Machine learning methods for categorization of hypersomnias/controls from PSG & MSLT data
Summer student mentorship project: These are the files for implementation and analysis of stepwise multinomial logistic regression, decision trees, random forest, gradient boosting machine, and recursive partitioning and regression trees from a project assessing better diagnostic differentiation between controls (from Wisconsin Sleep Cohort), type 1 narcolepsy, type 2 narcolepsy, and idiopathic hypersomnia (latter 3 from the Stanford Narcolepsy Center database) based on features from the MSLT and preceding nocturnal polysomnogram.