Optimizing Discharge Decisions to Reduce Surgery Re-Admissions
Pl: Ioannis Paschalidis, Electrical & Computer Engineering, ENG
Co-Pls: George Kasotakis, Surgery, MED
This project aims to develop predictive analytics to predict re-admissions within 30-days from discharge after surgery. Researchers will begin with a dataset of 5,769 BMC patients and then expand the project’s scope to 2,275,452 patients in order to improve prediction accuracy. The research project will include interpretable methods that associate each positive re-admission prediction with a sparse list of features that led to that prediction. The project will develop prescriptive analytics, offering specific discharge recommendations to optimize the trade-off between the cost of extended stay and/or follow-up actions against the cost of a 30-day readmission.
This work is funded by a Digital Health Initiative Research Award made in June, 2017.