Response to: 'On using machine learning algorithms to define clinically meaningful patient subgroups' by Pinal-Fernandez and Mammen.