Tests to be done and parameters to be set

[RobinGuillard]

We are hopefully getting close to performing our first complete run on our dataset. Here are my suggestions on how we should perform the final tests: 1) Parameters of the pipeline: a) Preprocessing: i ) Impedance thresholding : Here is what was intially proposed : Amplitude thr, abs_thr : personnalized, rel THR: 0. Epochs 62 samples no overlap ii ) is_good : THR: 5e-5, epochs 62 samples iii ) OMA THR : Amplitude Thr : abs_Thr : 0, rel_thr : 3 (to be validated), Epochs 248 samples, no overlap, episode joining with delim of 3s

b) Bruxism classification : series of [abs_THR, Rel_Thr] : [0 , 2 to 10], on valid_labels data epochs 62 samples, no overlap, episode delim 3s c) MEMA : series of [abs_THR, Rel_Thr] : [0 , [3, 3.5, 4]], on valid_labels data epochs 248 samples, no overlap, delim 3s.

Tests to be made: I) Bruxism correlation with tinnitus variation: Main test : Nb of episodes per hour VS variation in percentage for tinnitus masking Secondary tests: Nb of episodes per hour VS variation in percentage for EVA I and EVA G Third category tests : Nb of episodes per hour VS absolute differences of tinnitus masking, EVA I, EVA G. Same analyses can be conducted for Pure and combined bruxism.

Other features to investigate (VS tinnitus measures) : Nb burst/h, Nb episode tonic, phasic, mixed per hour, total bruxism duration. An analysis can be made regarding the pure/combine ratio for patients.

II) MEMA correlation with tinnitus variation: Main test : Nb of episodes per hour VS variation in percentage for tinnitus masking Secondary tests: Nb of episodes per hour VS variation in percentage for EVA I and EVA G. Third category tests : Nb of episodes per hour VS absolute differences of tinnitus masking, EVA I, EVA G. Same analyses can be conducted for Pure and combined MEMA.

Other features to investigate (VS tinnitus measures) : total MEMA duration, Mean power of an episode. An analysis can be made regarding the pure/combine ratio for patients.

III) {Patient with overnight tinnitus rise} VS {Controls and patients without tinnitus rise overnight}: For all preceding features cited for bruxism and MEMA with the same order of priority.

[RobinGuillard]

This issue is not set in stone and should be discussed before the start of the operations.