WithinaBayesianstatisticalframework,wejointlyestimatethesourceandlensparametersandevaluatetherelative

evidencebetweenthelensedandunlensedmodels.Weinvestigatethewave-opticseffectsinducedbyapoint-mass

lensongravitationalwaves(GWs)frommassivebinaryblackholeswithcomparablecomponentmasses(q=0.9)

andassessthecapabilityofthespace-basedGWdetectorTaijitodetectsucheffects.Specifically,weinvestigatethe

impactoftheredshiftedlensmassMLz∈[3×105,3×107]M⊙,thedimensionlessimpactparameteryisscaledin

termsoftheEinsteinradiusofthelenswithy∈[10,50], thesourceredshiftzs∈[4,6],andthetotalsourcemass

Ms∈[105,107]M⊙onparameterestimationandmodelselection.OurresultsindicatethatlargerMLzvaluesreduce

thefittingfactor (FF), therebyamplifyingthewaveformresidualanditsassociatedresidual signal-to-noiseratio

(SNRres),whichenhancestheabilitytodistinguishbetweenlensedandunlensedmodels.Conversely,for y≥50,the

FF approaches 1 andSNRres becomes too small to alloweffectivemodel discrimination. For y≤30, the

degeneracybetweentheluminositydistancedLandMLzcanbeeffectivelybroken.WhiletheBayesfactordecreases

withincreasingsourceredshift zs, lensingsignatures remainidentifiableuptozs=6.The influenceof the total

sourcemassMsdependsontheoverlapoftheGWsignalwiththedetector’ssensitiveband.Overall,effectivemodel

discriminationgenerallyrequiresanSNR 5 res (correspondingtoFF 0.9999994fortheresultspresentedin

thiswork).