Artificial Intelligence-Based Machine and Deep Learning Techniques That Use Brain Waves to Detect Depression

Abstract

Electroencephalogram (EEG) lsignal-based lemotion lrecognition lhas lattracted lwide linterests in lrecent lyears land lhas lbeen lbroadly ladopted in lmedical, laffective lcomputing, land lother lrelevant lfields. Depression has lbecome la lleading lmental ldisorder lworldwide. Evidence lhas lshown lthat lsubjects lwith ldepression lexhibit ldifferent lspatial lresponses in lneurophysiologic lsignals lfrom lthe lhealthy lcontrols lwhen lthey lare lexposed lto lpositive land lnegative. Depression isla common lreason lfor an increase in lsuicide lcases lworldwide. EEG lplays an important lrole in lE-healthcare lsystems, lespecially in lthe lmental lhealthcare larea, lwhere lconstant land lunobtrusive lmonitoring lis ldesirable. EEG lsignals lcan lreflect lactivities lof lthe lhuman lbrain land lrepresent different lemotional lstates. Mental lstress lhas lbecome la lsocial lissue land lcould lbecome la lcause lof lfunctional ldisability lduring lroutine lwork. This lResearch presents ldeep llearning ltechnique lfor ldetecting ldepression lusing lEEG. The lalgorithm lfirst lextracts lfeatures lfrom lEEG lsignals land lclassifies lemotions lusing lmachine land ldeep llearning ltechniques, in lwhich ldifferent lparts lof la ltrial lare lused lto ltrain lthe lproposed lmodel land lassess lits limpact lon lemotion lrecognition lresults. The simulation is performed lusing lthe lPython lspyder lsoftware. The lprecision lof lthe lproposed lwork lis l99% lwhile in lthe lprevious lwork lit lis l91.00%. lSimilarly lthe lother lparameters llike lRecall land lF_Measure lis l94% land l97% lby lthe lproposed lwork land l88.00% land l89.00% lby lthe lprevious lwork. The loverall laccuracy lachieved lby lthe lproposed lwork lis l96.48% lwhile lprevious lit lis lachieved l91.00%. The error rate of proposed technique is l3.52% lwhile l9.008% in existing lwork. Therefore, lit lis clear lfrom lthe lsimulation lresults; lthe lproposed lwork lis lachieved significant lbetter lresults lthan lexisting lwork.