Interpretability of machine learning based prediction models in healthcare