lkorczowski / Tinnitus-n-Sleep

Detecting events in sleeping tinnitus patients
MIT License
1 stars 0 forks source link

Sprint 2 #16

Closed RobinGuillard closed 4 years ago

RobinGuillard commented 4 years ago

Milestone to reach: 0.0.2

backlog for this sprint

Electrode disconnection

Bruxism event classification

Build automatic reporting for classification algorithm comparison

Build automatic reporting for clinical bruxism assessment

NOTES:

lkorczowski commented 4 years ago

Note that if use in multidimensional context, the very simple rule of for AmplitudeThresholding allows to discard independent event happening on a single electrode: https://github.com/lkorczowski/Tinnitus-n-Sleep/blob/6f71ce59a3a91986ee881b5726d536b491735241/tinnsleep/classification.py#L69-L78

Therefore, I believe the most urgent element of this sprint should be proper validation of the methods we develop. To do so, I propose:

Build automatic reporting for classification algorithm comparison

see tasks in #17

After that, any additional method (e.g. electrode disconnection) should be benchmarked using this report. It will save us time and sanity.

lkorczowski commented 4 years ago

not sure it is completely done

lkorczowski commented 4 years ago

Electrode disconnection added to Sprint 3 #28