Ronald Poppe and I published a paper in STATOR about the possibilities of machine learning for lie detection research purposes. You can read the entire article (in Dutch) here.