Artificial intelligence and automation in flexible endoscopy

Dr. Kumbhari and his team are working to move artificial intelligence (AI) in endoscopy beyond its current limits. While AI has already been helpful in endoscopy, most existing systems rely almost entirely on video images. Video is important, but it captures only part of how endoscopy is performed.

Human endoscopists do not rely on vision alone. They also depend on how the scope moves, how much force is applied, tactile feedback and constant real-time decision-making. When these elements are not captured, AI systems remain limited in what they can truly learn or replicate.

To address this gap, Dr. Kumbhari and his team are developing the Comprehensive Acquisition Platform for Training Universal Robotic Endoscopy (CAPTURE) research platform. This novel research platform is designed to collect synchronized data from multiple sources, first in simulated environments and then during real-world endoscopic procedures.

CAPTURE brings together the full spectrum of endoscopy data, including:

  • Endoscopic video.
  • Scope motion and positioning.
  • Applied forces.
  • Audio signals.

CAPTURE creates a much more complete picture of how endoscopy is performed. This rich dataset — the first of its kind — will train robotic platforms to learn how endoscopy is done. CAPTURE could improve physician training and guide the creation of assistive robotic technologies.

Besides CAPTURE, Dr. Kumbhari's work also examines how AI affects human performance. As AI becomes more common in clinical practice, there is growing concern about overreliance on technology and the potential loss of essential endoscopic skills. This research includes ongoing work studying whether AI use may be associated with lower adenoma detection rates in some settings or may lead to unnecessary interventions or biopsies.

By studying real-world data before and after AI adoption, the research aims to ensure that AI supports clinicians, strengthens decision-making and enhances patient care rather than replaces critical physician judgment.