Many molecular aging biomarkers have been developed to capture heterogeneity in individual aging rates. Yet, systematic comparison of the modeling choices underlying these biomarkers has been limited. In this study, we trained aging biomarkers on the Rockwood frailty index (FI) and all-cause mortality using UK Biobank Olink proteomics and metabolomics (¹H-NMR) data (n = 40,696). We systematically established the impact of model choice, target outcome, and molecular data source on several age-related outcomes. From this, we developed two aging biomarkers, ProteinFrailty (ProtFI) and ProteinMortality (ProtMort), which are both ElasticNet models that use a minimal set of proteins to predict FI and mortality, respectively. In particular, ProtFI outperformed established aging biomarkers in relation to diverse outcomes, including incident cardiovascular disease, handgrip strength, and self-rated health, both in internal validation and two Dutch external cohorts (n = 995, n = 500). Our findings show that an efficient frailty-trained proteomic biomarker robustly predicts age-related decline.

Integrating multi-omics data into predictive models has the potential to enhance accuracy, which is essential for precision medicine. In this study, we developed interpretable predictive models for multi-omics data by employing neural networks informed by prior biological knowledge, referred to as visible networks. These neural networks offer insights into the decision-making process and can unveil novel perspectives on the underlying biological mechanisms associated with traits and complex diseases. We tested the performance, interpretability and generalizability for inferring smoking status, subject age and LDL levels using genome-wide RNA expression and CpG methylation data from the blood of the BIOS consortium (four population cohorts, Ntotal = 2940). In a cohort-wise cross-validation setting, the consistency of the diagnostic performance and interpretation was assessed. Performance was consistently high for predicting smoking status with an overall mean AUC of 0.95 (95% CI: 0.90-1.00) and interpretation revealed the involvement of well-replicated genes such as AHRR, GPR15 and LRRN3. LDL-level predictions were only generalized in a single cohort with an R2 of 0.07 (95% CI: 0.05-0.08). Age was inferred with a mean error of 5.16 (95% CI: 3.97-6.35) years with the genes COL11A2, AFAP1, OTUD7A, PTPRN2, ADARB2 and CD34 consistently predictive. For both regression tasks, we found that using multi-omics networks improved performance, stability and generalizability compared to interpretable single omic networks. We believe that visible neural networks have great potential for multi-omics analysis; they combine multi-omic data elegantly, are interpretable, and generalize well to data from different cohorts.