In mice/rats, a non invasive way, is to monitored food intake (hr or every 2 hour) in a manual or automatic manner. If you have the material, spontaneous locomotor activity is also a parameter. Slightly more invasive, blood collection can be made at different time point and for example cortico, glutamine or albumin measured.
I second the locomotor suggestion, and would expand that to behavioral observation in general. Fully automated systems are expensive, and doing hours of behavioral observation scoring is cumbersome and time consuming. One noninvasive option would to be focus on and video record key transition periods such as the hour after lights off and the hour after lights on, to check for behavioral changes. While you could do scan sampling of behaviors every minute, continuous scoring for frequency and duration of behavior is ideal. You could start with scan sampling, but go back and analyze the videos if needed.