Stratification of asthmatic adolescents based on their short-term sensitivity to airborne particulates using sensor data from a pair of wearable devices

Health effects of air pollution on vulnerable populations is a concern for planetary health. A novel approach is presented for stratifying asthmatics based on their short-term sensitivity to airborne particulates using miniature wearable sensors coupled with causal discovery-based analysis [1] of 1-minute-level, time-series sensor data [2]. 137 asthmatic adolescents in Delhi were invited to wear for 48h: Airspeck [3] sensor (PM10/2.5/1), and Respeck [4] worn as a plaster on the chest (respiratory rate (RR) - breaths/minute, respiratory flow/effort and cough episodes [5]). The two sensor data streams were time-synchronised (1-minute resolution) and spatially-resolved (GPS) for causal discovery and machine learning-based analysis. For each subject a metric of their sensitivity to air pollution was introduced, by counting the number of causal links from PM2.5 exposures to respiratory rate in the 48h time-series sensor data: for 60-minute lags at 1-minute intervals, and for lags up to 8 hours at 15-minute intervals. The resulting stratification of short-term sensitivity of asthma subjects to PM2.5 into three categories: highly-, moderately-, and lightly-sensitive, was substantiated (attached table) by pulmonary function tests taken at the end of 48h data collection. RR could serve as proxy for lung function tests in clinical studies of air pollution effects on respiratory health in asthmatics. This could be used in primary care settings resulting in different care pathways for the three categories, without the need for spirometry tests using trained personnel. This approach could also serve as triage in the general population in low-resource LMIC settings to screen subjects highly-sensitive to airborne particulates who could be referred to tertiary centres to investigate their undiagnosed causes, possibly asthma, COPD, pulmonary TB or lung cancer. [1] doi: 10.1109/BSN51625.2021.9507030 [2] doi: 10.48550/arXiv.2301.06300 [3] doi: 10.1109/BHI.2018.8333376 [4] doi: 10.1093/bja/aer153 [5] Arvind DK et al “Monitoring coughs using a chest-wearable Respeck”, Proc. 2023 IEEE/EMBS BSN, Boston.