Autor: Ricardo Alfredo Quintano Neira
Orientador: Silvio Hamacher
Sepsis is considered a worldwide health and economic burden.In Brazil, sepsis is the major cause of death in Intensive CareUnits, as well as, one of the main causes of late hospital mortality.
In this thesis, we first provide a 10 years population-level epidemiologyreport of sepsis in Brazil, using data from the Brazilian Unified HealthSystem. Secondly, we present a research study that supports health care facilities in the evaluation and optimization of their sepsis clinical pathways (CP) using process mining techniques. A CP consists of a well-defined care plan, which includes a clear order and time for the execution of interventions with expected outcomes. During the execution of this study, it became clear for us the lack of existing process mining techniques for the optimization of CPs. Thus, we proposed, implemented and testeda novel process mining technique that automatically provides a set ofrecommendations for any process and applied it for CP improvement.
Our developed technique (Multi-CAT) identifies and highlights a set ofactivities and sub-sequences that provide positive or negative outcomesconsidering multiple simultaneous criteria. We successfully applied ourtechnique in a real sepsis CP, and we acquired more optimization insightsthat we got in our previous manual analysis. We conclude that Multi-CAThas high potential to help in the optimization of processes with a goodperformance. In the most complex test scenario, with 58 activities, 50,000cases and 14,451 variants, Multi-CAT took 2.04 minutes to execute theanalysis. Finally, the tool was validated in two different processes, indicating potential capability to be applicable to other business areas.