We introduce a new paradigm for evolving classification of event data streams, such as patient data in Intensive Care Units.

We present several alternative data mining approaches to evolving classification of event data streams.

The alternative approaches are evaluated on a dataset of 3,452 episodes of adult patients (≥16 years of age).

An incremental algorithm has produced the simplest and the most accurate models on Days 0 and 1.

The regenerative approaches have reached better performance in terms of predictive accuracy starting with Day 2.