Using anticipative hybrid extreme rotation forest to predict emergency service readmission risk

Abstract

This paper provides a real life application of the recently published Anticipative Hybrid Extreme RotationForest (AHERF), which is an heterogeneous ensemble classifier that anticipates the correct fraction ofinstances from each basic classifier architecture to be included in the ensemble. Heterogeneous classi-fier ensembles aim to profit from the diverse problem domain specificities of each classifier architecturein order to achieve improved generalization over a larger spectrum of problem domains. Given a prob-lem dataset, anticipative determination of the desired ensemble composition is carried out as follows:First, we estimate the performance of each classifier architecture by independent pilot cross-validationexperiments on a small subsample of the data. Next, classifier architectures are ranked according to theiraccuracy results. The likelihood of each classifier architecture instance appearing in the ensemble is com-puted from this ranking. Finally, while building the ensemble, the architecture of each individual classifieris decided by sampling this likelihood probability distribution. In this paper we provide an applicationof AHERF to a real life problem. Readmission of patients short time (i.e. 72 h) after being released poses agreat economical and social challenge, so that many efforts are being addressed to predict and avoid read-mission events. We present the results of the application of AHERF over a real life dataset composed of156,120 admission cases recorded between January 2013 and August 2015. AHERF archives results overor close to 70% sensitivity in the prediction of readmissions for adults and pediatric cases, suggesting thatit can be used to build institution specific prediction systems.