Skip to main content
. 2019 Dec 28;157(5):1147–1157. doi: 10.1016/j.chest.2019.11.039

Table 2.

Comparison of K-Means Clustering and NBS Clustering Results Shows Little Overlap

Variable NBS1 NBS2 NBS3 NBS4
Relatively resistant smokers 20 10 3 2
Upper lobe predominant emphysema 5 9 1 1
Airway predominant 15 11 0 3
Severe COPD 15 13 1 11

For 120 COPDGene subjects analyzed in both the Castaldi et al18 phenotype clustering article and the Chang et al31 gene expression clustering article, the overlap between clustering assignments was modest. Network-based stratification (NBS) clusters are ordered as in the original manuscript by average level of FEV1 (ie, NBS1 has highest average FEV1).