EYE penalty – a novel regularization penalty that incorporates domain expertise [code]

RL4BG – Reinforcement Learning for Blood Glucose Control [code]

Blood Glucose forecasting – [code]

FIDDLE – a preprocessing pipeline that transforms structured EHR data into feature vectors that can be used with ML algorithms [code]

Preprocessed data for MIMIC-III and eICU: link

mixLSTM – Relaxed Parameter Sharing in RNNs– mixture-based LSTM model to handle time-varying relationships [code] [paper]

CNNs for Structural MRI Data – modified CNN tailored to brain MRI data that includes region-specific filters; demonstrated use for the task of predicting brain age from MRI data [code][paper]