From ageing clocks to human digital twins in personalising healthcare through biological age analysis

Abstract

Age is the most important risk factor for the majority human diseases, leading to the exploration of innovative approaches, including the development of predictors to estimate biological age (BA). These predictors offer promising insights into the ageing process and age-related diseases. With real-time, multi-modal data streams and continuous patient monitoring, these BA can also inform the construction of 'human digital twins', quantifying how age-related changes impact health trajectories. This study highlights the significance of BA within a deeply phenotyped longitudinal cohort, using omics-based approaches alongside gold-standard clinical risk predictors. BA and health traits predictions were computed from 29 epigenetics, 4 clinical-biochemistry, 2 proteomics, and 3 metabolomics clocks. The study reveals that ageing is different between individuals but relatively stable within individuals. We suggest that BA should be considered crucial biomarkers complementing routine clinical tests. Regular updates of BA predictions within digital twin frameworks can also help guiding individualised treatment plans.

Fig. 1. Study design of the IAM Frontier.

Longitudinal comprehensive data were collected over the course of the study including monthly physiological and clinical biochemistry data (13 time points), bimonthly proteomics and metabolomics data (7 time points), and 6-monthly DNA methylation and microbiome data (3 time points). The study participants consist of 15 males and 15 females, with a (chronological) age range of 45–59 years old.

Fig. 2. The overview of published clinical and omics ageing clocks and predictors discussed in this article.

The predictors are grouped based on their purpose on predicting age, health traits, and telomere length. Each node represents one predictor connected to another. In the left-hand panel, the edges represent the ratio of shared parameters, e.g. there are 26 shared parameters between the MetaboAge-vdA clock and the MetaboAge-MD clock, hence the thickest edge. The size of each node represents the total number of parameters used in each clock. The right-hand panel shows all DNA methylation predictors discussed in this study. Edges indicate the parameters used for each predictor. Due to the high number of DNA methylation sites used by the predictors, nodes and edges are shown in standard sizes, with no relation to the number of features used or shared.

Fig. 3. Density plots and correlations of BA predictions.

MethylDetectRAge and Skin & blood clock have the highest correlations with the chronological age (CA) (left panel); the Skin & blood clock predicted age gives the most similar kernel density as CA, both in position and shape (right panel).

Fig. 4. Longitudinal exploration of BA predictions in three different omics modalities.

a Unsupervised analysis of BA predictions (PCA and cluster analysis) for all subjects in three shared time points, b ICC and Var_ratio of each BA clocks; large ICC and small Var_ratio indicate high within-person similarity, c Age acceleration of all IAM Frontier participants in different clocks; epigenetics clocks with ${\stackrel{\circledR }{S}}_E=0.164$: multi-tissue clock (pink), skin & blood clock (yellow), henoAge (green), GrimAge2 (orange), MethylDetectR (steel blue, top), CausAge(light blue, bottom) and Hannum clock (black). Clinical clocks with ${\stackrel{\circledR }{S}}_C=0.118$: MLR-Levine clock (brown), PCA-Levine clock (light pink), and KDM clock (blue). Proteomics clocks with ${\stackrel{\circledR }{S}}_P=0.116$: ProtAge-Tanaka (grey) and ProtAge-MD (tosca).

Fig. 5. Individual analysis of BA predictions in all available time points.

a Reduced ordered heatmap based on MAD thresholds. b Age acceleration of ID06, ID08, and ID27. Prediction values that are significantly different from the rest of the cohort are marked with full circles. c Clinical and blood cell profiles of ID06, ID08, and ID27. Measurements that are outside the cohort reference intervals are marked with full circles. The blood cell counts are shown in percentages.