A deep learning score for cellular senescence predicts human mortality and disease risk
The proteomic biomarker tracked chronic illness risk in the UK Biobank and shifted following an 18-month multimodal exercise intervention.
Aging cell · Zhao S et al. · Paper published 26 Sep 2026
In human participants from the UK Biobank and an independent randomized clinical trial, researchers developed and validated a deep learning biomarker of cellular senescence. The team integrated curated senescence-associated secretory phenotype (SASP) proteins from blood proteomics data using a Guided autoencoder with Transformer model. The resulting composite SASP Score strongly and independently predicted mortality risk and incident chronic conditions, including dementia, chronic obstructive pulmonary disease, myocardial infarction, and stroke. In an independent randomized trial cohort, an 18-month multimodal exercise program significantly altered the trajectory of the SASP Score. The authors note the framework offers cross-platform utility for tracking senescence burden.
Why it matters
Blood-based SASP measurements provide a noninvasive proxy for systemic cellular senescence burden. Establishing such biomarkers is essential for tracking biological aging and evaluating the success of geroscience-guided interventions in clinical trials.
Caveats
The abstract does not report the sample sizes, participant demographics, or specific composition of the multimodal exercise intervention. Additional studies are needed to determine how well the score generalizes across other diverse populations and clinical settings.
Written from the paper’s abstract, and every claim checked against it before publishing. Read the paper for the full methods and data.
The paper
A Deep-Learning Based Biomarker of Systemic Cellular Senescence Burden to Predict Mortality and Health Outcomes
Zhao S, Kuo CL, Lenze EJ et al.
Aging cell · 26 Sep 2026 · Peer-reviewed
- Relevance
- Core geroscience
- News value
- Important
- Evidence
- Human trial
- Status
- Peer-reviewed
More on Cellular senescence
See allDasatinib and quercetin reduce liver fibrosis in a phase-2 MASH trial
The senolytic combination improved liver fibrosis without disease worsening in nearly half of treated participants while lowering cellular senescence markers.
Nature metabolism · Koning M et al.
A distinct p21-driven macrophage senotype accumulates during aging and liver disease
Researchers identified a p53-p21-Cyclin D2 axis that governs senescent macrophages in mice and human liver cirrhosis and can be cleared with senolytics.
bioRxiv · Torres G et al.
DNA methylation proxies of senescence markers track human disease and mortality
The new blood-based tools estimate senescence gene activity to predict incident diseases, all-cause mortality, and clinical treatment responses.
medRxiv · Simpson DJ et al.
Bispecific T cell engager eliminates senescent cells in mice and nonhuman primates
Monitored by serum transaminase levels, low doses of the drug alleviated age-related pathologies without causing the hepatotoxicity seen at higher doses.
Aging cell · Deng J et al.
Lactate receptor activation reduces cellular senescence and protects vascular health in mice
Loss of the lactate receptor GPR81 drives arterial senescence and lipid accumulation, while pharmacological stimulation restores vascular homeostasis in progeroid mice.
bioRxiv · Wu Y et al.