Signal artefacts in wearable-based biosignal acquisition in patients with ventricular assist devices: the SMART-VADCH trial

Telemonitoring is increasingly used in cardiovascular care, yet data on the reliability of smartwatch-based biosignal acquisition in patients with advanced heart failure and LVAD support remain limited. This study evaluated the feasibility and accuracy of smartwatch-derived heart rate (HR), oxygen saturation (SpO₂), and single-lead ECG measurements in this population.

In this prospective, single-centre pilot study, patients without LVAD or with durable (HeartMate 3) or temporary (Impella 5.5) LVADs underwent parallel resting measurements using a consumer smartwatch and standard clinical. Measurements unsuccessful after three attempts were classified as failures. Thirty-eight patients were enrolled (15 heart failure, 15 durable LVAD, 8 temporary LVAD). In patients without LVAD, HR was feasible in all patients and accurate within a 5% tolerance in 93% (95% CI 70–99%). At least one interpretable ECG was obtained in all, although QRS duration met a 10% tolerance in 33%. SpO₂ was feasible and accurate in 93%. In LVAD, HR was feasible in all patients and accurate in 82%. ECG was technically successful in all. QRS complexes were distinguishable in 73% of durable and all temporary LVAD. SpO₂ feasibility was overall limited (36%), but higher in temporary than durable LVAD patients (63% vs. 20%), and accurate when successful. SpO₂ failure was observed in patients without aortic valve opening, with tachycardia >100 bpm, and with darkly pigmented skin.

The evaluated wrist-worn smartwatch reliably provided HR and ECG measurements in advanced heart failure, including LVAD patients, whereas SpO₂ performance remains limited.