Preprocessing before gaitpy

[IRuotsa]

Dear @matt002 and other GaitPy developers, thanks for making GaitPy!

I have been trying out GaitPy in our accelerometer data concerning older adults and have few questions.

1. In your article concerning GaitPy , it is mentioned that "Prior to feature extraction, raw vertical acceleration data is down-sampled to 50Hz and band-pass filtered using a 0.5-3Hz 1st order Butterworth filter." But do you recommend any other preprocessing. Despite this, I still found values that are impossible (> 1000 m/s**2). Do you suggest some sort of a outlier removal before GaitPy analysis?

2. Second question is related to the first one. I did not found reference why you chose to use 0.5-3Hz 1st order Butterworth filter instead of e.g. 4th order Butterworth filter. Would be interesting to hear the reason for this as this appears to differ from what is typically used in gait analyses. (Sorry if this is silly question, but my knowledge on butterworth filters is very very limited)

3. Do you plan to publish validation results in future? This would help us and maybe also some other researchers to use this tool, as published validation results are required by some ethical committees.

Thanks again!
Ilona