Ensemble DNA methylation clock predicts all-cause mortality in Framingham and validation cohorts
A stacked ensemble survival model built from DNA methylation data from the Framingham Heart Study predicted all-cause mortality in an external validation cohort better than PhenoAge and comparably to GrimAge; results apply to European-ancestry and postmenopausal women aged 50–79. [1]
Automatic summary · not editorially reviewed
The key findings
The authors built a stacked ensemble survival model from Framingham Heart Study DNA methylation data, first selecting 190 CpG loci with elastic net Cox regression and then fusing five survival models via a neural network meta-learner. [1]
In an external validation cohort, the model predicted all-cause mortality significantly better than PhenoAge and statistically comparably to GrimAge. [1]
Performance estimates came from European-ancestry cohorts and were externally validated in postmenopausal women aged 50–79 years, so the authors limit applicability to demographically similar populations. [1]
The source at a glance
- Study type / format
- Abstract of a modeling study using a stacked ensemble survival model with external validation
- Population / context
- Framingham Heart Study DNA methylation data; external validation in European-ancestry cohorts and postmenopausal women aged 50–79 years.
- Summary based on
- Study abstract
- Original publication
- August 22, 2026
Framingham DNAm data and ensemble clock design
The authors note that most DNA methylation clocks use linear models, which may not capture the heterogeneity and non-linearity of survival-related methylation signatures. Using DNA methylation data from the Framingham Heart Study, they first identified 190 CpG loci with elastic net Cox regression. They then combined five complementary survival models into one prediction model using a neural network meta-learner.
Supporting source [1]
Mortality prediction versus PhenoAge and GrimAge
The model's predictive power was tested in an external validation cohort. There it showed strong performance for predicting all-cause mortality, significantly exceeding PhenoAge and statistically comparable to GrimAge. The outcome was all-cause mortality; the comparison was with two existing clocks, not a test of whether the clock causes or changes mortality.
Supporting source [1]
European-ancestry and postmenopausal women scope
The performance estimates came from cohorts of European ancestry and were externally validated in postmenopausal women aged 50–79 years. The authors state the results should be interpreted as applicable only to demographically similar populations. This abstract reports no follow-up duration, no effect sizes and no full-text details.
Supporting source [1]
How to interpret this
Abstract only; the full text was not reviewed. The abstract reports no follow-up duration, effect sizes or detailed model statistics, and the authors restrict applicability to European-ancestry and demographically similar populations, externally validated in postmenopausal women aged 50–79 years.
Based on the study abstract; the full paper, tables and supplementary data were not reviewed.
Source & supporting passages
Muthukumar YG, Vijayakumar KA, Cho GW. · europepmc.org · August 22, 2026