Machine vision-based frailty assessment for genetically diverse mice.

Area of Science:

• Gerontology
• Animal Behavior
• Biomedical Engineering

Background:

• Frailty indexes (FIs) assess health and aging in humans and model organisms.
• High-throughput methods are crucial for aging research and interventional studies.
• Previous work developed a visual frailty index (vFI) using C57BL/6J mouse behavior.

Purpose of the Study:

• To extend the visual frailty index (vFI) to genetically diverse mouse populations.
• To enable scalable, high-throughput aging studies in diverse genetic backgrounds.
• To develop accurate predictive models for frailty and aging in mice.

Main Methods:

• Collected frailty and open-field behavior data from a large cohort of aged Diversity Outbred (DO) mice.
• Combined new data with previous datasets to create a large video-based aging behavior dataset.
• Developed predictive models using machine vision and behavioral data.

Main Results:

• Established accurate predictive models for frailty, chronological age, and proportion of life lived in DO mice.
• Demonstrated the applicability of the vFI to genetically diverse mouse populations.
• Created one of the largest video-based aging behavior datasets to date.

Conclusions:

• Automated frailty assessment tools can be extended to genetically diverse mice.
• This advancement will facilitate better modeling of aging mechanisms.
• Enables high-throughput interventional aging studies in varied genetic contexts.