Researchers have developed an algorithm that groups patients with chronic obstructive pulmonary disease (COPD) into clinically relevant phenotypes, which may aid the exploration of underlying disease mechanisms and improve the development of novel care strategies.
Published today (2 November, 2017) in the European Respiratory Journal, the study authors claim the algorithm is the first to integrate comorbidities including cardiovascular diseases, diabetes and obesity with age, as well as traditional respiratory variables such as FEV1 and dyspnoea.
Researchers conducted cluster analysis using data from 2,409 COPD patients that identified five COPD subgroups. Classification and regression tree (CART) analysis was then used to develop an algorithm that allocated patients into five classes; these classes corresponded to the five subgroups identified by the cluster analysis.
The algorithm was further tested using patient data from the COPD Cohorts Collaborative International Assessment (3CIA) initiative; application to the 3CIA cohort confirmed that the algorithm was able to consistently identify subgroups of patients with different clinical characteristics.
The authors speculate that the algorithm may be beneficial in helping to identify biological pathways that may have previously been missed and in developing phenotype-specific therapeutic strategies.
The article is accompanied by an editorial also published in the ERJ, written by Rosa Faner and Alvar Agustí, which discusses the study and debates the role of algorithms in helping clinicians to prescribe treatments.