A Machine Learning Analysis of Physiological Monitoring Signals to Detect Small Airway Narrowing Due to Cold Air Exposure in Asthma.

Abstract

Asthma is a chronic inflammatory disease of the small airways, affecting over 200 million people globally. Cold air exposure is a potential risk factor for asthma exacerbations. We hypothesized that monitoring physiological signals during exposure to cold would help to detect potential worsening in asthma and can be used to help persons with asthma adjust their daily routine. Non-smoker adults (18-80 years) with asthma were asked to sit in a cold room of 0°C temperature for 10 minutes. During this period, Electrocardiogram (ECG) and thoraco-abdominal motion/respiration belt signals were measured continuously. At 0 and 10 min, small airway narrowing was assessed with oscillometry to estimate respiratory system impedance. Based on changes in respiratory impedance from 0 to 10 min, participants were grouped into with or without airway narrowing. After signal processing, we extracted time and frequency domain features from ECG and respiration signals. To classify airway narrowing, different machine learning classifiers were fine-tuned and evaluated using a leave-one-subject-out cross-validation approach. A total of 23 individuals (11 females, age: 56.3 ± 10.9 years, BMI: 27.4 ± 5.7 kg/m$^{2}$) with asthma were enrolled in the study. Up to 42% and 58% windows of signals were from individuals with and without airway narrowing, respectively. The support vector machine classifier performed the best compared to other models with an accuracy of 85%, precision of 87%, recall of 76%, specificity of 91%, and F1 score of 81%. These results provided proof of concept that technologies with embedded respiratory and cardiac signal monitoring may be able to predict airway narrowing during exposure to cold air in individuals with asthma.