Approaches to identifying phenotypes or endotypes in asthma have become increasingly relevant. However, in the majority of published approaches, the characterising parameters are only assessed at a single point in time, yielding phenotypes that might not remain stable as time progresses. We hypothesised that we could identify asthma and functional healthy phenotypes by investigating the patterns of fluctuation in airway function measured over a predetermined, sufficiently long time window of observation.

We have developed and applied a computational data-driven method that allows us to classify healthy individuals and different types of asthmatic patients according to the fluctuation patterns in their lung function. By applying this methodology to three different patient cohorts, we were able to explore the potential clinical usefulness of our approach. Indeed, we found evidence for the existence of subtypes of asthma patients, who, if properly identified, would benefit from therapeutic strategies that differ from the commonly used anti-inflammatory treatment schemes.