Student Modeling in Orthopedic Surgery Training: Exploiting Symbiosis between Temporal Bayesian Networks and Fine-grained Didactic Analysis
Cognitive approaches have been used for student modeling in intelligent tutoring systems (ITSs). Many of those systems have tackled fundamental subjects such as mathematics, physics, and computer programming. The change of the student’s cognitive behavior over time, however, has not been considered and modeled systematically. Furthermore, the nature of domain knowledge in specific subjects such as orthopedic surgery, in which pragmatic knowledge could play an important role, has also not been taken into account deliberately. We believe that the temporal dimension in modeling the student’s knowledge state and cognitive behavior is critical, especially in such domains. In this paper, we propose an approach for student modeling and diagnosis, which is based on a symbiosis between temporal Bayesian networks and fine-grained didactic analysis. The latter may help building a powerful domain knowledge model and the former may help modeling the learner’s complex cognitive behavior, so as to be able to provide him or her with relevant feedback during a problem-solving process. To illustrate the application of the approach, we designed and developed several key components of an intelligent learning environment for teaching the concept of sacro-iliac screw fixation in orthopedic surgery, for which we videotaped and analyzed six surgical interventions in a French hospital. A preliminary gold-standard validation suggests that our diagnosis component is able to produce coherent diagnosis with acceptable response time.