Table 1 Comparison between STL and MTL in heart disease prediction. MAND-LR, MAND-MLP, MAND-LSTM, and MAND-MHSA denote the MAND architecture integrated with logistic regression, multilayer perceptron (MLP), LSTM, and multi-head self-attention as ICD feature extraction modules, respectively. FM and DCN represent CTR-based approaches. BAC: balanced accuracy; FPR: false positive rate; FNR: false negative rate.

Backbone model	STL/MTL	Log loss	AUC	BAC	Precision	Recall	F1 score	FPR	FNR
MAND-LR¹⁸	STL	0.3838	0.8602	0.7067	0.7422	0.4663	0.5728	0.0529	0.5337
MAND-LR¹⁸	MTL	0.3861	0.8580	0.7002	0.7352	0.4535	0.5609	0.0531	0.5465
MAND-MLP¹⁸	STL	0.3581	0.8698	0.7204	0.8041	0.4789	0.6003	0.0381	0.5211
MAND-MLP¹⁸	MTL	0.3640	0.8695	0.7239	0.7507	0.5019	0.6016	0.0541	0.4981
MAND-LSTM¹⁸	STL	0.3543	0.8774	0.7162	0.8555	0.4576	0.5962	0.0252	0.5424
MAND-LSTM¹⁸	MTL	0.3467	0.8787	0.7205	0.8366	0.4710	0.6026	0.0300	0.5290
MAND-MHSA¹⁸	STL	0.3621	0.8765	0.7512	0.7224	0.5744	0.6400	0.0720	0.4256
MAND-MHSA¹⁸	MTL	0.3614	0.8579	0.7411	0.7286	0.5486	0.6259	0.0664	0.4514
FM¹⁹	STL	0.4421	0.8467	0.7012	0.7056	0.4658	0.5612	0.0634	0.5342
FM¹⁹	MTL	0.3915	0.8576	0.7215	0.7084	0.5114	0.5940	0.0684	0.4886
DCN²²	STL	0.3566	0.8745	0.7232	0.8123	0.4829	0.6057	0.0365	0.5171
DCN²²	MTL	0.3523	0.8749	0.7308	0.7867	0.5062	0.6160	0.0446	0.4938

Quick links

Search