search

lkorczowski / Tinnitus-n-Sleep

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

Add ratio of power in the reporting for middle ear events #51

Closed RobinGuillard closed 4 years ago

RobinGuillard commented 4 years ago

It could be cool to report the mean ratio of power between events and (adaptive) baseline values: in fact, one hypothesis to test is "are MEMA activity for tinnitus patients more powerful than for controls". When I inspect data of "pure" middle ear events, I think it's possible this hypothesis is true.

lkorczowski commented 4 years ago

something like :

amplitude = rms(epoch)
power_num = np.mean(amplitude[labels]**2)
power_denum = np.mean(amplitude[np.invert(labels)]**2)
power_ratio = power_num / power_denum

?

lkorczowski commented 4 years ago

option 2

amplitude = rms(epoch)
power_num = np.mean(amplitude[labels]**2)
power_denum = np.mean(amplitude**2)
power_ratio = power_num / power_denum
RobinGuillard commented 4 years ago

Better option 1