An Application of Machine Learning to Forecast Hypertension Signals in Intracranial Pressure

The objective of the work presented in this article is to investigate the applicability of lightweight machine learning (ML) algorithms capable of detecting and forecasting hypertensive (HT) episodes from historical intracranial pressure (ICP) signals. Specifically, we aim at identifying noncomputationally dependent algorithms, which can be supported by lightweight hardware such as medical monitoring devices. We also propose applicable algorithms, which can be trained with a limited number of labeled samples due to the unfeasibility of manually labeling large volumes of ICP signals in most instances.