Interpreting artificial intelligence models: a systematic review on the application of LIME and SHAP in Alzheimer’s disease detection

Vimbi, Viswan; Shaffi, Noushath; Mahmud, Mufti

doi:10.1186/s40708-024-00222-1

Brain Informatics

Table 5 Studies incorporating SHAP framework for explaining model predictions

From: Interpreting artificial intelligence models: a systematic review on the application of LIME and SHAP in Alzheimer’s disease detection

References	Task	Data type	Sig. features	Classifier	Blackbox
[94]	HC versus sMCI versus	Numeric	Cognitive, PET, MRI, CSF	ML	RF
	pMCI versus AD		Genetics, Medical history
			Other Individual modalities
			Neuropsychological battery
[95]	HC versus MCI versus AD	Numeric	Volumetric measurements	ML	RF
			Cognitive tests, ApoE allele		XGBoost
			Demographic features
[96]	HC versus MCI versus AD	Numeric	Demographic, Clinical	ML	RF
			Neuropsychological
[97]	HC versus MCI versus AD	Numeric	Clinical history, Cognitive features	ML	XGBoost
			Anatomical Metabolic features		SVM, RF
			CSF biomarkers,ApoE4
[98]	HC versus AD	Numeric	Endoplasmic Reticulum stress related differentially expressed genes measures	ML	AdaBoost, RF
					LGBM, XGBoost
					kNN, NB, SVM
					LR
[99]	HC versus erMCI versus	Numeric	CDRSB, Age, MMSE, RAVLT	ML	XGBoost
	ltMCI versus AD	Categoric	measure, MRI middle temporal artery measure		RF
			Gender, ApoE
			FDG, MRI whole brain
			MRI entorhinal measur
			MRI hippocampus measure
[100]	aMCI versus AD	Numeric	Clinical, Demographic, ApoE genotype, Neuropsychological	ML	LR, RF
					XGBoost,SVM
[101]	HC versus MCI versus AD	Numeric	CDRSB, MMSE, EcogSPTotal	ML	Random Seeds
			RAVLT-perc-for-getting		SVM-SMOTE
			FAQ, ADAS11, MOCA		RF
			LDELTOTAL
[102]	HC versus erMCI versus	Numeric	MRI Volumetric measures, Age	ML	DT, LGBM
	ltMCI versus AD	Categorical	Gender, Education, ApoE		RF, SVM
[103]	HC versus sMCI versus	Numeric	MRI Volumetric measures	ML	XGBoost
	pMCI versus AD		ApoE4 alleles, Cognitive results		RF, SVM
			Socio-demographic data
[104]	HC versus AD	Numeric	Socio-demographic data medical history, Life Style measures	ML	RF
					XGBoost
[105]	HC versus ltMCI versus AD	Numeric	Amyloid beta features, glucose	ML	RF
			MRI measures
[106]	noDem versus vmDem versus mdDem versus moDem	Image	Image patterns	DL	CNN
				ML	KNN, SVM
[107]	HC versus AD	Numeric	Image patterns	DL	3DCNN
		Image	Demographic and Cognitive biomarkers
[108]	HC versus MCI	Numeric	Clinical information	ML	XGBoost, RF
			Neuropsychological test		AdaBoost, NB
			Data, Neuromaging-extracted biomarkers, gene data
			APOE-\(\epsilon\)4

Back to article page