Overview
- Stanford researchers continuously filmed 81 African turquoise killifish across their 4–8 month lifespans, generating billions of video frames.
- Machine-learning identified about 100 short behavioural syllables and built a behavioural clock that forecast individual lifespan from a few days of midlife data.
- By roughly 70–100 days, longer-lived fish were more active and slept mainly at night, whereas shorter-lived fish moved less and napped during the day.
- Analysis indicated ageing occurs in roughly two to six discrete stages rather than as a smooth, continuous decline.
- The team notes potential to adapt similar passive tracking with consumer wearables, while stressing that validation beyond this single lab species is needed.