Machine learning predicts lifespan and underlying causes of death in aging C. elegans

Abstract

Senescence (aging) leads to senescent pathology that causes death, and genes control aging by determining such pathology. Here we investigate how senescent pathology mediates the effect of genotype on lifespan in C. elegans by means of a data-driven approach, using machine learning (ML). To achieve this we gathered extensive data on how diverse determinants of lifespan (sex, nutrition, genotype) affect patterns of age-related pathology. Our findings show that different life-extending treatments result in distinct patterns of suppression of senescent pathology. By analysing the differential effects on pathology and lifespan, our ML models were able to predict >70% of lifespan variation. Extent of pathology in the pharynx and intestine were the most important predictors of lifespan, arguing that elderly C. elegans die in part due to late-life disease in these organs. Notably, the mid-life pathogenetic burst characteristic of hermaphrodite senescence is absent from males.