How reliable are epigenetic clocks in measuring biological age? This is a critical question for the research community, as accurate clocks can accelerate the development of rejuvenation therapies. However, the construction of epigenetic clocks through machine learning poses challenges in quantifying the risk of inaccuracies. Uncertainty in aging clock predictions stems from model choice, out-of-distribution data, and data variations. Existing clocks lack the ability to estimate uncertainty effectively, raising concerns about their reliability in predicting biological age. A new analytical framework highlights the high uncertainty in age estimations, casting doubt on the validity of rejuvenation effects observed in current models. Including uncertainty estimation in future aging clock models is crucial to avoid misinterpretation of biological age predictions.
Link: https://doi.org/10.1111/acel.14283
