A miRNA-based epigenetic molecular clock for biological skin-age prediction

Abstract

Skin aging is one of the visible characteristics of the aging process in humans. In recent years, different biological clocks have been generated based on protein or epigenetic markers, but few have focused on biological age in the skin. Arrest the aging process or even being able to restore an organism from an older to a younger stage is one of the main challenges in the last 20 years in biomedical research. We have implemented several machine learning models, including regression and classification algorithms, in order to create an epigenetic molecular clock based on miRNA expression profiles of healthy subjects to predict biological age-related to skin. Our best models are capable of classifying skin samples according to age groups (18-28; 29-39; 40-50; 51-60 or 61-83 years old) with an accuracy of 80% or predict age with a mean absolute error of 10.89 years using the expression levels of 1856 unique miRNAs. Our results suggest that this kind of epigenetic clocks arises as a promising tool with several applications in the pharmaco-cosmetic industry.

Keywords: Aging; ElasticNet; Machine learning; Prediction; Skin; Suppor Vector classifier.

Fig. 1

Principal Component Analysis (PCA) of miRNAs database. a) PCA depicting the distribution of samples before batch effect correction and adjusting for surrogate variables. b) PCA depicting the distribution of samples after batch effect correction. Study_ID 1: GSE31037 (Joyce et al. 2011).Study_ID 2: GSE72193 (Gulati et al. 2016). Study_ID 3: GSE84193 (Chitsazzadeh et al. 2016). Study_ID 4: GSE142582 (Yu et al. 2020b)

Fig. 2

Biological age prediction from miRNA expression by regression analysis. Results of the biological age prediction, using regression models, computed by 10 cross-validation. The x-axis shows the chronological age in years. The y-axis show the biological age in years (predicted age) computed for each regression model a) ElasticNet, b) AdaBoost regressor, c) Lasso, d) Support Vector Regressor with radial basis kernel function (SVR-RBF), (e) SVR with lineal kernel function, (f) SVR with polynomial kernel function. Every blue dot depicts one prediction for one subject of the testing set during the cross-validation. The dotted line shows the perfect linear correlation

Fig. 3

Confusion Matrix of classification models for biological age group. a, c, e, g. Confusion matrix for multiclass classification models (Random Forest, Gradient Boosting, kNN, and Support Vector Classifier) after age group prediction of a common training set (13 samples). The x-axis shows the true label (real age group) for a given sample. The y-axis shows the predicted label (predicted age group) for a given sample according to age groups. Group 1: 18–28 years old. Group 2: 29–39 years old. Group 3: 40–50 years old. Group 4: 51–60 years old. Group 5: 61–83 years old. b, d, f, h. Confusion matrix for multiclass classification models (Random Forest, Gradient Boosting, kNN, and Support Vector Classifier) with SMOTE application after age group prediction for a common training set (21 samples). The x-axis shows the true label (real age group) for a given sample. The y-axis shows the predicted label (predicted age group) for a given sample. The optimum confusion matrix should show a perfect diagonal line