Abstract Rationale Early detection of asthma attacks in children is limited by subjective symptom reporting by the carer or the child. Moreover, the time taken for resolution of symptoms post attack is unknown. The Albus device - a contactless bedside device which continuously monitors nocturnal respiratory parameters - can detect asthma attacks early and their resolution by tracking changes in respiratory rate and cough frequency (Nagakumar et al, 2024). We hypothesised that the Albus Home device can detect changes in a novel abnormal breathing metric pre and post attacks in children. Methods A new ‘abnormal breathing’ metric was defined as any added or audible sounds of expiration, including wheezing, rattling or coarse sounds. Children aged 6-16 years with asthma were recruited into the Childhood Home Asthma Monitoring Study (CHAMP). Participants with 3 months data meeting the quality control criteria were included. Changes in nocturnal abnormal breathing sound counts were analysed as the daily aggregate during two weeks before and after an asthma attack. Counts were automatically detected from audio recorded from the Albus Home. The data for each attack is normalised between 0 and 1 to focus on the trends and eliminate difference in baseline counts between participants. Asthma attack was defined as a course of systemic steroids taken for asthma, and the attack day defined as the first day of starting steroids. Steroid courses which occurred within 3 weeks of each other were treated as one event, with the first pre-attack and last post-attack period included in analysis. Results Forty-nine attacks from 24 children (17 males; mean ±SD age of 10.2 ±2.8 years) met data requirements before and after an asthma attack. There was increased nocturnal abnormal breathing sounds in the two-week period leading up to an attack (Fig. 1a). Following an attack, a decrease was seen but with higher variability (Fig. 1b). Conclusion We demonstrated previously the Albus Home device detected increased respiratory rate and cough prior to asthma attacks in children. The novel metric presented here provides an additional parameter to detect attacks and monitor subsequent resolution. This could help in developing a composite multi-metric score for detecting attacks early, providing a potential therapeutic window to prevent asthma attacks, as well as monitoring responses to treatment. This abstract is funded by: Artificial Intelligence in Health and Care Award (NIHR, NHS AI Lab and Department of Health and Social Care, U.K.)
Nagakumar et al. (Fri,) studied this question.