Unsupervised phenotypic clustering for determining clinical status in children with cystic fibrosis

Abstract

Background: Cystic fibrosis (CF) is a multisystem disease in which the assessment of disease severity based on lung function alone may not be appropriate. The aim of the study was to develop a comprehensive machine-learning algorithm to assess clinical status independent of lung function in children.

Methods: A comprehensive prospectively collected clinical database (Toronto, Canada) was used to apply unsupervised cluster analysis. The defined clusters were then compared by current and future lung function, risk of future hospitalisation, and risk of future pulmonary exacerbation treated with oral antibiotics. A k-nearest-neighbours (KNN) algorithm was used to prospectively assign clusters. The methods were validated in a paediatric clinical CF dataset from Great Ormond Street Hospital (GOSH).

Results: The optimal cluster model identified four (A-D) phenotypic clusters based on 12 200 encounters from 530 individuals. Two clusters (A and B) consistent with mild disease were identified with high forced expiratory volume in 1 s (FEV₁), and low risk of both hospitalisation and pulmonary exacerbation treated with oral antibiotics. Two clusters (C and D) consistent with severe disease were also identified with low FEV₁. Cluster D had the shortest time to both hospitalisation and pulmonary exacerbation treated with oral antibiotics. The outcomes were consistent in 3124 encounters from 171 children at GOSH. The KNN cluster allocation error rate was low, at 2.5% (Toronto) and 3.5% (GOSH).

Conclusion: Machine learning derived phenotypic clusters can predict disease severity independent of lung function and could be used in conjunction with functional measures to predict future disease trajectories in CF patients.