Smartphone-based cough detection predicts asthma control – description of a novel, scalable digital biomarker

Abstract

Introduction: The frequency of nocturnal cough might be a predictor for asthma control. Information regarding longitudinal trends and cause-effect relationships of nocturnal cough in asthma is lacking. This study investigated nocturnal cough in asthmatic individuals over the course of 4 weeks by means of a smartphone and explored the interplay between nocturnal cough frequency and asthma control. Methods: Machine learning models for nocturnal cough detection were developed. 94 patients in 2 centers were recruited; study duration per patient was 29 days with in-person appointments on first and last study day. In between, patient-reported outcomes and nocturnal sensor data were collected by a smartphone with a chat-based study app. Asthma control was assessed by weekly ACT (Asthma Control Test). Separate mixed effect models were calculated to understand whether nocturnal cough could predict asthma control. Results: The mixed effect model demonstrates that weekly aggregated measurements of nocturnal cough were associated with asthma control of the same and the subsequent week. An increase of 100 coughs per week was associated with a decrease of 0.56 points in ACT for this week (p=0) and a decrease of 0.25 points in ACT for the following week (p=0.024). Conclusion: Our study is the first to describe the independent predictive value of changes of nocturnal cough frequency for asthma control. Nocturnal cough could therefore be a useful parameter for timely medical interventions to prevent asthma deteriorations. By using conventional smartphones, this study lays the groundwork for scalable, validated and broadly available digital biomarkers for asthma.