Kim H, et al. - Researchers sought to determine how improved signal quality obtained by a machine-learning model during the monitoring of physiological signals actually influences the performance of the clinical parameters after the elimination of the artifacts in subjects with traumatic brain injury. By constructing machine-learning deep belief network, automatic detection and removal of false incidences of hypotension, hypertension, bradycardia, tachycardia, and alterations in cerebral perfusion pressure (CPP) was achieved. They found that the use of machine-learning could result in a significantly reduced prevalence of false incidents due to signal artifacts. Particularly high predictive capacity of some clinical events, such as hypotension and alterations in CPP, for patient outcomes was seen after artifacts were eliminated from physiological signals.