Table 2.

Performance of Random Forest Federated Learning with different levels of model aggregation.

Models	HIV prediction			STI prediction
	Global model capacity	AUC	Accuracy	Global model capacity	AUC	Accuracy
Local model		0.89	0.82		0.96	0.92
FL with 100% aggregated	4.36 MB	0.91	0.84	2.28 MB	0.98	0.94
FL with 70% aggregated	3.25 MB	0.91	0.83	1.67 MB	0.98	0.93
FL with 50% aggregated	1.75 MB	0.90	0.84	1.09 MB	0.97	0.93
FL with 30% aggregated	1.42 MB	0.90	0.83	840 KB	0.97	0.93
FL with 20% aggregated	769 KB	0.90	0.83	508 KB	0.97	0.93
FL with 10% aggregated	348 KB	0.89	0.82	345 KB	0.96	0.92
FLCNN	964 MB	0.87	0.78	964 MB	0.74	0.70
Xu et al.7	NA	0.72	NA	NA	0.67–0.75	NA