Predicting renal recovery from acute kidney injury using artificial intelligence
Acute kidney injury (AKI) is a frequent and severe complication in critically ill patients, and is especially associated with systemic inflammatory syndromes. AKI leads to increased morbidity and mortality and has implications both on short- and long-term outcomes, such as chronic kidney disease (CKD). Renal recovery from AKI is a key factor on the AKI-to-CKD-continuum, however clinical prediction of renal recovery is difficult, and the scientific understanding of these processes is still very limited.
Increasingly, methods of artificial intelligence and machine learning are applied to clinical research projects, including AKI. However, so far, no machine-learning models have been developed for the prediction of renal recovery from AKI.
This project aims to establish such prediction models, supporting clinicians in their bedside decision-making in this regard. Furthermore, we aim to apply causal modelling approaches and interpretable machine-learning algorithms to generate new knowledge about the pathophysiology of renal recovery from AKI. To enable these data-driven models, we have established a research-compatible data-warehouse which is based on the AMDS dataset and integrates state-of-the-art machine learning analyses and data visualisation tools.