Demo: multi-device cuffless blood pressure monitoring via PPG-to-ECG generation (MobiCom '26)

Live demo of a PPG-conditioned diffusion model that generates ECG for cuffless BP estimation. The ECG-targeted generative process uses a QRS-adaptive forward process plus scale- and frequency-alignment modules that constrain R-peak position, amplitude range and heart rate. It runs on heterogeneous commodity wearables: two smart rings (Circular Ring, OmniRing), a Mi Band 4 and the authors' own PPG wristband.

• Form factor: ring + wrist

Cuffless blood pressure (BP) monitoring is central to managing cardiovascular diseases, and is commonly based on photoplethysmography (PPG) or electrocardiography (ECG). ECG delivers better BP accuracy, but straps and patches make it impractical for daily use; PPG is already present on the rings and watches people wear all day, yet carries far less cardiac information. We bridge the two by generating ECG from PPG: a diffusion model performs a selective, ECG-targeted generative process conditioned on PPG, using a QRS-adaptive forward process together with scale- and frequency-alignment modules that constrain R-peak position, amplitude range and heart rate. This demo runs that pipeline live on heterogeneous commodity wearables—two smart rings (Circular Ring, OmniRing), a smart watch (Mi Band 4) and our own PPG wristband. Attendees wear the device of their choice, watch an ECG being synthesized from their own PPG in real time, and compare the resulting estimate against an Omron cuff, shown alongside a signal-quality index and the running error against the cuff.

Verbatim from Semantic Scholar (opens in a new tab) · Abstract via Semantic Scholar; © the authors/publisher

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