Advanced Colonoscopy Training Developed Through Manikin Sensorization and Computational Optimization Modeling

Project Summary/Abstract More than 19 million colonoscopies are performed in the US each year, with 60% of adults between 55-75 years having at least one within the last decade (1-3). These procedures provide physicians with minimally invasive access to diagnose and treat diseases such as colon cancer. While routine, colonoscopies are the most expensive screening test routinely performed on healthy Americans, with a mean cost of $2,125 (4). Furthering the cost is the wide variation in successful completion rates that occur based on the physician skill– While the acceptable physician success rate is only 90%, this modest target today is only achieved after physicians complete hundreds of patient cases (5-7). In light of this, the investigators have developed the Manikin Advanced Feedback Trainer (MAFT) to improve patient health and reduce healthcare costs through efficient and effective colonoscopy training. Efficient learning is produced using a high fidelity sensorized manikin trainer and a virtual “coach” that uses optimization-based feedback developed through optimization modeling. MAFT is rigorously developed through 3 Aims below: Specific Aim 1: Develop and validate an extensive colon geometry dataset and a refined sensorization system for the MAFT. Colon geometry of 959 patient virtual colonoscopies from The Cancer Imaging Archive (TCIA) will be computationally measured with a user-friendly automated organ geometry centerline program. In addition, this aim will advance MAFT through the refinement of sensor processing. Specific Aim 2: Develop, validate, and execute a 3D optimization model to solve for optimal endoscope steering control. For each specific colon geometry there exists an optimal set of endoscope steering controls to successfully insert an endoscope to reach the cecum in optimal time and with minimal trauma. An Optimization model will be developed to solve for this optimum, and then this program will be executed across the large colon dataset. Specific Aim 3: Develop and experimentally assess virtual coach to provide personalized and progressive learning for the Manikin Advanced Feedback Trainer. Hypothesis (H1): MAFT system of interactive feedback provided by a virtual coach will improve a physician’s cecal intubation rate (CIR), Adenoma Detection Rate (ADR) through improved visualizations, and reduce endoscope forces that can cause colon damage. The virtual coach will be developed and validated through Human Study 1 and 2 that utilizes both attending physicians and residents. In Human Study 3 MAFT will be implemented into the surgical residency program at Hershey Medical Center in Years 4 and 5, and residents’ skills at performing colonoscopy will be assessed based on the well-established metrics of CIR, ADR, and GAGES assessment to validate H1.