Table 1.

AUC of the seven models using four central datasets of patients in the relapse and nonrelapse models

Method	Evaluation	Centre A	Centre B	Centre C	Centre D
FedAvg	AUC	0.672	0.726	0.711	0.798
	Accuracy	0.634	0.536	0.660	0.639
FedProx	AUC	0.658	0.718	0.731	0.766
	Accuracy	0.607	0.681	0.660	0.557
Moon	AUC	0.663	0.661	0.724	0.775
	Accuracy	0.634	0.565	0.679	0.689
HarmoFL	AUC	0.684	0.723	0.707	0.773
	Accuracy	0.616	0.696	0.604	0.820
pFedMe	AUC	0.689	0.706	0.744	0.769
	Accuracy	0.634	0.681	0.623	0.787
pFedFSL	AUC	0.649	0.742	0.728	0.796
	Accuracy	0.652	0.623	0.679	0.656
RFLM	AUC	0.710	0.798	0.809	0.869
	Accuracy	0.598	0.710	0.755	0.689

RFLM robust federated learning model.