ICU Patient Flow Prediction via Discriminative Learning of Mutually-Correcting Processes

Over the past decade the rate of intensive care unit (ICU) use in the United States has been increasing, with a recent study reporting almost one in three Medicare beneficiaries experiencing an ICU visit during the last month of their lives. With an aging population and ever-growing demand for critical care, effective management of patient flow and transition among different care facilities will prove indispensible for shortening lengths of hospital stays, improving patient outcomes, allocating critical resources, and reducing preventable re-admissions. In this paper, we focus on a new problem of predicting the so-called ICU patient flow from longitudinal electronic health records (EHRs), which is not explored via existing machine learning techniques. By treating a sequence of transition events as a point process, we develop a novel framework for modeling patient flow through various ICU care units and predict patients' destination ICUs and duration days jointly. Instead of learning a generative point process model via maximum likelihood estimation, we propose a novel discriminative learning algorithm aiming at improving the prediction of transition events. By parameterizing the proposed model as a mutually-correcting process, we formulate the problem as a generalized linear model, i.e., multinomial logistic regression, which yields itself to efficient learning via alternating direction method of multipliers (ADMM). Furthermore, we achieve simultaneous feature selection and learning by adding a group-lasso regularizer to the ADMM algorithm. Using real-world data of ICU patients, we show that our method obtains superior performance in terms of accuracy of predicting the destination ICU transition and duration of each ICU occupancy.