This repository hosts pre-trained models for decoding of Bradykinesia, Tremor and Dyskinesia Parkinson's disease symptoms. Models are split into classification and regression. Regression outputs are based on PKG-wearable scores. For further information for training, see train_final_models.py. Features were computed using py-neuromodulation. The feature computation is provided in run_features.py.

Seemodel_inference.ipynb for exemplary inference.

Contact information: Timon Merk [email protected]