Predicting Intensive Care Unit Length of Stay via Supervised Learning

Abstract

[en] Healthcare is a traditional sector that is demanding, nowadays, a profound change regarding tasks and ways of work. The explotation of data-based analytical techniques together with computational capabilities are potential candidates to lead part of that demanding change. This can cause an innovation to the sector with considerable social impact. In any case, it is necessary to take into account the specific characteristics of the clinical data: quality, volume, access and multimodality. In this Master Thesis, an analysis of the data from critical patients was carried out in order to study the influence of several observables to determine their Length of Stay in the Intensive Care Unit. Try to solve that problem can help a lot not only the physicians from the mere investigation purposes point of view but also the healthcare sector because Intensive Care Unit logistics counts and it can become very important.