We address the problem of multi-vehicle tracking and motion prediction in highway scenarios using information from sensors and perception systems widely used in automated driving. In particular, we focus on the detection of lane change maneuvers. Dangerous lane changing constitutes the main cause of highway accidents and a reliable detection system is still lacking on modern cars. Our prediction approach is two-fold. First, a driver model learned from demonstrations via Inverse Reinforcement Learning is used to equip a host vehicle with the anticipatory behavior reasoning capability of common drivers. Second, inference on an interaction-aware augmented Switching State-Space Model allows the approach to account for behaviors that deviate from those learned from demonstrations. In this paper, we show how to combine model-based behavior prediction and filtering-based state and maneuver tracking in order to detect lane changes in highway scenarios, and present the results obtained on real data gathered with an instrumented vehicle.