ABSTRACT
Background: Viral and bacterial pneumonia are characterized by exaggerated inflammation, dyspnea, and cough. We investigated cough evolution in hospitalized patients with COVID-19 and non-COVID-19 pneumonia using a novel, automated cough count detector. Material(s) and Method(s): In this observational cohort study, we used an automated, machine learning-based, contact-free, near real-time, smartphone-enabled cough counter to quantify cough in 46 patients with pneumonia in a non-ICU setting. Cough frequencies were correlated to clinical and laboratory markers of pneumonia disease activity. Result(s): We observed steady declines of cough frequencies during hospital stay in both COVID-19 and non-COVID19 pneumonia. Cough frequencies were associated with the degree of oxygenation/ oxygen supplementation (ROX index, FiO2, SpO2, breathing rate) as well as markers of inflammation (ferritin, CRP, LDH, body temperature) (Figure 1). No association with markers of multi-systemic disease (ASAT, ALAT, D-dimer) was found. Conclusion(s): Mobile technology leverages cough detection and allows for unobtrusive, long-term monitoring of patients in aerosol isolation. Results from this study suggest that cough frequency represents a surrogate marker of pneumonia disease activity. Future studies are warranted to assess cough frequency as a clinically actionable digital biomarker for lower respiratory tract infections.