Workflow Optimization in Operating Rooms and Intervention Labs Through AI-Driven Data Analytics

Operating rooms (ORs) and Image-Guided Therapy (IGT) labs are high-stakes environments that require seamless coordination across multiple disciplines. These settings are often marred by complexities, time constraints, and resource limitations, intensifying the demands on healthcare professionals. The challenges of administrative burden and information overload further hinder the ability to achieve efficient and economically viable patient care. A key element in overcoming these issues is real-time situational awareness, coupled with adaptive scheduling techniques to manage both anticipated and unforeseen events. In an increasingly strained healthcare system, there is an immediate need to address these inefficiencies through innovative solutions.

This research aims to fill two critical gaps: the absence of real-time insights into the progression of clinical procedures and the inability to adapt to the dynamic and unpredictable nature of these procedures. Our goal is to harness AI-powered data analytics and real-time end-time predictions to optimize complex workflows in ORs and IGT labs. This study employs computer vision algorithms to analyze multi-camera video footage from cardiac angiogram (CAG) and gynecologic procedures in two Dutch hospitals. By identifying objects, patients, and personnel, we are able to isolate key events within each procedure. These data points are then integrated with clinical phase annotations, enabling accurate predictions for procedure completion times.

In this talk we present our latest results for detecting clinical phases for the use case in cardiology