B24-24 Continuous Cough Monitoring for Early Detection of COPD Exacerbations Using Hyfe (COPD-HYFE): Interim Results From a Prospective Multi-center Cohort Study

Rationale

COPD exacerbations drive disease progression, hospitalization, and mortality, yet early detection remains limited by reliance on subjective symptom reporting. Cough, a hallmark symptom of COPD, may provide an objective, continuous biomarker for exacerbation prediction. However, longitudinal cough patterns in COPD have never been characterized using continuous 24/7 monitoring. We present interim findings from an ongoing prospective study evaluating passive cough monitoring using the Hyfe CoughMonitor, a validated AI-powered wearable system, to characterize cough variability in COPD and assess its potential for early exacerbation detection.

Methods

This prospective, multi-center, observational cohort study (NCT07212439) recruits COPD patients across clinical sites in Navarra, Spain. Participants wear the Hyfe CoughMonitor during waking hours and charge it bedside overnight, where it continues cough monitoring. A convolutional neural network classifies cough events on-device, transmitting only timestamps to fully preserve privacy - no audio or conversations can be listened to. Weekly CAT questionnaires and paraclinical data (spirometry, imaging, 6-min walking test) supplement continuous cough monitoring. Clinicians remain blinded to cough rates and study-administered CAT scores; exacerbation diagnosis follows standard care using GOLD 2025 criteria. The study targets 50 participants contributing over 100,000 monitoring hours. We report interim analyses of cough rate variability, adherence, and temporal patterns surrounding clinically-confirmed exacerbations.

Results

To date, 11 participants have contributed 16,201 hours of cough monitoring with 98.5% adherence. Average hourly cough rates varied substantially between individuals (range: 1.2-16.0 coughs per hour [cph]) and were higher during daytime (6.4 cph) than nighttime (2.6 cph). Cough rates showed minimal variation across GOLD stages (range: 5.1-5.5 cph) but greater heterogeneity across BODE scores (range: 3.0-10.5 cph). Five exacerbations occurred in five participants (4 severe, 1 moderate). Hourly cough rate trajectories around exacerbation events reveal heterogeneous individual patterns, with several participants demonstrating apparent increases in cough frequency in the days preceding clinical presentation (Figure 1). Marked within-individual day-to-day variability was observed, representing the first continuous characterization of cough dynamics in COPD.

Conclusions

Continuous passive cough monitoring in COPD is feasible with excellent adherence. Preliminary data reveal substantial inter-individual and day-to-day cough variability, previously undescribed in this population. The observation that cough frequency appears independent of GOLD stage suggests cough burden may reflect disease dimensions beyond airflow limitation. Early findings indicate detectable cough pattern changes preceding exacerbations. Completion of enrollment will enable determining whether cough monitoring provides clinically meaningful lead time for intervention.

This abstract is funded by: Hyfe Inc.