Skip to main content
Fig. 1 | Respiratory Research

Fig. 1

From: Metabolomic fingerprinting and systemic inflammatory profiling of asthma COPD overlap (ACO)

Fig. 1

Orthogonal projections to latent structures discriminant analysis (OPLS-DA) model removes outliers which do not contribute to class separation. It is an extension to the supervised partial least squares (PLS) regression method that filters out some variance in the X-matrix unrelated to Y thereby producing results which are easier to interpret. OPLS-DA model shows optimized discrimination between (a) obstructive lung diseases and healthy controls (R2Y = 0.937 and Q2 = 0.919, CV-ANOVA score p = 0) (b) ACO and COPD (R2Y = 0.931 and Q2 = 0.877, CV-ANOVA score p = 0) (c) ACO and Asthma (R2Y = 0.95 and Q2 = 0.89, CV-ANOVA score p = 0). Parameters including R2 (goodness of the fit), Q2 (predictive ability), and analysis of variance testing of cross validated predictive residuals (CV-ANOVA) score were used to validate the robustness of the OPLS-DA model

Back to article page