A new cognitive clock matching phenotypic and epigenetic ages

Abstract

Cognitive abilities decline with age, constituting a major manifestation of aging. The quantitative biomarkers of this process, as well as the correspondence to different biological clocks, remain largely an open problem. In this paper we employ the following cognitive tests: 1. differentiation of shades (campimetry); 2. evaluation of the arithmetic correctness and 3. detection of reversed letters and identify the most significant age-related cognitive indices. Based on their subsets we construct a machine learning-based Cognitive Clock that predicts chronological age with a mean absolute error of 8.62 years. Remarkably, epigenetic and phenotypic ages are predicted by Cognitive Clock with an even better accuracy. We also demonstrate the presence of correlations between cognitive, phenotypic and epigenetic age accelerations that suggests a deep connection between cognitive performance and aging status of an individual.

A cost‐effective blood DNA methylation‐based age estimation method in domestic cats, Tsushima leopard cats (Prionailurus bengalensis euptilurus) and Panthera species, using targeted bisulphite sequencing and machine learning models

“…As shown in Figure 1, after feature selection, we created regression models with elastic net regression and support vector machine radial (SVMr) in the training data. SVMr, similar to elastic net regression, has been found to produce high estimation accuracy (Krivonosov et al, 2022;Nakamura et al, 2023;Qi et al, 2021;Xu et al, 2015). We performed elastic net regression and its hyperparameter tuning with cv.glmnet in the R package glmnet 4.1-8 (Friedman et al, 2010).…”

Blood DNA methylation-based age estimation in domestic cats, Tsushima leopard cats (Prionailurus bengalensis euptilurus), and Panthera species using machine learning models

“…After the feature selection (i.e., the selection of CpG sites), to build age estimation regression models, we applied two methods-elastic net regression and Support Vector Machine radial (SVMr) on new, featureextracted datasets. Both regression models are widely applied in many DNA methylation-based age estimation studies with high estimation accuracy (Bors et al, 2021;Krivonosov et al, 2022;Qi et al, 2021;Raj et al, 2021;Thompson et al, 2017;Vidaki et al, 2021;Xu et al, 2015). The setting of elastic net regression is the same as what was described in the above feature selection.…”