AI reveals pulmonary fibrosis is age-linked yet biologically distinct, pointing to new therapies

AI tools link pulmonary fibrosis to aging biology-without reducing IPF to "accelerated aging"

BUFFALO, NY - September 11, 2025 - A study in Aging-US (Volume 17, Issue 8; August 8, 2025) connects idiopathic pulmonary fibrosis (IPF) with aging biology using an AI-driven toolset. The work shows IPF is a distinct disease state influenced by age-related dysfunction, not a simple fast-forward of normal aging.

IPF primarily affects people over 60 and scars lung tissue, leading to breathlessness and, often, respiratory failure. Available drugs slow decline but rarely halt or reverse it. This study points to new targets and biomarkers that could change how research and clinical teams approach fibrosis.

"Our findings establish novel connections between aging biology and IPF pathogenesis while demonstrating the potential of AI-guided approaches in therapeutic development for age-related diseases."

Key findings

• Proteomic aging clock: An AI model trained on 55,319 Olink NPX proteomic profiles from the UK Biobank estimated biological age with high fidelity. Severe COVID-19 cases-at higher risk for lung fibrosis-showed accelerated aging, indicating fibrosis leaves a measurable biological trace.
• ipf-P3GPT transcriptomic comparator: Comparing aging lungs with IPF lungs revealed overlapping genes with opposing directions of change in more than half of shared cases. IPF alters core pathways differently than normal aging.
• Distinct molecular signatures: Both aging and IPF involve inflammation and tissue remodeling, but IPF drives more harmful changes to lung structure and repair systems-marking it as a separate biological state.
• Generalizable toolset: The AI approaches can be applied to other fibrotic conditions (e.g., liver, kidney), enabling broader investigations into age-linked disease processes.

Why this matters for researchers and clinicians

• Target discovery: Focus on IPF-specific nodes shared with aging but moving in opposite directions to reduce off-target effects on healthy aging processes.
• Biomarkers and trial design: Proteomic aging clocks can support patient stratification, endpoint selection, and monitoring of fibrosis burden or treatment response.
• Risk and screening: Accelerated biological aging signatures in severe COVID-19 suggest a path to identify high-risk groups for earlier intervention.

Methods in brief

• Proteomics: Aging clock trained on UK Biobank Olink NPX profiles (n=55,319) annotated with age and gender. The model quantifies biological age shifts relevant to fibrotic risk.
• Transcriptomics: Custom AI (ipf-P3GPT) contrasted gene activity in aging versus IPF lung tissue, highlighting divergent regulation and disease-specific pathways.

Implications and next steps

• Validate signatures across independent cohorts and longitudinal datasets to link biological age trajectories with clinical outcomes.
• Map divergent genes to druggable pathways and prioritize targets that modulate fibrosis without suppressing protective aging biology.
• Extend aging clock readouts to screening and post-viral fibrosis surveillance, with clear thresholds for action.

Citation

AI-driven toolset for IPF and aging research associates lung fibrosis with accelerated aging - Aging-US, Impact Journals LLC (Volume 17, Issue 8). DOI: https://doi.org/10.18632/aging.206295

Resources

• UK Biobank
• Idiopathic Pulmonary Fibrosis overview (NHLBI)

Research team

Researchers from Insilico Medicine-Fedor Galkin, Shan Chen, Alex Aliper, Alex Zhavoronkov, and Feng Ren-developed the proteomic aging clock and the ipf-P3GPT model to interrogate aging and IPF biology at scale.

For teams building AI capability

If you are setting up internal training for AI in biomedical data analysis, see job-focused options here: Complete AI Training - Courses by Job.