Machine learning models for predicting metabolic dysfunction-associated steatotic liver disease prevalence using basic demographic and clinical characteristics

Zhu, Gangfeng; Song, Yipeng; Lu, Zenghong; Yi, Qiang; Xu, Rui; Xie, Yi; Geng, Shi; Yang, Na; Zheng, Liangjian; Feng, Xiaofei; Zhu, Rui; Wang, Xiangcai; Huang, Li; Xiang, Yi

doi:10.1186/s12967-025-06387-5

Table 2 Performance of various machine learning models for predicting MASLD across internal validation, internal testing, and external validation cohorts

From: Machine learning models for predicting metabolic dysfunction-associated steatotic liver disease prevalence using basic demographic and clinical characteristics

Cohort	Model	AUC	95% CI	Optimal cutoff	Accuracy	Sensitivity	Specificity	Precision	F1 score
Internal validation cohort	LR	0.806	(0.787–0.825)	0.571	0.728	0.749	0.713	0.659	0.701
	MLP	0.809	(0.790–0.827)	0.531	0.735	0.724	0.744	0.676	0.699
	XGBoost	0.806	(0.788–0.825)	0.566	0.727	0.730	0.725	0.662	0.694
	Bagging	0.801	(0.782–0.820)	0.578	0.727	0.709	0.740	0.669	0.688
	RF	0.801	(0.782–0.820)	0.578	0.719	0.750	0.697	0.646	0.694
	NB	0.797	(0.778–0.816)	0.464	0.714	0.747	0.689	0.640	0.689
	LightGBM	0.796	(0.776–0.815)	0.577	0.716	0.731	0.705	0.647	0.687
	SVM	0.769	(0.747–0.791)	0.594	0.736	0.704	0.760	0.685	0.694
	KNN	0.767	(0.746–0.788)	0.619	0.694	0.753	0.650	0.614	0.677
	DT	0.714	(0.694–0.734)	0.569	0.711	0.736	0.692	0.639	0.684
Internal testing cohort	LR	0.798	(0.771–0.825)	0.617	0.711	0.793	0.649	0.626	0.700
	MLP	0.796	(0.769–0.823)	0.579	0.697	0.847	0.585	0.602	0.704
	XGBoost	0.798	(0.771–0.825)	0.641	0.709	0.857	0.599	0.612	0.714
	Bagging	0.791	(0.763–0.818)	0.597	0.722	0.718	0.724	0.658	0.687
	RF	0.792	(0.764–0.819)	0.556	0.720	0.700	0.734	0.661	0.680
	NB	0.786	(0.759–0.814)	0.664	0.695	0.826	0.597	0.603	0.697
	LightGBM	0.789	(0.762–0.817)	0.582	0.719	0.749	0.696	0.646	0.693
	SVM	0.739	(0.707–0.770)	0.610	0.717	0.674	0.748	0.664	0.669
	KNN	0.748	(0.718–0.778)	0.614	0.685	0.704	0.670	0.612	0.655
	DT	0.721	(0.693–0.749)	0.569	0.718	0.746	0.696	0.645	0.692
External validation cohort	LR	0.831	(0.825–0.837)	0.923	0.777	0.756	0.790	0.695	0.724
	MLP	0.823	(0.817–0.829)	0.691	0.755	0.734	0.769	0.667	0.699
	XGBoost	0.784	(0.777–0.791)	0.788	0.763	0.574	0.883	0.755	0.652
	Bagging	0.757	(0.751–0.764)	0.874	0.758	0.633	0.837	0.710	0.669
	RF	0.763	(0.755–0.770)	0.726	0.756	0.571	0.873	0.740	0.645
	NB	0.798	(0.792–0.805)	0.997	0.742	0.762	0.729	0.640	0.696
	LightGBM	0.807	(0.800-0.814)	0.912	0.766	0.723	0.794	0.689	0.706
	SVM	0.607	(0.598–0.615)	0.641	0.731	0.436	0.917	0.769	0.556
	KNN	0.774	(0.767–0.781)	0.880	0.738	0.628	0.807	0.673	0.650
	DT	0.670	(0.664–0.676)	0.569	0.733	0.391	0.949	0.829	0.532

Abbreviations: LR, logistic regression; MLP, multilayer perceptron; XGBoost, extreme gradient boosting; Bagging, bootstrap aggregating; DT, decision tree; KNN, k-nearest neighbors; LightGBM, light gradient boosting machine; NB, naive bayes; RF, random forest; SVM, support vector machine; AUC, area under the curve; MASLD, metabolic dysfunction-associated steatotic liver disease

Back to article page

ISSN: 1479-5876

Contact us

Submission enquiries: Access here and click Contact Us
General enquiries: journalsubmissions@springernature.com

Journal of Translational Medicine

Contact us