Architecture of aging seen in lifelong behavior of short-lived fish.

Researchers continuously tracked African turquoise killifish through adulthood and found that midlife behaviors—especially sleep patterns and daytime activity—predicted individual lifespan and that aging unfolded in distinct, rapid transitions separated by stable stages.

Researchers at Stanford continuously monitored individual African turquoise killifish throughout their adult lives to study how daily behavior and aging are linked. The team used automated, camera‑monitored tanks and analyzed posture, speed, rest and movement across billions of video frames. From those data they identified about 100 recurring behavioral syllables and followed 81 fish from early adulthood to death. The study aimed to see when individual aging paths diverge and whether behavior alone can predict lifespan. Key findings: - The researchers tracked 81 killifish continuously from adulthood and generated billions of video frames for analysis. - They identified about 100 recurring behavioral syllables capturing posture, speed, rest and movement. - By early midlife (around 70–100 days), behavioral differences correlated with eventual lifespan. - Fish that later had shorter lives tended to sleep more during the day; longer‑lived fish were more active in daylight and showed higher darting speeds. - Machine‑learning models could use a few days of midlife behavior to forecast lifespan for individual fish. - Aging proceeded in two to six rapid behavioral transitions separated by longer stable stages, and liver gene activity differed between aging paths, with higher activity of protein‑production and maintenance genes in shorter‑lived fish. Summary: The researchers report that continuous behavioral monitoring offers a sensitive, whole‑organism readout of aging and that individual trajectories can diverge by midlife. The team plans further experiments to test targeted interventions (including diet and genetic changes), to study brain activity alongside behavior, and to explore whether similar principles appear in humans via long‑term tracking. The broader implications and the effectiveness of interventions are undetermined at this time.