Robust Functional Statistics applied to Probability Density Function Shape screening of sEMG data

Archive ouverte

Boudaoud, Sofiane | Rix, Hervé | Al Harrach, Mariam | Marin, Frédéric

Edité par CCSD -

International audience. Recent studies pointed out possible shapemodifications of the Probability Density Function (PDF) ofsurface electromyographical (sEMG) data according to severalcontexts like fatigue and muscle force increase. Following thisidea, criteria have been proposed to monitor these shapemodifications mainly using High Order Statistics (HOS)parameters like skewness and kurtosis. In experimentalconditions, these parameters are confronted with small samplesize in the estimation process. This small sample size induceserrors in the estimated HOS parameters restraining real-timeand precise sEMG PDF shape monitoring. Recently, afunctional formalism, the Core Shape Model (CSM), has beenused to analyse shape modifications of PDF curves. In thiswork, taking inspiration from CSM method, robust functionalstatistics are proposed to emulate both skewness and kurtosisbehaviors. These functional statistics combine both kerneldensity estimation and PDF shape distances to evaluate shapemodifications even in presence of small sample size. Then, theproposed statistics are tested, using Monte Carlo simulations,on both normal and Log-normal PDFs that mimic observedsEMG PDF shape behavior during muscle contraction.According to the obtained results, the functional statistics seemto be more robust than HOS parameters to small sample sizeeffect and more accurate in sEMG PDF shape screeningapplications.

• Description
• Sujet/Public
• Outil
• Outil d'anticipation
• Mémoire et thèse
• Gestion