Clocks
Characteristics of the 100 benchmarked clocks, from Supplementary Table 1.
100 of 100 clocksDownload CSV
| Koch (5 CpG) | Koch | 2011 | Multiple tissues | 5 | Chronological age | 16-72 | Multi-dataset | 130.0 | Linear regression | 27k |
|---|---|---|---|---|---|---|---|---|---|---|
| Hannum clock (69 CpG) | Hannum | 2013 | Blood | 69 | Chronological age | 19-101 | GSE40279 | 482 | Elastic net regression | 450k |
| Hannum clock (88 CpG) | Hannum | 2013 | Blood | 88 | Chronological age | 19-101 | GSE40279 | 656 | Elastic net regression | 450k |
| Horvath clock (353 CpG) | Horvath | 2013 | Multiple tissues | 353 | Chronological age | −0.5-100 (training) | Multi-dataset | 3931 (training); 7844 (total) | Elastic net regression | 27k, 450k |
| Horvath clock (110 CpG) | Horvath | 2013 | Multiple tissues | 110 | Chronological age | −0.5-100 (training) | Multi-dataset | 3931 (training); 7844 (total) | Elastic net regression | 27k, 450k |
| Weidner (3 CpG) | Weidner | 2014 | Blood | 3 | Chronological age | Training-subset range not verified | Targeted blood pyrosequencing cohort | 82 training; 69 independent validation | Multiple linear regression | 27K discovery; targeted bisulfite pyrosequencing |
| Lin (99 CpG) | Lin | 2016 | Blood | 99 | Chronological age | 19-101 | GSE40279 | 656 | Linear regression | 450k |
| Lin (3 CpG) | Lin | 2016 | Blood | 3 | Chronological age | 19-101 | GSE40279 | 656 | Linear regression | 450k |
| Vidal-Bralo (8 CpG) | Vidal-Bralo | 2016 | Blood | 8 | Chronological age | 20-78 | Multi-dataset | 390.0 | Linear regression | 27k |
| Horvath2 clock | Horvath | 2018 | Multiple tissues | 391 | Chronological age | −0.28–94 (training; cord blood age offset included) | Multi-dataset | 896 | Elastic net regression | 450k, EPIC |
| DNAm PhenoAge | Levine | 2018 | Blood | 513 | Mortality-related age | 21-100 | InCHIANTI | 456.0 | Elastic net regression | 27k, 450k, EPIC |
| DNAm GrimAge | Lu | 2019 | Blood | 1030 | Mortality-related age | Not reported (DNAm visit IQR 59–73; mean 66.2) | Framingham Heart Study Offspring Cohort | 1731.0 | Elastic net Cox regression | 450k, EPIC |
| Zhang (514 CpG) | Zhang | 2019 | Multiple tissues | 514 | Chronological age | 2-104 | Multi-dataset | 13566 (released predictors); 13661 (study total) | Elastic net regression | 450k, EPIC |
| Wu (111 CpG) | Wu | 2019 | Blood | 111 | Chronological age | 9-212 months | Multi-dataset | 716.0 | SIS + Elastic net regression | 27k, 450k |
| BLUP | Zhang | 2019 | Multiple tissues | 318395 | Chronological age | 2-104 | Multi-dataset | 13566 (released predictors); 13661 (study total) | Best linear unbiased prediction | 450k, EPIC |
| ABEC | Lee | 2020 | Blood | 1694 | Chronological age | 19-59 | MoBa-START | 1592.0 | Elastic net regression | EPIC |
| cABEC | Lee | 2020 | Blood | 1890 | Chronological age | 19-88 | MoBa-START + GEO | 2227.0 | Elastic net regression | 450K/EPIC shared CpGs |
| eABEC | Lee | 2020 | Blood | 1791 | Chronological age | 19-88 | MoBa-START + GEO | 2227.0 | Elastic net regression | EPIC |
| Han (65 CpG) | Han | 2020 | Blood | 65 | Chronological age | Exact training range not verified | Public blood methylation datasets | Training N not independently verified in this addition | Multiple linear regression on transformed age | 450K; EPIC-compatible sites |
| Schmidt (7 CpG) | Schmidt | 2021 | Multiple tissues | 7 | Chronological age | 18-91 (age-annotated training subset) | Public EPIC profiles from 13 studies / nine tissues (Table S4) | 104 age-annotated training profiles; 128 training / 301 total reference profiles | Multivariable linear model | EPIC |
| AltumAge | de Lima Camillo | 2022 | Multiple tissues | 20318 | Chronological age | Fetal to centenarian samples (pooled datasets) | 142 publicly available human methylation datasets | Approximately 13000 total; training-only N not verified | Deep neural network (multilayer perceptron) | 27k, 450k, EPIC common probes |
| GrimAge2 | Lu | 2022 | Blood | 1030 | Mortality-related age | 40-92 | Framingham Heart Study (FHS) Offspring Cohort | 1833.0 | Elastic net Cox regression | 450k |
| HRSInCHPhenoAge | Higgins-Chen | 2022 | Blood | 959 | Chronological age | 21-100 (package metadata) | HRS and InCHIANTI | Exact fitted-model N not verified | CpG-weighted linear model | 450K, EPIC |
| MonoDNAmAge | Liang | 2022 | Blood | 186 | Chronological age | 44-83 | MESA | 721.0 | Elastic Net Regularization (ENR) | 450k |
| PCGrimAge | Higgins-Chen | 2022 | Blood | 78464 | Mortality-related age | 24-92 | FHS Offspring and Gen3; Supplementary Table 6 | 2754 clock training; 3935 PCA reference samples | PCA and elastic-net biomarker models; calibrated GrimAge combination | 450K and EPIC |
| PCHannum | Higgins-Chen | 2022 | Blood | 78464 | Chronological age | 19-101 | GSE40279 | 656.0 | PCA and elastic net | 450K and EPIC |
| PCHorvath | Higgins-Chen | 2022 | Multiple tissues | 78464 | Chronological age | −0.5-105 | Multiple GEO, FHS and TCGA datasets; Supplementary Table 6 | 4297 listed in Table S6; paper PCA count implies 4281 | PCA and elastic net; developmental-age inverse transformation | 450K and EPIC |
| PCHorvath2 | Higgins-Chen | 2022 | Multiple tissues | 78464 | Chronological age | −0.3-101 | GEO and E-MTAB-4385; Supplementary Table 6 | 895.0 | PCA and elastic net; developmental-age inverse transformation | 450K and EPIC |
| PCPhenoAge | Higgins-Chen | 2022 | Blood | 78464 | Chronological age | 21-100 | InCHIANTI and HRS | 4505 samples (912 InCHIANTI + 3593 HRS; Supplementary Table 6) | PCA and Elastic net regression | 450k and EPIC |
| Pipek ElasticNet | Pipek | 2022 | Multiple tissues | 239 | Chronological age | Exact training range not verified | Combined GEO and GDC datasets | 5964.0 | Elastic net regression | 27K, 450K, EPIC shared sites |
| Pipek Filtered H | Pipek | 2022 | Multiple tissues | 272 | Chronological age | Exact training range not verified | Combined GEO and GDC datasets | 5964.0 | Elastic net regression restricted to Horvath sites | 27K, 450K, EPIC shared sites |
| Pipek Retrained H | Pipek | 2022 | Multiple tissues | 308 | Chronological age | Exact training range not verified | Combined GEO and GDC datasets | 5964.0 | Unpenalized linear refit of Horvath sites | 27K, 450K, EPIC shared sites |
| DNAmFitAge | McGreevy | 2023 | Blood | 1338 | Other age estimate | Not reported for combined training set | FHS + BLSA; Budapest for VO2max component | 2650 | Klemera Doubal method | 450k, EPIC |
| GP-age (30 CpG) | Varshavsky | 2023 | Blood | 30 | Chronological age | 0-103 | 19 array studies (final model) | 8337 final training; 7860 in initial 18-study analysis | Gaussian process regression (GPR) | 450k and EPIC; also evaluated on WGBS |
| GP-age (10 CpG) | Varshavsky | 2023 | Blood | 10 | Chronological age | 0-103 | 19 array studies (final model) | 8337 fitted training observations | Gaussian process regression (GPR) | 450k and EPIC; also evaluated on WGBS |
| GP-age (71 CpG) | Varshavsky | 2023 | Blood | 71 | Chronological age | 0-103 | 19 array studies (final model) | 8085 fitted training observations | Gaussian process regression (GPR) | 450k and EPIC; also evaluated on WGBS |
| GP-age (variant a) | Varshavsky | 2023 | Blood | 18 | Chronological age | 0-103 | 19 array studies (final model) | 8337 fitted training observations | Gaussian process regression (GPR) | 450k and EPIC; also evaluated on WGBS |
| GP-age (variant b) | Varshavsky | 2023 | Blood | 32 | Chronological age | 0-103 | 19 array studies (final model) | 8337 fitted training observations | Gaussian process regression (GPR) | 450k and EPIC; also evaluated on WGBS |
| GP-age (variant c) | Varshavsky | 2023 | Blood | 21 | Chronological age | 0-103 | 19 array studies (final model) | 8337 fitted training observations | Gaussian process regression (GPR) | 450k and EPIC; also evaluated on WGBS |
| Systems Age | Sehgal | 2023 | Blood | 125175 | Mortality-related age | 51-100 years (HRS biomarker-training cohort) | Health and Retirement Study and Framingham Heart Study | 3593 biomarker training; 4018 HRS PCA; 2748 FHS mortality training | PCA, elastic-net biomarker proxies and elastic-net Cox regression | Illumina EPIC and 450K common CpGs |
| SystemsAge Blood | Sehgal | 2023 | Blood | 125175 | Mortality-related age | 51-100 years (HRS biomarker-training cohort) | Health and Retirement Study and Framingham Heart Study | 3593 biomarker training; 4018 HRS PCA; 2748 FHS mortality training | PCA, elastic-net biomarker proxies and elastic-net Cox regression | Illumina EPIC and 450K common CpGs |
| SystemsAge Brain | Sehgal | 2023 | Blood | 125175 | Mortality-related age | 51-100 years (HRS biomarker-training cohort) | Health and Retirement Study and Framingham Heart Study | 3593 biomarker training; 4018 HRS PCA; 2748 FHS mortality training | PCA, elastic-net biomarker proxies and elastic-net Cox regression | Illumina EPIC and 450K common CpGs |
| SystemsAge Heart | Sehgal | 2023 | Blood | 125175 | Mortality-related age | 51-100 years (HRS biomarker-training cohort) | Health and Retirement Study and Framingham Heart Study | 3593 biomarker training; 4018 HRS PCA; 2748 FHS mortality training | PCA, elastic-net biomarker proxies and elastic-net Cox regression | Illumina EPIC and 450K common CpGs |
| SystemsAge Hormone | Sehgal | 2023 | Blood | 125175 | Mortality-related age | 51-100 years (HRS biomarker-training cohort) | Health and Retirement Study and Framingham Heart Study | 3593 biomarker training; 4018 HRS PCA; 2748 FHS mortality training | PCA, elastic-net biomarker proxies and elastic-net Cox regression | Illumina EPIC and 450K common CpGs |
| SystemsAge Immune | Sehgal | 2023 | Blood | 125175 | Mortality-related age | 51-100 years (HRS biomarker-training cohort) | Health and Retirement Study and Framingham Heart Study | 3593 biomarker training; 4018 HRS PCA; 2748 FHS mortality training | PCA, elastic-net biomarker proxies and elastic-net Cox regression | Illumina EPIC and 450K common CpGs |
| SystemsAge Inflammation | Sehgal | 2023 | Blood | 125175 | Mortality-related age | 51-100 years (HRS biomarker-training cohort) | Health and Retirement Study and Framingham Heart Study | 3593 biomarker training; 4018 HRS PCA; 2748 FHS mortality training | PCA, elastic-net biomarker proxies and elastic-net Cox regression | Illumina EPIC and 450K common CpGs |
| SystemsAge Kidney | Sehgal | 2023 | Blood | 125175 | Mortality-related age | 51-100 years (HRS biomarker-training cohort) | Health and Retirement Study and Framingham Heart Study | 3593 biomarker training; 4018 HRS PCA; 2748 FHS mortality training | PCA, elastic-net biomarker proxies and elastic-net Cox regression | Illumina EPIC and 450K common CpGs |
| SystemsAge Liver | Sehgal | 2023 | Blood | 125175 | Mortality-related age | 51-100 years (HRS biomarker-training cohort) | Health and Retirement Study and Framingham Heart Study | 3593 biomarker training; 4018 HRS PCA; 2748 FHS mortality training | PCA, elastic-net biomarker proxies and elastic-net Cox regression | Illumina EPIC and 450K common CpGs |
| SystemsAge Lung | Sehgal | 2023 | Blood | 125175 | Mortality-related age | 51-100 years (HRS biomarker-training cohort) | Health and Retirement Study and Framingham Heart Study | 3593 biomarker training; 4018 HRS PCA; 2748 FHS mortality training | PCA, elastic-net biomarker proxies and elastic-net Cox regression | Illumina EPIC and 450K common CpGs |
| SystemsAge Metabolic | Sehgal | 2023 | Blood | 125175 | Mortality-related age | 51-100 years (HRS biomarker-training cohort) | Health and Retirement Study and Framingham Heart Study | 3593 biomarker training; 4018 HRS PCA; 2748 FHS mortality training | PCA, elastic-net biomarker proxies and elastic-net Cox regression | Illumina EPIC and 450K common CpGs |
| SystemsAge MusculoSkeletal | Sehgal | 2023 | Blood | 125175 | Mortality-related age | 51-100 years (HRS biomarker-training cohort) | Health and Retirement Study and Framingham Heart Study | 3593 biomarker training; 4018 HRS PCA; 2748 FHS mortality training | PCA, elastic-net biomarker proxies and elastic-net Cox regression | Illumina EPIC and 450K common CpGs |
| ENCen100 | Dec | 2023 | Multiple tissues | 198 | Chronological age | 100-115 | Centenarian subset | 184.0 | Elastic net regression | 450K, EPIC |
| ENCen40 | Dec | 2023 | Multiple tissues | 559 | Chronological age | 40-115 | Combined public datasets | 7039.0 | Elastic net regression | 450K, EPIC |
| AdaptAge | Ying | 2024 | Blood | 998 | Chronological age | Training cohort age limits: see primary paper | LOLIPOP | 2664.0 | Elastic net | Illumina 450K |
| CausAge | Ying | 2024 | Blood | 581 | Chronological age | Not reported for exact training subset | LOLIPOP (training dataset cited to Lehne et al. 2015); Generation Scotland used for age-associated methylation effects | 2664 | Causality-enriched elastic net (tau = 0.3) | 450k |
| DamAge | Ying | 2024 | Blood | 1089 | Chronological age | Training cohort age limits: see primary paper | LOLIPOP | 2664.0 | Elastic net | Illumina 450K |
| DiseaseAge Circulatory | Fuentealba | 2024 | Blood | 654 | Mortality-related age | NA | Health and Retirement Study (HRS) | NA | Elastic net regression | Illumina MethylationEPIC |
| DiseaseAge Digestive | Fuentealba | 2024 | Blood | 322 | Mortality-related age | NA | Health and Retirement Study (HRS) | NA | Elastic net regression | Illumina MethylationEPIC |
| DiseaseAge Genitourinary | Fuentealba | 2024 | Blood | 124 | Mortality-related age | NA | Health and Retirement Study (HRS) | NA | Elastic net regression | Illumina MethylationEPIC |
| DiseaseAge Infectious | Fuentealba | 2024 | Blood | 126 | Mortality-related age | NA | Health and Retirement Study (HRS) | NA | Elastic net regression | Illumina MethylationEPIC |
| DiseaseAge Mental | Fuentealba | 2024 | Blood | 648 | Mortality-related age | NA | Health and Retirement Study (HRS) | NA | Elastic net regression | Illumina MethylationEPIC |
| DiseaseAge Metabolic | Fuentealba | 2024 | Blood | 454 | Mortality-related age | NA | Health and Retirement Study (HRS) | NA | Elastic net regression | Illumina MethylationEPIC |
| DiseaseAge Musculoskeletal | Fuentealba | 2024 | Blood | 183 | Mortality-related age | NA | Health and Retirement Study (HRS) | NA | Elastic net regression | Illumina MethylationEPIC |
| DiseaseAge Nervous | Fuentealba | 2024 | Blood | 374 | Mortality-related age | NA | Health and Retirement Study (HRS) | NA | Elastic net regression | Illumina MethylationEPIC |
| DiseaseAge Respiratory | Fuentealba | 2024 | Blood | 400 | Mortality-related age | NA | Health and Retirement Study (HRS) | NA | Elastic net regression | Illumina MethylationEPIC |
| DiseaseAge Systemic | Fuentealba | 2024 | Blood | 610 | Mortality-related age | NA | Health and Retirement Study (HRS) | NA | Elastic net regression | Illumina MethylationEPIC |
| IntrinClock | Tomusiak | 2024 | Multiple tissues | 380 | Chronological age | Full training age range not explicitly reported | Database of 71 datasets (GEO/GTEx and study-generated data) | 9104 | Elastic net (two rounds, alpha 0.5) | 450k/EPIC training; EPIC v2-compatible sites |
| Retro-age 450K | Ndhlovu | 2024 | Blood | 1282 | Chronological age | 12-100 | TruDiagnostic Biobank | 10138 training; 12670 total | Elastic net; linear weighted score | 450K and EPIC v1 common probes |
| Retro-age V2 | Ndhlovu | 2024 | Blood | 1376 | Chronological age | 12-100 | TruDiagnostic Biobank | 10138.0 | Elastic net regression | EPIC v1/v2-compatible retroelement CpGs |
| Retro-age V1 | Ndhlovu | 2024 | Blood | 1309 | Chronological age | 12-100 | TruDiagnostic Biobank | 10138 | Elastic net | EPIC |
| XAI-AGE | Prosz | 2024 | Multiple tissues | 14403 | Chronological age | 0–101 in released evaluation data; complete training range not specified | 54 pan-tissue cohorts from TCGA/GEO | 6547 total; each cohort split 75% training / 25% testing | Biologically informed explainable deep neural network | 27k, 450k |
| iCAS CompositeAge | Zheng | 2024 | Blood | 23 | Other age estimate | 20-87 | Quzhou cohort | 112 female participants; full Quzhou multi-modal training set | Elastic net regression | EPIC, EPIC v2 |
| iCAS DNAmAge | Zheng | 2024 | Blood | 65 | Chronological age | 20-87 | Quzhou cohort | 125 | Elastic net regression | EPIC, EPIC v2 |
| iCAS FacialAge | Zheng | 2024 | Blood | 70 | Other age estimate | 20-87 | Quzhou cohort | 112 female participants; full Quzhou multi-modal training set | Elastic net regression | EPIC, EPIC v2 |
| iCAS HormoneAge | Zheng | 2024 | Blood | 58 | Other age estimate | 20-87 | Quzhou cohort | 112 female participants; full Quzhou multi-modal training set | Elastic net regression | EPIC, EPIC v2 |
| iCAS ImmuneAge | Zheng | 2024 | Blood | 18 | Other age estimate | 20-87 | Quzhou cohort | 112 female participants; full Quzhou multi-modal training set | Elastic net regression | EPIC, EPIC v2 |
| iCAS LipidAge | Zheng | 2024 | Blood | 94 | Other age estimate | 20-87 | Quzhou cohort | 112 female participants; full Quzhou multi-modal training set | Elastic net regression | EPIC, EPIC v2 |
| iCAS MetabAge | Zheng | 2024 | Blood | 71 | Other age estimate | 20-87 | Quzhou cohort | 112 female participants; full Quzhou multi-modal training set | Elastic net regression | EPIC, EPIC v2 |
| iCAS PhenoAge | Zheng | 2024 | Blood | 20 | Other age estimate | 20-87 | Quzhou cohort | 112 female participants; full Quzhou multi-modal training set | Elastic net regression | EPIC, EPIC v2 |
| iCAS ProteinAge | Zheng | 2024 | Blood | 107 | Other age estimate | 20-87 | Quzhou cohort | 112 female participants; full Quzhou multi-modal training set | Elastic net regression | EPIC, EPIC v2 |
| iCAS TransAge | Zheng | 2024 | Blood | 60 | Other age estimate | 20-87 | Quzhou cohort | 112 female participants; full Quzhou multi-modal training set | Elastic net regression | EPIC, EPIC v2 |
| StocH | Tong | 2024 | Blood | 353 | Chronological age | MESA source donors: 44–83; simulated training age grid not verified | Simulated DNAm trajectories informed by MESA monocytes | Simulation training N not verified; MESA source: 1202 donors | Elastic net regression on simulated methylation | 450K |
| StocP | Tong | 2024 | Blood | 513 | Chronological age | MESA source donors: 44–83; simulated training age grid not verified | Simulated DNAm trajectories informed by MESA monocytes | Simulation training N not verified; MESA source: 1202 donors | Elastic net regression on simulated methylation | 450K |
| StocZ | Tong | 2024 | Blood | 514 | Chronological age | MESA source donors: 44–83; simulated training age grid not verified | Simulated DNAm trajectories informed by MESA monocytes | Simulation training N not verified; MESA source: 1202 donors | Elastic net regression on simulated methylation | 450K |
| Childhood CpG clock | Khodasevich | 2025 | Multiple tissues | 42 | Chronological age | Birth-21 years | 859 public samples from GEO cohorts (not 859 separate datasets) | 859 | Elastic net | Illumina 450K; sites shared with EPIC/EPICv2 |
| Childhood PCA clock | Khodasevich | 2025 | Multiple tissues | 290207 | Chronological age | Birth-21 years | Childhood methylation datasets | 859 samples | PCA and elastic net | Illumina 450K and EPIC common CpGs |
| Childhood SL PCA clock | Khodasevich | 2025 | Multiple tissues | 290207 | Chronological age | Birth-21 years | Childhood methylation datasets | 859 samples | PCA and SuperLearner ensemble | Illumina 450K and EPIC common CpGs |
| EnsembleLR | Anand | 2025 | Multiple tissues | 21639 | Chronological age | 20-80 years | GTEx; 987 original tissue samples | 80% of SMOTE-oversampled dataset; original sample count 987 | L2-regularized linear stacking; alpha 0.1 | Illumina EPIC |
| EnsembleNaive | Anand | 2025 | Multiple tissues | 21623 | Chronological age | Base-model age ranges; evaluated GTEx 20–80 | Six published clocks; evaluated in GTEx | No new fitted weights: arithmetic mean | Mean of AltumAge, Han2020, Horvath, SkinBlood, CausAge and Zhang2019 | Platforms supported by component clocks; evaluated EPIC |
| EpInflammAge | Kalyakulina | 2025 | Blood | 2211 | Chronological age | Few months to 101 years | 72 GEO datasets, 17195 healthy samples total; inflammatory predictors trained in 329 Russian participants | 13756 training/validation pool (80% of 17195); 3:1 training/validation folds | DANet age model plus neural inflammatory-marker predictors | 450k, EPIC |
| EpiAgePublic | Cheishvili | 2025 | Blood | 3 | Chronological age | 0-103 years | GSE55763, GSE157131, GSE40279, GSE30870 | 4625 | Linear regression | Illumina 450K; targeted NGS validation |
| GOLD-R DNAmAge | Zhang | 2025 | Multiple tissues | 34 | Other age estimate | 18-114 years (released training metadata) | EWAS public datasets | 7313 | Two-stage LASSO / Gompertz residual approach | Illumina 450K |
| IR mFSS | Skinner | 2025 | Blood | 46 | Chronological age | 19-101 years (GSE40279) | GSE40279 fit; RA/MS/Parkinson cohorts define rectification | 85% of 656 age-training samples | Modified forward selection and ordinary least squares with feature reflection | Illumina 450K |
| MskAge | Green | 2025 | Multiple tissues | 3365 | Chronological age | 2-95 years (training; Supplementary Table S5) | CIMA and public GEO datasets | 738 in Supplementary Table S5 (305 test); released fitted object records 921 observations | Genetic-algorithm feature selection, PCA and elastic net (alpha 0.5) | Illumina 450K + EPIC |
| PhysAge | Arpawong | 2025 | Blood | 1708 | Other age estimate | Training mean 68.2 years (SD 9.6); 49–93 is reported for test sample | US Health and Retirement Study (HRS) | 1589 | Elastic net biomarker predictors; standardized surrogate sum | EPIC (sites shared with EPIC v2) |
| RA mFSS | Skinner | 2025 | Blood | 124 | Chronological age | 19-101 years (GSE40279) | GSE40279 age fit; GSE42861 feature rectification | 85% of 656 age-training samples | Modified forward selection and ordinary least squares with feature reflection | Illumina 450K |
| WKDE | Perez-Correa | 2025 | Blood | 27 | Chronological age | 1-101 years (parent training set) | Seven GEO studies; six independent validation studies | 285 for kernel construction; 1029 parent training samples | Weighted 2D kernel density estimation; genetic-algorithm weight optimization | Illumina 450K |
| OLS mFSS | Skinner | 2025 | Blood | 74 | Chronological age | Not explicitly reported for training split | GSE40279 | 85% of 656 (15% validation; external test GSE55763) | Modified forward stepwise selection + OLS | Illumina 450K |
| MAPLE | Zhang | 2026 | Multiple tissues | 70110 | Chronological age | Wide age range; exact training limits not reported in Supplementary Table 1 | Public GEO datasets; Supplementary Table 1 | 11313 (sum of training samples in Supplementary Table 1) | Siamese pairwise encoder followed by age-prediction neural network | Illumina 450K and EPIC |
| SenChronoAge | Kasamoto | 2026 | Blood | 185 | Chronological age | 19-101 (package metadata) | UCSD and WCH / GSE40279 (package metadata) | Exact fitted-model N not verified | Elastic net regression | 450K, EPIC |