Network-level enrichment provides a framework for biological interpretation of machine learning results.

While the difference in Pearson r correlations between actual and predicted ages appeared highly significant, when we instead compared the actual r correlations within a network block to actual r correlations from an equal count of randomly selected features outside the network block, we observed that there were no network blocks that predicted age better than randomly selected features ( Figure 8A ).

also in 132,142 other papers