PD Digital Biomarker DREAM Challenge: Feature Clustering
This site acts as a supplementary material for PD Digital Biomarker DREAM Challenge concluded in October 2017.
Here, you can freely freely explore the results of clustering analysis for the features from all the subchallenges visualized as interactive charts. Clustering was performed by two standard methods: k-means and bisecting k-means. To map an input feature space to two dimensions we employed two manifold projection techniques: MDS and t-SNE, whose outcomes were fed as an input to clustering.
Please use the menu on the left to navigate through the subchallenges.