Describir: Machine learning models for Parkinson's disease detection and stage classification based on spatial-temporal gait parameters.