A NEW study finds that machine learning models using data from digital inhalers can successfully predict chronic obstructive pulmonary disease (COPD) exacerbations up to 5 days in advance, offering a promising tool for proactive disease management.
COPD exacerbations are a major driver of morbidity and hospitalisation. Timely identification of impending events could allow for earlier intervention and improved outcomes. In this 12-week, open-label clinical trial, patients aged 40 years and older with COPD were prescribed ProAir Digihaler, a digital dry powder inhaler equipped with sensors to monitor reliever medication use. The device continuously recorded inhalation metrics including peak inspiratory flow (PIF), inhalation volume and duration, time to PIF, and timestamps. Researchers used this high-resolution data alongside clinical and demographic variables to develop a machine learning model to predict exacerbations.
Of the 336 patients included in the analysis, 98 experienced 111 exacerbations. Prior to exacerbation events, a decline in PIF and inhalation volume was consistently observed. A predictive model using gradient-boosting trees achieved a receiver operating characteristic area under the curve (AUC) of 0.77 (95% CI: 0.71–0.83) for forecasting exacerbations within a 5-day window. The most influential features in the model included both baseline inhalation metrics and changes from baseline prior to the event, highlighting the clinical value of longitudinal inhalation tracking.
These findings establish proof of concept for using data from digital reliever inhalers to anticipate clinical deterioration in COPD. With further validation, these models could be integrated into routine care to enhance disease monitoring, reduce emergency interventions, and support more personalised management strategies.
Reference
Snyder LD et al. Predictive machine learning algorithm for COPD exacerbations using a digital inhaler with integrated sensors. BMJ Open Respir Res. 2025;12(1):e002577.
