Epigenetic age estimation in saliva and in buccal cells

The study developed and validated age prediction models using epigenetic markers from saliva and buccal cells, which are useful for criminal investigations. Using 184 samples (91 saliva, 93 buccal cells) for training and another 184 for testing, the tissue prediction model achieved an 88.59% correct classification rate for the training set and 83.69% for the testing set. The combined age prediction model, using seven CpG sites, had a median absolute error of ±3.54 years for the training set and ±3.66 years for the testing set, with correct classification rates of 76.08% and 71.19%, respectively. Adding tissue origin as a co-variate did not improve age predictions. The model's robustness was confirmed with a minimum of 10 ng of DNA, and it is available on the Snipper forensic classification website.