The Canon M50 has 1.5 times the sensor area and 1.5 times the pixel number (24MP instead of 16MP) so the pixel size is the same. That gives the cameras basic equal photon noise level per pixel. The comparison was done with video which often uses a crop (don't know whether this was the case here) where the pixel noise would be relevant also for the noise per framed area. So that makes the decision depend on the brightness of the lenses used. Those were probably kit lenses and looked rather compact. With similarly compact kit lenses for APS-C and 4/3, chances are that the maximum aperture for APS-C ends up smaller (larger aperture number) in order to fit the size.

Framing was similar in both cases; with larger sensor areas there is a tendency to support wider angles in a default lens. Which would mean that for equal framing the larger sensor lens might be zoomed in more, which might reduce the aperture size available compared to the maximum aperture size.

So basically you need to look at a number of details more before being able to make a guess regarding better low-light behavior, and then the possibly different sensor quality has not even yet factored in.