Namespace:    neuralmtpp

PI: Christian Shelton
Institution: University of California, Riverside
Project description:

Learn statistical models of timings, values, and the heterogenous types of events found in a medical event data from intensive care units. Using a modified continuous-time Long Short Term Memory network, we empirically evaluate the efficacy of jointly learning the sequence distribution and optimizing for a target cost at the end of the sequence. Our evaluation is performed on a large cohort of patients on four clinically relevant tasks, which requires powerful GPU computation.

Software: Python, CUDA, PyTorch, NumPy, CometML, Conda

Back to namespaces