Research Paper Advance Articles

Figure 1. Deep learning models for studying IPF and fibrotic diseases. (A) This paper presents two deep learning models: an omics-transformer that generates differential gene expression profiles from text prompts and a pathway-aware aging clock trained on UK Biobank proteomics data. The models focus on IPF-relevant biological pathways including TGF-β signaling, oxidative stress, inflammation, and ECM remodeling. (B) Architecture of the pathway-aware proteomic aging clock. The neural network processes protein measurements through feature extraction layers that branch into age prediction and pathway-specific attention mechanisms, enabling interpretable aging predictions with pathway awareness.