PPG-based algorithm achieves 98.93% accuracy in detecting device wear status with negligible battery impact, eliminating the critical problem of misclassifying patient stillness as non-compliance in continuous monitoring applications.
Simulation evidence supporting continuous, automated monitoring as a superior approach for clinical research.
A study demonstrating the use of synchronized wearables to determine when a cough monitor detects non-user coughs
Addressing privacy risks in clinical trials - edge computing and on-device cough analytics safeguard participant privacy, ensure regulatory compliance, and optimize clinical trial scalability.
16.06.2025
This paper addresses a critical limitation in wearable-based cough monitoring: distinguishing coughs emitted by the user versus those from nearby individuals (related: Other People's Coughs). In a controlled experiment with cohabiting couples (n=4), synchronized smartwatches running Hyfe's cough detection software captured cough events alongside their root mean square (RMS) acoustic energy—a proxy for source proximity.
Results showed that the device worn by the cougher detected 100% of events with zero false positives, while the non-cougher’s device detected 95% of events, also without false positives. Crucially, the RMS amplitude was consistently and significantly higher on the source device (median RMS: 0.0925 vs. 0.0235; p < 0.001), enabling 100% correct attribution of coughs using a simple energy threshold.
This method offers a scalable, non-intrusive alternative to contact microphones or manual annotation, making accurate source attribution viable in decentralized trials and real-world settings. By leveraging synchronized acoustic data alone, the approach enhances the scientific rigor of remote respiratory monitoring - especially in shared environments where attribution has historically been uncertain.
