Table 3 Performance scores for different learning paradigms across countries

From: Personalized federated learning for predicting disability progression in multiple sclerosis using real-world routine clinical data

		ROC–AUC ↑					AUC–PR ↑
Country	Dataset Size	Federated	Adaptive	Fine-tuned	Centralized	Local	Federated	Adaptive	Fine-tuned	Centralized	Local
CZ	55435	0.8909	0.9211	0.9193	0.8781	0.8669	0.5875	0.7214	0.6990	0.5568	0.4600
IT	54354	0.7946	0.8168	0.8108	0.7769	0.7603	0.4590	0.5099	0.4924	0.4312	0.3491
TR	37853	0.8365	0.8887	0.8854	0.8467	0.8498	0.4427	0.5953	0.5716	0.483	0.4161
ES	33396	0.7680	0.8027	0.7969	0.7713	0.7609	0.3819	0.4564	0.4393	0.4055	0.3381
CA	27131	0.7370	0.7619	0.7615	0.7488	0.7332	0.3383	0.3965	0.3949	0.3893	0.3407
AU	23906	0.7300	0.7679	0.7657	0.7490	0.7287	0.3098	0.3962	0.3805	0.3882	0.3129
PT	6884	0.7449	0.8475	0.8525	0.8252	0.7972	0.2627	0.4524	0.4774	0.449	0.3562
BE	6534	0.6495	0.8115	0.8156	0.7963	0.7987	0.1559	0.3980	0.4050	0.3553	0.2987
KW	5725	0.7445	0.9137	0.9128	0.8761	0.9164	0.1661	0.5104	0.5111	0.4558	0.4850
HU	4892	0.7128	0.9608	0.9632	0.9495	0.9549	0.3099	0.7810	0.779	0.7311	0.5924
NL	4869	0.5614	0.6595	0.6782	0.6873	0.7107	0.1797	0.2539	0.2756	0.2826	0.2650
TN	4780	0.7857	0.9535	0.9535	0.9319	0.9312	0.503	0.8639	0.8664	0.8178	0.8040
CH	3836	0.6212	0.7650	0.7700	0.7925	0.7274	0.1232	0.3042	0.3196	0.3084	0.1952
IR	2980	0.6396	0.8269	0.8330	0.8158	0.7471	0.2514	0.5570	0.5702	0.5345	0.3682
AR	2440	0.6714	0.8856	0.8719	0.8274	0.8784	0.2625	0.6287	0.6331	0.5801	0.5946
LB	1937	0.5955	0.7589	0.7398	0.7314	0.6553	0.1171	0.3343	0.3187	0.2756	0.2255
US	1344	0.5627	0.7303	0.7368	0.7437	0.7044	0.1383	0.2493	0.2797	0.3128	0.2433
IL	1140	0.6937	0.8750	0.8782	0.8537	0.8503	0.2604	0.5986	0.6310	0.5198	0.5181
OM	969	0.5339	0.8472	0.8731	0.8093	0.7981	0.0962	0.5421	0.5897	0.4763	0.2995
CU	782	0.5625	0.7971	0.8050	0.8062	0.8266	0.1864	0.4190	0.4744	0.4775	0.5260
BR	578	0.5768	0.8063	0.7680	0.7307	0.7070	0.1434	0.4308	0.4200	0.4677	0.4605
SA	256	0.6749	0.8915	0.8677	0.9374	0.8827	0.2466	0.659	0.5619	0.6851	0.7674
GB	221	0.6520	0.6060	0.6880	0.8510	0.5333	0.2576	0.3250	0.3007	0.4774	0.2529
NZ	110	0.3286	0.6095	0.4857	0.4190	0.5873	0.0738	0.1247	0.1094	0.0810	0.1057
GR	99	0.7240	0.8714	0.8703	0.9292	0.7998	0.6495	0.8794	0.8777	0.9082	0.8048
WA		0.7835	0.8398	0.8375	0.8092	0.7983	0.4081	0.5346	0.5221	0.4605	0.3874

The table reports ROC–AUC and AUC–PR metrics for federated learning, adaptive, fine-tuned, centralized, and local approaches. The bold values indicate the best-performing results within each row, where higher values are better, as shown by the arrows in the column headers.

Back to article page

Table 3 Performance scores for different learning paradigms across countries

Search

Quick links