Feasibility of continuous finger arterial pressure recording with synchronised facial video in a hypertension clinic cohort: the BP-WAVE pilot study

Introduction. Whether the arterial pressure waveform carries structural vascular information that brachial pressure does not, and whether its shape can be read from facial video, are both uncertain, and recording a continuous pressure reference with synchronised video in a hypertension clinic has not been reported. We sought to determine whether continuous finger arterial pressure recording with synchronised facial video, linked to existing coronary calcium imaging, is feasible in adults attending a hypertension clinic, whether a consumer camera recovers heart rate and pulse shape in that setting, and to generate exploratory estimates of association between waveform morphology and coronary calcium. Methods. BP-WAVE is a single-site prospective observational pilot study. The primary outcome was feasibility, pre-specified as recruitment, consent, completeness of waveform recording, imaging availability and data completeness. Secondary outcomes were waveform quality, device agreement, feature extraction, and recovery of heart rate and pulse shape from video. The exploratory outcome was the Spearman correlation of 14 waveform features with the Agatston calcium score, with exact permutation p values adjusted within each family of tests. Results. Nine of ten planned participants were recorded over 12 days (median age 55 years, range 40 to 74; five women); all nine completed the recording with video and had a calcium score (median 351, range 33 to 1816). Of 4,178 detected beats, 2,965 (71%) passed quality control. Video recovered heart rate within 0.4 to 4.1 beats per minute in eight of nine participants and pulse shape above a phase-randomised noise bound in six of nine (held-out r 0.48 to 0.76). Form factor had the strongest correlation with calcium (rho -0.68, exact p 0.050, q 0.68); no feature was associated with calcium after Benjamini-Hochberg adjustment within any of the nine families of 14 tests (126 tests in all; smallest q 0.17). Conclusion. In this single-site pilot of nine adults from a hypertension clinic, continuous finger arterial pressure recording with synchronised facial video and linkage to coronary calcium was feasible, a consumer camera recovered heart rate in eight and pulse shape in six participants, and no waveform feature was associated with coronary calcium after correction for multiple testing.