Mean absolute error

1. intercept-only adjustment decreases mean absolute error in silico (from 8.83 to 5.56 years, 4.28 to 2.46 years, and 7.77 to 3.92 years).

   “intercept-only adjustment reduced the eight-CpG model mean held-out MAE from 8.83 to 5.56 years, 4.28 to 2.46 years, and 7.77 to 3.92 years, respectively.”

   In silicorepeated-split calibration

   Compact DNA methylation panels for forensic age estimation: cross-cohort evaluation and calibration using public whole-blood data

   Research Square·27 Aug 2026Preprint

1. four-CpG model predicts mean absolute error in silico (MAE 8.62 years and 3.90 years).

   “the four-CpG model achieved the lowest MAE in GSE40279 (8.62 years) and GSE69270 (3.90 years)”

   In silicoraw external evaluation

   Compact DNA methylation panels for forensic age estimation: cross-cohort evaluation and calibration using public whole-blood data

   Research Square·27 Aug 2026Preprint

More on four-CpG model

1. ResNet-18 is associated with mean absolute error in silico (MAE 3.61 years).

   “ResNet-18 was the superior architecture, achieving an internal test Mean Absolute Error (MAE) of 3.61 years”

   In silicointernal test on Brazilian OPGsn = 10,036

   External Validation and Interpretability Analysis of a Deep Learning Model for Forensic Age Estimation Across Brazilian and Romanian Populations

   Research Square·4 Aug 2026Preprint