Passive Heart-Rate Monitoring During Smartphone Use in Everyday Life

Why It Matters

Passive smartphone monitoring expands affordable, continuous cardiovascular tracking to billions, reducing reliance on wearables and addressing equity gaps in health data collection.

Key Takeaways

• PHRM achieves <5% MAPE in lab and free‑living tests
• Accuracy holds across all Fitzpatrick skin tones, meeting FDA equity targets
• Daily resting heart rate estimates within 5 bpm MAE of wearable trackers
• System runs passively on 26 smartphone models during normal screen unlocks
• Researchers released pretrained model and 162 k video dataset for public use

Pulse Analysis

Smartphones are now in the hands of 90 % of U.S. adults, making them a powerful platform for health monitoring that rivals dedicated wearables. Traditional resting heart‑rate measurement requires a period of stillness and specialized hardware, limiting its frequency and accessibility. By leveraging remote photoplethysmography (rPPG) through the front‑facing camera, the new PHRM system captures eight‑second facial videos whenever a user unlocks their device, turning everyday interactions into a continuous stream of cardiovascular data. This approach sidesteps the need for extra devices while maintaining privacy, as processing occurs locally on the phone.

The technical breakthrough lies in an ensemble of temporal shift convolutional neural networks that treat heart‑rate estimation as a classification problem across a 40‑180 bpm range, outputting a confidence‑weighted probability distribution. A Kalman filter aggregates valid measurements throughout the day to produce a reliable daily resting heart‑rate (RHR) estimate. In laboratory trials, PHRM delivered a mean absolute error of 4.09 bpm and a MAPE of 5.65 %, comfortably below the ANSI/CTA‑2065 10 % threshold. Real‑world testing across 26 phone models and diverse lighting conditions yielded a video‑level MAPE of 4.83 % and demonstrated equitable performance across Fitzpatrick I‑VI skin tones, meeting FDA non‑inferiority standards.

Clinically, accurate, passive RHR tracking can flag early signs of cardiovascular stress, inform fitness coaching, and enhance remote patient monitoring programs without additional hardware costs. For insurers and employers, the technology offers a scalable way to incorporate biometric data into wellness incentives. The public release of the pretrained model and a 162 k video dataset encourages open‑source innovation, potentially accelerating integration with telehealth platforms and electronic health records. As smartphones continue to evolve with better cameras and processing power, passive heart‑rate monitoring could become a standard feature of digital health ecosystems, democratizing access to vital sign analytics across socioeconomic and ethnic groups.