Healthcare operations technology company Opmed and Mayo Clinic have presented results from a multi-year collaborative study evaluating the use of artificial intelligence to improve operating room (OR) scheduling efficiency. The findings were unveiled at the American College of Cardiology Expo (ACC.26) and focused on cardiovascular (CV) procedures, an area where scheduling variability can create significant operational and financial challenges for health systems.
The study assessed Opmed’s multimodal deep learning platform, which was designed to predict procedure durations more accurately than traditional scheduling methods. Many hospitals continue to rely on historical averages, manual planning processes and clinician judgment when allocating OR time. Variations in patient complexity, surgical techniques and intraoperative developments can lead to inaccurate estimates, resulting in delays, overtime costs or underutilized operating room capacity.
To address these challenges, Opmed developed an AI-driven scheduling platform trained on historical cardiovascular procedure data from Mayo Clinic collected between 2022 and 2025. The system analyzes multiple data sources simultaneously, including patient characteristics, procedure types, operational workflow metrics and unstructured clinical information such as physician notes and preoperative electrocardiogram (ECG) reports.
Researchers evaluated four model configurations using a separate validation cohort of 643 cardiovascular procedures performed between November 2025 and January 2026. The conventional scheduling approach produced a Mean Absolute Error (MAE) of 1.13 hours per case. Opmed’s highest-performing model, which combined structured clinical data with unstructured physician notes, achieved an MAE of 0.564 hours, reducing scheduling error by approximately half.
The study also reported a root mean square error (RMSE) of 0.799 hours and an R² score of 0.721 for the top-performing model, compared with an R² score of 0.31 for traditional human-based scheduling approaches. According to the companies, all AI model variations maintained prediction errors below 0.60 hours.
The findings suggest potential operational benefits for hospitals seeking to improve OR utilization. More accurate scheduling may reduce unused operating room time, minimize delays and support the addition of further procedures without expanding staffing levels or infrastructure.
Commenting on the results, Dr. Mor Brokman Meltzer, CEO and Co-founder of Opmed, said, “While much of the healthcare AI conversation focuses on areas such as medical imaging or surgical tools when it comes to AI’s impact, this study demonstrates the potential, to patients and medical centers alike, of AI scheduling.”
She added, “Being able to collaborate with the Mayo Clinic to study this impact over a long period serves as a major milestone in bringing this technology and its impact to the forefront of the medical AI conversation.”
Click here for the original news story.
