Accuracy Loss at Lower Sampling Frequencies

[StantonGodshall]

Hello,

I apologize if this question should be asked elsewhere. I have found that if I downsample my 100Hz data to 15Hz, there is a substantial reduction in the total number of steps detected (36% reduction). I know that the original paper validated the algorithm on both 100Hz and 15Hz datasets, so I just wanted to confirm that this loss is expected and not attributable to something else. I understand there are inherent challenges to sampling at lower rates. We have a cohort of patients with low sampling frequencies (12.5Hz) and with high frequency (100Hz), so I was hoping to compare them. I'm happy to share more details/data files if necessary.

Thank you, Stanton