Abstract
Little is known about tissue-specific changes that occur with aging in humans. Using the description of 33 million histological samples we extract thousands of age- and mortality-associated features from text narratives that we call The Human Pathome (pathoage.com). Notably, we can broadly determine when post-development aging starts at the organism and tissue level, indicating a sexual dimorphism with females aging earlier but slower and males aging later but faster. We employ unsupervised topic-modeling to identify terms and themes that predict age and mortality. As a proof of principle, we cross-reference these terms in PubMed to identify nintedanib as a potential aging intervention and show that nintedanib reduces markers of cellular senescence, reduces pro-fibrotic gene pathways in senescent cells and extends the lifespan of fruit flies. Our findings pave the way for expanded exploitation of population text datasets towards discovery of novel aging interventions.