Abstract-Text (inkl. Referenzen und Bildunterschriften)

Background: Managing patients with acute brain injury in the neurocritical care unit (NICU) has become increasingly complex due to technological advances and available information at bedside including information derived from multimodal neuromonitoring. Diverse data streams necessitate innovative approaches for clinicians to understand complex interactions between physiologic variables, neuroimaging parameters, electrophysiology and others. Clustering and machine learning techniques have been applied in NICU patients aiming to explore hidden phenotypes of patients defined by complex datasets. Clustering has also emerged as a valuable step to identify specific data features, and to visualize complex multidimensional data. It has therefore potential applications for the NICU. However, a systematic review dedicated to this field is lacking.

Methods: In this study we systematically review scientific articles using clustering in acutely brain injured patients and aim for a guidance/recommendation for future research. The primary objective is to provide an overview of clustering applications in NICU studies, including data types and identified clusters. As secondary objective we explore on differents study design, settings, and challenges in NICU clustering. Databases (Medline, Scopus, Web of Science) were searched for English studies involving NICU patients; traumatic brain injury, subarachnoid hemorrhage, intracerebral hemorrhage, acute ischemic stroke, hypoxic ischemic brain injury, that focused on clustering analysis. Abstracts were screened and study variables (publication, diagnosis, objectives, study-, patient, and clustering characteristics) extracted and summarized.

Results: So far, we identified 25 articles using clustering in NICU. Results will be presented in tables and figures and strengths, limitations, and recommendations will be discussed for future research.

Conclusion: We believe that this systematic review will help to understand current approaches in neurocritical care patients and provide potential directives for future developments.