UCLA

Biological age (BA) as opposed to chronological age (CA) is meant to measure the true aging process of an individual. The expectation is that biological age is better than chronological age when it comes to predicting mortality or age related functional decline. We evaluate and adapt techniques from Klemera and Doubal (KD) (2006), an adapted reverse regression approach, and propose a novel ensemble based approach in estimating biological age. However, several simulations has shown that the KD approach may produce unreliable estimates due to the non-robust reverse regression. Though the ensemble based approach mitigates the issue, it is not without limitations. Here, we propose a method to produce more reliable estimates for the KD approach and, consequently, improve the accuracy of our ensemble based approach as well. Lastly, we evaluate and compare the performance of each method including our improved ensemble based model in simulated scenarios and in a Down syndrome application. We found that the ensemble KD approach outperforms the KD approach and standard methods in estimating biological age. Thus, the development of the ensemble KD approach may be useful in examining phenotypic traits that affects aging patterns.