Machine vision-based frailty assessment for genetically diverse mice

Abstract

Frailty indexes (FIs) capture health status in humans and model organisms. To accelerate our understanding of biological aging and carry out scalable interventional studies, high-throughput approaches are necessary. We previously introduced a machine vision-based visual frailty index (vFI) that uses mouse behavior in the open field to assess frailty using C57BL/6J (B6J) data. Aging trajectories are highly genetic and are frequently modeled in genetically diverse animals. In order to extend the vFI to genetically diverse mouse populations, we collect frailty and behavior data on a large cohort of aged Diversity Outbred (DO) mice. Combined with previous data, this represents one of the largest video-based aging behavior datasets to date. Using these data, we build accurate predictive models of frailty, chronological age, and even the proportion of life lived. The extension of automated and objective frailty assessment tools to genetically diverse mice will enable better modeling of aging mechanisms and enable high-throughput interventional aging studies.