Table 5 Country-level dataset sizes and class distributions, sorted by dataset size and organized column-wise

Country	Dataset Size	Class		Country	Dataset Size	Class		Country	Dataset Size	Class
		Class 0 (%)	Class 1 (%)			Class 0 (%)	Class 1 (%)			Class 0 (%)	Class 1 (%)
CZ	55435	91.26	8.74	TN	4780	77.99	22.01	EG	241	99.59	0.41
IT	54354	85.60	14.40	CH	3836	92.39	7.61	GB	221	88.24	11.76
TR	37853	90.98	9.02	IR	2980	83.99	16.01	NZ	110	91.82	8.18
ES	33396	86.39	13.61	AR	2440	88.11	11.89	MK	103	97.09	2.91
CA	27131	85.65	14.35	LB	1937	93.86	6.14	GR	99	73.74	26.26
AU	23906	88.50	11.50	US	1344	89.96	10.04	RO	89	93.26	6.74
PT	6884	88.83	11.17	IL	1140	87.02	12.98	IE	69	97.10	2.90
BE	6534	90.54	9.46	OM	969	94.12	5.88	IN	59	91.53	8.47
KW	5725	93.15	6.85	CU	782	86.57	13.43	FR	55	100.00	0.00
HU	4892	93.40	6.60	BR	578	87.20	12.80	MT	45	100.00	0.00
NL	4869	84.86	15.14	SA	256	89.84	10.16

The table summarizes the dataset size and the proportion of Class 0 (MS worsening not confirmed) and Class 1 (MS worsening confirmed) for each participating country. Countries are sorted by dataset size in descending order. The data highlight substantial heterogeneity across countries, both in the number of available samples and in class balance. While Class 0 generally dominates, several countries exhibit severe class imbalance or complete absence of one class, underscoring the challenges of federated learning across non-identically distributed clinical datasets.

Quick links

Search