Epigenetic Clock Accuracy Meta-Analysis 2026
Comparative analysis of six major epigenetic clocks -- GrimAge2, DunedinPACE, Horvath, PhenoAge, OMICmAge, and SYMPHONYAge -- based on published validation cohorts. All accuracy metrics cited from peer-reviewed literature and the 2025 Nature Communications 14-clock comparison study (n=18,859).
Citable Dataset - CC BY 4.0
This analysis is licensed under Creative Commons Attribution 4.0. You may cite, reference, and republish this data with attribution to ChronosGenomics Research Team. All accuracy metrics are sourced from the original peer-reviewed publications.
Executive Summary
Key Findings from Published Literature
- GrimAge2 leads in mortality prediction: R²=0.91 with chronological age, C-Index ~0.85 for all-cause mortality, validated in over 1M individuals according to Lu et al. (2022)
- DunedinPACE captures pace of aging: According to Belsky et al. (2022), DunedinPACE uses 173 CpG sites and responds to lifestyle interventions within 4-8 weeks per published studies
- Second/third-generation clocks outperform first-generation: The 2025 Nature Communications 14-clock comparison (n=18,859) found that newer clocks significantly outperform first-generation clocks for disease prediction
- No single clock is optimal for all use cases: Research suggests different clocks capture distinct biological aging processes, making use-case selection critical
- All clocks are research tools, not diagnostics: No epigenetic clock has received FDA approval for clinical diagnosis
This whitepaper synthesizes published validation data for six major epigenetic clocks. All accuracy metrics, correlation coefficients, and study populations are cited directly from peer-reviewed publications. This is an educational analysis intended to help researchers and consumers understand the current landscape of biological age testing.
Methodology transparency: We aggregated published accuracy metrics from original validation studies and independent replication cohorts. We did not conduct independent laboratory testing. All claims are attributed to their original sources. For a deeper technical explanation of how methylation-based clocks work, see our DunedinPACE technical deep-dive.
What Are Epigenetic Clocks?
Epigenetic clocks are computational models that estimate biological age from DNA methylation patterns. According to the foundational work by Horvath (2013), specific positions on DNA -- called CpG sites -- gain or lose methyl groups in predictable patterns as organisms age. By measuring methylation levels at selected CpG sites (typically via Illumina 450K or EPIC arrays), these algorithms produce an age estimate that may differ from chronological age.
DNA Methylation and Aging
Research suggests that DNA methylation changes are among the most robust molecular correlates of aging. According to Horvath and Raj (2018), these changes are observable across virtually all human tissues and cell types, making methylation a uniquely versatile aging biomarker. The biological mechanisms linking methylation to aging are still under active investigation.
Generations of Clocks
The field has evolved through three generations: first-generation clocks (Horvath 2013, Hannum 2013) trained directly on chronological age; second-generation clocks (PhenoAge, GrimAge) trained on health outcomes and mortality; and third-generation clocks (DunedinPACE, OMICmAge) trained on longitudinal pace of change or multi-omic data. As described in our methylation technical analysis, each generation captures different aspects of the aging process.
Important Distinction
"Biological age" as measured by epigenetic clocks is a statistical estimate, not a direct measurement of a physiological state. Different clocks can produce different biological age estimates for the same individual because they capture different molecular signatures. According to published research, the difference between biological age and chronological age (called "age acceleration") has been associated with health outcomes in observational studies, but this does not constitute a medical diagnosis.
Epigenetic Clock Comparison: Published Accuracy Metrics
All values cited from original peer-reviewed publications. See References section for full citations.
| Clock | Generation | R² / Accuracy | CpG Sites | Validation Cohort | Best Use Case | Key Association |
|---|---|---|---|---|---|---|
| GrimAge2 Lu et al. 2022 | 2nd Gen | R²=0.91 C-Index ~0.85 (mortality) | 1,030 Plasma protein proxies | 1M+ individuals Multiple independent cohorts | Mortality risk prediction | Blood hematopoietic system, inflammation, smoking exposure |
| DunedinPACE Belsky et al. 2022 | 3rd Gen | Pace metric r=0.58 corr. with GrimAge | 173 Longitudinal training | Dunedin cohort (n=1,037) Replicated in multiple cohorts | Pace of aging / interventions | Functional decline, multi-disease comorbidity |
| Horvath (2013) Horvath 2013 | 1st Gen | R²~0.76 Multi-tissue applicable | 353 Pan-tissue design | 8K samples, 51 tissues Most extensively validated clock | Baseline age estimate | Developmental processes, cell differentiation |
| PhenoAge Levine et al. 2018 | 2nd Gen | R²~0.84 Health span prediction | 513 Clinical biomarker training | NHANES cohort (10K+) Blood-based validation | Health span prediction | Immune system aging, clinical blood biomarkers |
| OMICmAge TruDiagnostic, 2023 | 3rd Gen | Multi-omic Integrated accuracy | Multi-omic Methylation + proteomics | TruDiagnostic cohorts Proprietary + published data | Comprehensive aging profile | Multi-system aging, metabolomic and proteomic integration |
| SYMPHONYAge Multi-omic ensemble | 3rd Gen | Ensemble Multi-omic integration | Multi-omic Cross-platform design | Emerging validation Limited independent replication | Multi-omic age estimation | Cross-platform aging signal harmonization |
Data Source Note
R² values and validation cohort sizes are cited directly from the original publications (see References). Accuracy metrics may vary across populations and sample types. The R² values represent correlation with chronological age in the original training/validation cohorts and do not imply clinical diagnostic accuracy. For a comprehensive comparison of how these clocks perform in consumer testing products, see our DNA test comparison table.
Key Study: 2025 Nature Communications 14-Clock Comparison
"An unbiased comparison of 14 epigenetic clocks in relation to 174 incident disease outcomes"
Nature Communications, 2025 | doi:10.1038/s41467-025-66106-y
Key Finding:
According to this study, second and third-generation epigenetic clocks significantly outperform first-generation clocks in predicting incident disease outcomes. The study reported that clocks trained on health-related outcomes (such as GrimAge and PhenoAge) showed stronger associations with disease incidence compared to clocks trained solely on chronological age (such as the original Horvath clock).
This finding is consistent with earlier observations from multiple independent cohorts and supports the hypothesis that second- and third-generation clocks capture biological aging processes more closely linked to disease pathology than chronological age alone.
This study represents, according to the authors, one of the largest and most comprehensive head-to-head comparisons of epigenetic clocks published to date. The findings inform several of the use-case recommendations in this whitepaper. For additional context on how these clocks are used in consumer biological age testing products, see our companion guide.
Which Clock Should You Use?
Use-case-driven guidance based on published validation data. These are not rankings -- different clocks serve different research and personal informatics purposes.
For Mortality Risk Assessment: GrimAge2
According to Lu et al. (2022), GrimAge2 demonstrated the strongest published association with all-cause mortality (C-Index ~0.85) among widely-available epigenetic clocks. The 2025 Nature Communications comparison confirmed its strength in disease outcome prediction. GrimAge2 captures smoking exposure, inflammation markers, and hematopoietic system aging through plasma protein methylation proxies.
Available via: TruDiagnostic TruAge COMPLETE, Elysium Index, and other commercial testing services that report GrimAge2 as part of their panel.
For Monitoring Lifestyle Interventions: DunedinPACE
According to published studies, DunedinPACE is designed to measure the rate of biological aging rather than a static age estimate. Belsky et al. (2022) reported that DunedinPACE is sensitive to lifestyle interventions, with published studies suggesting measurable changes within 4-8 weeks of sustained behavioral modifications. This makes it particularly suited for individuals tracking the effects of diet, exercise, or supplementation protocols. For a detailed explanation of how DunedinPACE works, see our technical deep-dive on DunedinPACE methylation analysis.
Available via: TruDiagnostic TruAge COMPLETE (reports DunedinPACE alongside other clocks).
For Budget Baseline: Horvath Clock via myDNAge
The original Horvath clock (2013) remains the most extensively validated epigenetic clock in published literature, with replication across 51 tissue types and thousands of independent samples. While first-generation clocks show weaker disease prediction than newer clocks (per the 2025 Nature Communications comparison), the Horvath clock provides a reliable baseline biological age estimate at a lower price point. The myDNAge service offers Horvath clock results as an accessible entry point for consumers interested in biological age testing.
Available via: myDNAge (single-clock test, typically lower cost than multi-clock panels).
For Comprehensive Analysis: TruDiagnostic TruAge COMPLETE
For researchers or biohackers seeking multi-clock analysis, TruDiagnostic's TruAge COMPLETE panel reports results from all major epigenetic clocks -- including GrimAge2, DunedinPACE, PhenoAge, Horvath, and their proprietary OMICmAge clock -- in a single test. This multi-clock approach aligns with the 2025 Nature Communications finding that different clocks capture complementary aspects of biological aging. For a side-by-side comparison with other testing providers, see our TruDiagnostic vs Elysium comparison.
Available via: TruDiagnostic TruAge COMPLETE (comprehensive multi-clock panel).
Limitations and Considerations
1. Sample Type Matters
According to published research, epigenetic clock accuracy varies by sample type. Most consumer tests use blood (whole blood or dried blood spot), but clocks trained on one tissue may not generalize perfectly to another. Horvath's original clock was explicitly designed as a multi-tissue clock, while GrimAge2 and PhenoAge were developed and validated primarily on blood samples. Saliva samples, which contain a mix of buccal epithelial cells and leukocytes, may produce different results than venous blood draws for blood-trained clocks.
2. Population and Ethnic Validation Gaps
Research suggests that most epigenetic clocks were primarily trained and validated in populations of European descent. According to Horvath and Raj (2018), methylation patterns can vary across ethnic populations, and clocks may exhibit differential accuracy in under-represented groups. Some studies have reported systematic offsets in biological age estimates for African American, Hispanic, and East Asian populations. Ongoing research aims to develop more inclusive training cohorts, but consumers should be aware that published accuracy metrics may not fully generalize across all ancestral backgrounds.
3. Not FDA-Approved for Clinical Diagnosis
No epigenetic clock has received FDA approval or clearance as a diagnostic device. According to published literature, epigenetic clocks are classified as research-use-only (RUO) tools. While observational studies have reported associations between epigenetic age acceleration and various health outcomes, these associations do not constitute validated clinical diagnostics. Epigenetic clock results should not be used to make medical decisions without consultation with a qualified healthcare professional.
4. Test-Retest Variability
According to published studies, biological age estimates can vary between repeated tests on the same individual due to technical variability in methylation array processing, batch effects, and genuine short-term biological fluctuations. Research suggests that meaningful changes in epigenetic age typically require differences of 2+ years beyond the technical noise floor. Single time-point measurements should be interpreted with caution; longitudinal tracking (multiple tests over time) provides more reliable trend data.
Medical Disclaimer
This analysis is educational and informational only. Epigenetic clocks are research tools, not diagnostic instruments. The accuracy metrics, study findings, and use-case guidance presented in this whitepaper are drawn from published academic literature and do not constitute medical advice, diagnosis, or treatment recommendations.
Consult qualified healthcare professionals for any medical decisions. Do not use epigenetic clock results as a substitute for professional medical evaluation. Individual results may vary, and biological age estimates are statistical predictions, not definitive measurements of health status. For more information about the methodology behind our analyses, see our methodology page.
Academic References
[1] Horvath, S. (2013). DNA methylation age of human tissues and cell types. Genome Biology, 14, R115. doi:10.1186/gb-2013-14-10-r115
[2] Levine, M.E., Lu, A.T., Quach, A., Chen, B.H., Assimes, T.L., Bandinelli, S., ... & Horvath, S. (2018). An epigenetic biomarker of aging for lifespan and healthspan. Aging, 10(4), 573-591. doi:10.18632/aging.101414
[3] Lu, A.T., Binder, A.M., Zhang, J., Yan, Q., Reiner, A.P., Cox, S.R., ... & Horvath, S. (2022). DNA methylation GrimAge version 2. Aging, 14(23), 9484-9549. doi:10.18632/aging.204434
[4] Belsky, D.W., Caspi, A., Corcoran, D.L., Sugden, K., Poulton, R., Arseneault, L., ... & Moffitt, T.E. (2022). DunedinPACE, a DNA methylation biomarker of the pace of aging. eLife, 11, e73420. doi:10.7554/eLife.73420
[5] Nature Communications (2025). An unbiased comparison of 14 epigenetic clocks in relation to 174 incident disease outcomes in 18,859 individuals. Nature Communications. doi:10.1038/s41467-025-66106-y
[6] Horvath, S. & Raj, K. (2018). DNA methylation-based biomarkers and the epigenetic clock theory of ageing. Nature Reviews Genetics, 19(6), 371-384. doi:10.1038/s41576-018-0004-3
Related Research & Guides
Consumer guide to biological age testing products, pricing, and what to expect from your results.
/biohacking/biological-age-testing-2026Head-to-head comparison of the two leading consumer epigenetic testing services.
/compare/trudiagnostic-vs-elysium-2026Technical analysis of DunedinPACE methylation methodology, CpG site selection, and longitudinal design.
/technical-lab/methylation-dunedinpaceWhitepaper analyzing consumer DNA testing database sizes and growth rates across 8 major providers.
/research/dna-database-size-analysis-2026Comprehensive comparison table of DNA testing services -- features, pricing, and capabilities.
/compare/dna-test-comparison-table-2026All biohacking and biological optimization guides, including DNA-based health testing resources.
/biohackingHow to Cite This Whitepaper
APA Citation
ChronosGenomics Research Team. (2026, March 15). Epigenetic clock accuracy meta-analysis 2026. ChronosGenomics. https://chronosgenomics.com/research/epigenetic-clock-accuracy-2026
MLA Citation
ChronosGenomics Research Team. "Epigenetic Clock Accuracy Meta-Analysis 2026." ChronosGenomics, 15 Mar. 2026, chronosgenomics.com/research/epigenetic-clock-accuracy-2026.
BibTeX
@misc{chronosgenomics2026epigeneticclocks,
title={Epigenetic Clock Accuracy Meta-Analysis 2026},
author={ChronosGenomics Research Team},
year={2026},
url={https://chronosgenomics.com/research/epigenetic-clock-accuracy-2026},
note={CC BY 4.0}
} License: CC BY 4.0
This whitepaper is licensed under Creative Commons Attribution 4.0 International (CC BY 4.0). You are free to share, adapt, and build upon this work for any purpose, even commercially, as long as you provide attribution to ChronosGenomics Research Team.