docs: input/outputs

[pausz]

From Nathan.

Python Function (general form, Matlab will parallel)

Input:     EDF file with EEG of the correct dimensions (number of channels to align with ONNX models, longer than 60s)

              Age of Recording (to correct FBA)

Outputs: FBA – raw FBA

  FBA_corrected – corrected so that FBA at 50th centile equals age

  Centiles – from growth chart based on all our data

  PAD (predicted age difference) – difference between raw FBA and age

 

The last two variables will come from Kartik’s stuff

 

ONNX list: 4 models will be used – input epochs are number of channel x 1920 samples

D1_18ch_model.onnx – trained on D1, 18 channel data,

D1_2ch_model.onnx – trained on D1, 2 channel data

D2_2ch_model.onnx – trained on D2, 2 channel data

D1D2_2ch_model.onnx – trained on combined dataset of D1 and D2, 2 channel data

 

Each of these files will need their own lookup tables from Kartik’s work.