AI Model Improves Survival Prediction in Patients with Fibrotic Lung Disease- EMJ

AI Model Improves Survival Prediction in Patients with Fibrotic Lung Disease

AN AI-driven imaging model has demonstrated enhanced prognostic capability for fibrotic lung disease, outperforming conventional clinical and radiologic measures in a large retrospective study. The model, SABRE (Smart Airway Biomarker Recognition Engine), identifies structural airway features from high-resolution CT and was shown to independently predict disease progression and mortality.

The study analysed 1,744 patients, including 1,284 with diffuse fibrotic lung disease drawn from the Australian Idiopathic Pulmonary Fibrosis (IPF) Registry and Open Source Imaging Consortium (OSIC) datasets. SABRE was initially developed on a diverse training cohort of 460 patients with lung cancer, pneumonia, emphysema, or fibrosis. It quantitatively assessed airway branches by anatomic distribution and volumetric features. These outputs were then correlated with clinical outcomes using multivariable Cox regression analyses.

SABRE-derived variables remained statistically significant predictors of both mortality and disease progression even when adjusted for disease severity indices such as fibrosis extent, traction bronchiectasis, interstitial lung disease extent, and functional parameters including forced vital capacity (FVC%), diffusing capacity of the lungs for carbon monoxide (DLCO%), and composite physiologic index (CPI). Additionally, SABRE outperformed previously published deep learning models for fibrosis quantification and morphology.

Combining SABRE features with DLCO substantially improved prognostic discrimination, achieving an area under the curve of 0.852 for 1-year mortality and a concordance index (C-index) of 0.752. These results suggest SABRE captures mechanistic airway remodelling signals associated with disease pathogenesis and progression, not accounted for by traditional structural or functional metrics.

The authors conclude that SABRE-based airway phenotyping adds clinically meaningful information beyond current standards and may assist in risk stratification and therapeutic decision-making in fibrotic lung disease, including IPF. Further validation in prospective cohorts is needed, but the integration of AI-derived airway biomarkers may represent a step forward in precision medicine approaches for interstitial lung disease.

Reference

Nan Y et al. Prognostication in patients with idiopathic pulmonary fibrosis using quantitative airway analysis from HRCT: a retrospective study. Eur Respir J. 2025; DOI:10.1183/13993003.00981-2025.

Author:

Each article is made available under the terms of the Creative Commons Attribution-Non Commercial 4.0 License.

Rate this content's potential impact on patient outcomes

Average rating / 5. Vote count:

No votes so far! Be the first to rate this content.