Table 1.

Summary of the systematic analysis of the state-of-the-art thyroid disease studies.

Authors	Year	Sample Size	Dataset Source	Model	Classes	Evaluation Metrics	Results
[9]	2020	-	ToxCast	LR RF SVM XGB ANN	2	F1-score	(TPO) XGB-83% and (TR) RF-81%
[11]	2018	7200 samples, 21 attributes	UCI	SVM, Multiple Linear Regression(MLR), NB and DT	2	Accuracy	MLR 91.59% SVM 96.04% Naive Bayes 6.31% Decision Trees 99.23%
[12]	2020	7547, 30 features	UCI	multi-kernel SVM	3	Accuracy, Sensitivity, and Specificity	Accuracy (97.49%), Sensitivity (99.05%), and Specificity (94.5%)
[13]	2021	3771 samples, 30 attributes	UCI	DT, KNN, RF, and SVM	4	Accuracy	KNN 98.3% SVM 96.1% DT 99.5% RF 99.81%
[14]	2021	519 samples	diagnostic center Dhaka, Bangladesh	SVM, DT, RF, LR, and NB. Recursive Feature Selection (RFE), Univariate Feature Selection (UFS) and PCA	4	Accuracy	RFE, SVM, DT, RF, LR accuracy—99.35%
[15]	2021	1250 with 17 attributes	external hospitals and laboratories	SVM, RF, DT, NB, LR, KNN, MLP, linear discriminant analysis (LDA) and DT	3	Accuracy	DT 90.13, SVM 92.53 RF 91.2 NB 90.67 LR 91.73 LDA 83.2 KNN 91.47 MLP 96.4
[16]	2021	7200 patients, with 21 features	UCI	multiple MLP	3	Accuracy	multiple MLP 99%
[17]	2021	690 samples, 13 features	datasets from KEEL repo and District Headquarters teaching hospital, Pakistan	KNN without feature selection, KNN using L1-based feature selection, and KNN using chi-square-based feature selection	3	Accuracy	KNN 98%
[18]	2021	3772 and 30 attributes	UCI	RF, sequential minimal optimization (SMO), DT, and K-star classifier	2	Accuracy	K = 6, RF 99.44%, DT 98.97%, K-star 94.67%, and SMO 93.67%
[19]	2022	3163	UCI	DT, RF, KNN, and ANN	2	Accuracy	Best performance Accuracy RF 94.8%
[21]	2022	215 with 5 features	UCI	KNN, XGB, LR, DT	3	Accuracy	KNN 81.25 XGBoost 87.5 LR 96.875 DT 98.59
[20]	2022	3152, 23 features	UCI	DNN	2	Accuracy	Accuracy 99.95%