Photographic age gap is associated with frailty and predicts survival in laboratory mice

A research study has found that age gaps in photographs of laboratory mice are associated with frailty and can predict survival. The study, which involved 1,292 mice from four different strains aged between one and 34 months, utilized photographs taken at seven facilities. By employing a frozen self-supervised vision model called DINOv2 and a ridge regression, researchers were able to predict the age of the mice. However, the absolute age transferred poorly between facilities, with a median absolute error of 138 days.

Despite this, relative age was found to transfer well among mice photographed in the same session. The older mouse was predicted to be older in 65% of pairs, and this accuracy increased to above 95% for age differences of over 500 days. Additionally, the later photograph of the same mouse was predicted to be older in 63% of repeat pairs. In comparison to sex, which was inferred more reliably at 87%, the age information conveyed in photographs proved valuable.

Interestingly, mice predicted to be older than their actual age were found to be frailer, exhibiting a 0.12 standard deviation increase per 100 days. Furthermore, these predicted older mice also had shorter survival rates, with a hazard ratio of 1.14 per 100 days. Additionally, fourth generation (G4) telomerase-deficient mice appeared 151 days older than their wild-type counterparts in photographs.

The study concludes that routine photographs of mice carry a real biological age signal that can be read without any additional cost. This finding suggests that photographs could be best utilized to compare mice that have been photographed together.