Cough monitoring during the covid-19 pandemic

Abstract

<b>Background:</b> The VitaloJAK cough monitor is an approved ambulatory medical device that records sound simultaneously from a chest wall contact microphone and free-field lapel microphone to quantify coughing in clinical trials. A digital signal processing algorithm (WH03_V3.0) utilises both microphones data (dual processing) to identify coughs and remove non-cough sounds, thereby reducing the time taken to count the number of coughs. The wearing of face masks/coverings during the Covid-19 pandemic distorts sounds from the lapel microphone but not the contact sensor. <b>Aim:</b> We investigated the utility of an algorithm using only the contact microphone data for cough detection and removal of non-cough sounds (WH03_V3.03). <b>Methods:</b> 24h sound recordings from 60 patients with refractory chronic cough underwent both single and dual channel processing. The performance of each algorithm was assessed by the ability to retain coughs pre-identified by manual counting of the 24h recordings (sensitivity) and filter out non-cough sounds (reduction in recording length). <b>Results:</b> Dual channel processing had a median sensitivity of 99.7%(IQR 99.0-100) and shortened the 24h file to a median 114.5mins(154.8-70.8), 7.9% of the 24h. Single channel processing (contact sensor only) maintained sensitivity, median 100%(99.7-100.0), but reduced filtering efficiency, file length 197.3mins(151.5-262.7), 13.7%. Single channel processing has been implemented in up to 50% of monthly cough analysis during the pandemic, maintaining data integrity ≥98%. <b>Conclusion:</b> Processing of contact microphone data is an acceptable alternative for the VitaloJAK cough monitoring system, however the accuracy of systems reliant on free-field microphones may be compromised during the pandemic.