Fig. 5: The performance of omics-based scores for disease trajectories. | Nature Communications

Fig. 5: The performance of omics-based scores for disease trajectories.

From: Multiomics insight into disease trajectories of cardiometabolic diseases and cancer

Fig. 5: The performance of omics-based scores for disease trajectories.The alt text for this image may have been generated using AI.

a Overview of the model development procedures. Created in BioRender. Jiang, X. (2025) https://BioRender.com/u3bef2j. b Frequent genomics, metabolomics, and proteomics predictors identified through LASSO regression across 9 outcomes (n = 375,239 for genomics dataset, n = 234,452 for metabolomics dataset, and n = 44,816 for proteomics dataset). Predictors with non-zero coefficients were considered as selected features across the 9 outcomes. The Δ C-statistics between the base model and omics-based scores for c single morbidity outcomes, d multimorbidity outcomes, and e mortality outcomes (n = 375,239 for genomics dataset, n = 234,452 for metabolomics dataset, and n = 44,816 for proteomics dataset). ΔC-statistics and 95% confidence intervals are shown. The ΔC-statistics were estimated using time-dependent Cox models with 100 bootstrap resamples in the replication dataset. CA cancer, CMD cardiometabolic disease.

Back to article page