AI Deciphers Disease Mechanisms in Pulmonary Fibrosis

A new international study provides groundbreaking insights into the decoding of cellular processes in complex lung diseases. The research team presents an innovative AI-based model that analyzes the temporal development of cells during disease progression and can predict potential new therapeutic approaches. The publication, appearing in Nature Biomedical Engineering, was authored as co-first authors by Yumin Zheng and DZL researcher Prof. Dr. Jonas Schupp (DZL site BREATH).

The study introduces the UNAGI model (“Unsupervised Neural Analysis of Gene Interactions”) – a generative deep learning method that, for the first time, leverages longitudinal single-cell transcriptome data to reconstruct cellular dynamics over the course of chronic diseases. While classical analyses provide only static snapshots, UNAGI allows a temporally resolved view of disease progression. High-dimensional transcriptome data are used to generate so-called disease-informed embeddings, which map functional cell states according to disease stages.

Using data from idiopathic pulmonary fibrosis (IPF), the researchers demonstrated that UNAGI can accurately distinguish between early, intermediate, and late disease stages at the cellular level. The model identified fibroblastic subpopulations associated with progressive tissue remodeling and traced their transcriptional programs over the course of the disease. This information enables the identification of regulatory networks critical for the transition from reversible to irreversible fibrotic states.

Particularly innovative is the combination of modeling cellular dynamics with an in silico drug screening approach. Based on the generated disease profiles, UNAGI can predict which pharmacological compounds may modulate specific pathological cell states. Potential antifibrotic candidates were experimentally tested in human lung precision-cut tissue cultures and validated by measurable reductions in fibrotic markers.

“With approaches like these, we bring cellular mechanisms—and ultimately therapeutic options—much closer to the patient,” emphasizes Prof. Dr. Schupp. The findings highlight the potential of modern AI-driven methods not only to describe disease mechanisms at the single-cell level but also to make them therapeutically actionable.

Source: BREATH

Original publication: Zheng Y, Schupp JC, Adams T, Clair G, Justet A, Ahangari F, Yan X, Hansen P, Carlon M, Cortesi E, Vermant M, Vos R, De Sadeleer LJ, Rosas IO, Pineda R, Sembrat J, Königshoff M, McDonough JE, Vanaudenaerde BM, Wuyts WA, Kaminski N, Ding J. A deep generative model for deciphering cellular dynamics and in silico drug discovery in complex diseases. Nat Biomed Eng. 2025 Jun 20. doi: 10.1038/s41551-025-01423-7. Epub ahead of print. PMID: 40542107.