Operationalizing mucosal biopsies using machine learning to determine lung allograft dysfunction

Lung allografts suffer from a high incidence of acute as well as chronic rejection. Due to exposure to the external milieu, lung allografts are also uniquely susceptible to damage from noxious stimuli. The diagnosis of allograft injury and differentiation from rejection requires transbronchial biopsy which is associated with severe complications, such as pneumothorax and bleeding, and is frequently inaccurate due to the heterogeneity observed in histopathology. The study by Halloran et al (1) attempts to operationalize machine learning based microarray analysis of pre-validated rejection-associated transcripts within mucosal biopsies, in lieu of transbronchial biopsies, to improve diagnostic accuracy and safety.