Hospitalisation Prediction from Telemonitoring Data

Authors: Božidara Cvetković, Aljoša Vodopija, Mitja Luštrek, Drago Rudel and Zdravko Balorda.
Abstract: Congestive heart failure is a common disease in people aged over 65 whose management can benefit from telemonitoring. In this paper we analyse the data from a year-long telemonitoring trial of 141 patients. The trial itself already reduced the number of hospitalisations, but our goal was to use machine learning to build a classifier that could predict the remaining ones. Such a classifier could then be used for timely interventions that could further reduce the number and/or duration of hospitalisations. By engineering a large number of features from the telemonitored parameters, and experimenting with various feature-selection and machine-learning methods, we built a classifier that predicted 7 hospitalisations out of 9, and raised a false alarm in only 1 instance out of 117.

Date: 
Sunday, July 10, 2016
Event: 
IJCAI 2016 - Workshop on Knowledge Discovery in Healthcare Data