TABLE 4.

Comparison between single and ensemble models, regardless of the presence of crosslinks.

Metrics	Accuracy	Precision‐macro	Sensitivity‐macro	F1 score‐macro
MAIA models
AP model	0.8113	0.8067	0.8078	0.8004
Lat model	0.9057	0.902	0.9079	0.9008
Concat model	0.8585	0.8524	0.8590	0.8511
Merge model
Dual images	0.7972	0.7898	0.7906	0.7821
AP images	0.7075	0.6889	0.7014	0.6846
Lat images	0.8868	0.8894	0.8797	0.8774
Average	0.8278	0.8215	0.8244	0.8161
Ensemble models
All	0.9245	0.9226	0.9248	0.9181
AP + LP	0.9434	0.9405	0.9416	0.9388
AP + Lat + Concat	0.9340	0.9304	0.9332	0.9302
AP + Lat + Merge	0.9340	0.9337	0.9297	0.9289
Lat + Concat	0.9340	0.9304	0.9332	0.9299
Average	0.9340	0.9315	0.9325	0.9292

Abbreviations: Concat, concatenated; MAIA, Medical Artificial Intelligence Aggregator.