Events

• Dr. Pengyi Shi
  Assistant Professor of Management
  Krannert School of Management
  Purdue University

   

When to discharge a patient plays an important role in hospital patient flow management and patient outcomes. In this work, we develop and implement a practical decision support tool to aid hospitals in managing the delicate balance between readmission risk at discharge and ward congestion. We formulate the discharge decision framework as a large-scale Markov Decision Process (MDP) that integrates a personalized readmission prediction model to dynamically prescribe both how many and which patients to discharge each day. We overcome challenges from both the analytical and prediction sides. Due to patient heterogeneity and the fact that length-of-stay is not memoryless, the MDP suffers the curse of dimensionality. We derive useful structural properties and leverage an analytical solution for a special cost setting to transform the MDP into a univariate optimization; this leads to an efficient dynamic heuristic. Meanwhile, off-the-shelf prediction models alone could not provide adequate input for our decision support framework. To bridge this gap, we integrate several statistical methods to build a new readmission prediction model that allows us to implement our decision framework with existing hospital data systems.
Through extensive counterfactual analyses, we demonstrate the value of our recommended discharge policy over our partner hospital’s historical discharge behavior. We also discuss the implementation efforts of this discharge optimization tool at our partner hospital. This is joint work with Jonathan Helm at Kelley School of Business at Indiana University and industry partners. The paper received the Pierskalla Best Paper Award at INFORMS 2018 and can be accessed at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3202975