Large-scale bilateral cardiovascular monitoring via wearable rings

Bilateral ring-PPG benchmark: 97,559 samples, 1,810 participants, synchronized both hands. Bilateral-averaged MAE: HR 3.21 bpm, SBP 12.36, DBP 8.21 mmHg, age 9.46 y; BP barely beats cohort mean (13.81→12.36 SBP); pattern holds over 12 backbones.

• Form factor: ring

Wearable photoplethysmography (PPG) enables non-invasive cardiovascular monitoring, but most systems rely on single-site measurements and rarely test whether signals are spatially homogeneous across limbs. Here, we built a large-scale bilateral ring-PPG benchmark comprising 97,559 valid samples from 1810 participants with synchronized measurements from both hands. Bilateral averaged predictions showed target-dependent performance, with mean absolute errors of 3.21 bpm for heart rate (HR), 12.36 mmHg for systolic blood pressure (SBP), 8.21 mmHg for diastolic blood pressure (DBP), and 9.46 years for age. Relative to a cohort-mean baseline, HR showed the clearest improvement (10.65-3.21 bpm), whereas age improved more modestly (11.08-9.46 years). BP estimation performance remained limited, with MAE reductions from 13.81 to 12.36 mmHg for SBP and from 9.27 to 8.21 mmHg for DBP, while error dispersions remained close to target variability. Similar trends across 12 backbones suggested that the pattern was not architecture-specific. Samples with smaller bilateral differences further reduced HR MAE by 19.3%. Together, these findings establish a bilateral wearable PPG benchmark with strongest utility for HR, cautious utility for age, and limited, transparently quantified BP performance.

Verbatim from Europe PMC (opens in a new tab) · Licence: cc by-nc-nd

No results from this paper on ChatPPG leaderboards yet. Browse the leaderboards.