Have a UROP opening you would like to submit?
Please fill out the form.
Physiological Patient Monitoring Using Switching State Space Approaches
QI: MIT Quest for Intelligence
Switching state-space approaches, such as switching vector autoregressive processes, and switching linear dynamical systems, model complex dynamical phenomena as repeated returns to a set of simpler linear dynamic systems. This project aims to apply switching state space approaches to jointly model time-varying changes in multivariate physiological time series of a patient cohort for informed clinical treatment decision making. More specifically, this project will involve applying switching state space techniques to model real-world physiological time series of a patient cohort, using the learned switching dynamics to characterize the changing physiological states of patients, and analyzing the learned time-varying dynamics in the context of clinical treatment and outcomes. Related Works (see http://web.mit.edu/lilehman/www/) "A Physiological Time Series Dynamics-Based Approach to Patient Monitoring and Outcome Prediction", Li-wei Lehman et al., IEEE JBHI 2014. "Bayesian nonparametric learning of switching dynamics in cohort physiological time series: application in critical care patient monitoring", Li-wei Lehman et al., Chapter in Advanced State Space Methods for Neural and Clinical Data, Cambridge University Press, 2015.
The candidate should have experience in machine learning. Knowledge and experience in one or more of the following areas would be desirable: signal processing, probabilistic graphical models, state space models, dynamical systems, and deep learning.