The paper “Knowledge Graph Embeddings for ICU readmission prediction”, authored by LASIGE’s student Ricardo Carvalho, Daniela Oliveira (former LASIGE member now at Novartis) and Cátia Pesquita (integrated researcher at LASIGE) has been published in BMC Medical Informatics and Decision Making journal (h-5 index=53).

The paper presents a novel approach to predict Intensive care unit (ICU) readmissions using machine learning over a Knowledge Graph that integrates Electronic Health Records (EHR) data with scientific and clinical ontologies for a more meaningful representation of patients and their ICU stays. ICU readmissions represent a four-fold mortality risk increase and place a financial strain on healthcare institutions. Preventing too-early discharges for high-risk patients without delaying the discharge of low-risk ones is a pressing need. The application of this novel approach to real EHR data was able to prevent the readmission of 40% of Intensive Care Unit patients, without unnecessarily prolonging the stay of those who would not require it, outperforming both baseline and state-of-the-art works. This work demonstrates the potential impact of integrating ontologies and Knowledge Graphs into clinical machine-learning applications to improve patient outcomes and hopes to pave the way for more explainable approaches that explore the meaning encoded in ontologies to better explain predictions to clinicians.

The paper is available here.