Intelligent Tutoring Systems for Adaptive Learning Pathways in Healthcare Training

Effective training is essential for successful surgical outcomes, yet traditional methods are often resource-intensive and inefficient. This study explores the use of AI-powered Recommender Systems to provide personalized and scalable training in robotic-assisted surgery (RAS) by offering adaptive task choices. The developed Recommender System divides the task selection process into two steps: a decision base and a decision algorithm. For the decision base data is collected and further enriched by artificial intelligence. This decision base is then used by the decision algorithm, an RL agent, which selects the next task with the aim of accelerating the student's learning. A synthetic dataset based on the Item Response Theory Knowledge Tracing Model was used to simulate task interactions and learning progress for individuals with varying skill levels across tasks of differing difficulty and requirements. Results show that a graph-based knowledge tracing model, which reveals latent structures among tasks, effectively supports the decision basis, while reinforcement learning enhances task selection within the decision algorithm. This framework demonstrates a promising approach for AI-driven RAS training, with future research focused on optimizing these components and preparing them for real-world implementation.