective feedback on techniques most closely linked to successful recovery.

The work builds on prior studies showing a strong relationship between specific surgical gestures and postoperative outcomes. The findings were published in npj Digital Medicine on June 23, 2026 and suggest that automated analysis of surgical video could help surgeons learn which techniques correlate with better functional recovery and apply those insights consistently. The investigators report that the technology could eventually offer real-time, objective guidance to improve surgical performance.

“Our goal isn’t simply to predict who will recover. We want to identify the surgical techniques with the best outcomes so surgeons can learn, refine and improve care for our future patients,” said Andrew Hung, MD, professor of urology at Cedars-Sinai.

Related Links
Cedars-Sinai Medical Center