Objective: In this study we aimed to predict the time to syncope occurrence (TSO) in patients with vasovagal syncope (VVS), solely based on measurements recorded during the supine position of the head-up tilt (HUT) testing protocol.
Methods: We extracted various time and frequency domain features related to morphological aspects of arterial blood pressure (ABP) and the electrocardiogram (ECG) raw signals as well as to dynamic interactions between beat-to-beat ABP, heart rate, and cerebral blood flow velocity. From these we identified the most predictive features related to TSO.
Results: Specifically, when no orthostatic stress is involved, TSO in VVS patients can be predicted with high accuracy from a set of only five ECG features.
Keywords: Feature selection; Head-up tilt; Random forest; Vasovagal syncope.