Powered by state-of-the-art deep learning and computer vision technology, the proposed platform will leverage previous work on automated markerless motion capture (i.e., whole body movement tracking from video) and behavioral phenotyping (i.e., detection of behavior events such as walking or eating). This technology has been used for video-based behavioral analysis in insects, fish, plants, rodents, and humans, including for applications such as health monitoring in animal studies of cancer and neurodegeneration.

The proposed work will demonstrate the feasibility of using this technology to automate spaceflight behavioral health monitoring by applying it to previously collected videos from the NASA Rodent Research-1 mission. These have been painstakingly manually annotated by human expert observers with frame-by-frame labels of behaviors (e.g., feeding, grooming) to enable quantitative analysis of rodent behavior during spaceflight. This laborious effort was necessary due to the challenging imaging conditions inherent in spaceflight videography – but which are overcome through the use of AI in the proposed work. By developing a platform capable of automating this process, we will establish the foundation for future systems that will be able to monitor behavioral health in research missions – automatically and in real-time, opening the door to interventional studies aimed at maintaining positive behavioral health conditions. Future work may adapt this technology for behavioral monitoring in humans to detect and mitigate the risks of crewed deep space missions.

Furthermore, this research lays the groundwork for adapting the technology to enable human health and behavioral performance monitoring in upcoming crewed deep space missions. The proposed work is also valuable to the Institutional Animal Care and Use Committee and the Rodent Research Project as it fills a critical technology gap.

By leveraging existing flight data from the RR-1 mission, this project will validate the applicability of AI-based behavioral analysis tools to other existing and upcoming RR missions. This approach not only benefits future missions but also increases the utility of data collected in past missions, supporting NASA's goals of maximizing the scientific return from spaceflight experiments and advancing our understanding of biological responses to the space environment.

The accomplishments we have made in these areas demonstrate the feasibility of automating motion capture-based analysis of animal behavior in microgravity under challenging imaging conditions. Our work has set a foundation for future improvements in behavioral monitoring systems for spaceflight research.

Future plans: This work will be fully described in a manuscript set to be submitted this year, and is touched on in a perspective that is under revision at npj Microgravity. Future plans also include completed estimation of 3D Kinematics.

Powered by state-of-the-art deep learning and computer vision technology, the proposed platform will leverage previous work on automated markerless motion capture (i.e., whole body movement tracking from video) and behavioral phenotyping (i.e., detection of behavior events such as walking or eating). This technology has been used for video-based behavioral analysis in insects, fish, plants, rodents, and humans, including for applications such as health monitoring in animal studies of cancer and neurodegeneration.

The proposed work will demonstrate the feasibility of using this technology to automate spaceflight behavioral health monitoring by applying it to previously collected videos from the NASA Rodent Research-1 mission. These have been painstakingly manually annotated by human expert observers with frame-by-frame labels of behaviors (e.g., feeding, grooming) to enable quantitative analysis of rodent behavior during spaceflight. This laborious effort was necessary due to the challenging imaging conditions inherent in spaceflight videography – but which are overcome through the use of AI in the proposed work. By developing a platform capable of automating this process, we will establish the foundation for future systems that will be able to monitor behavioral health in research missions – automatically and in real-time, opening the door to interventional studies aimed at maintaining positive behavioral health conditions. Future work may adapt this technology for behavioral monitoring in humans to detect and mitigate the risks of crewed deep space missions.