RLMD-PA: A Reinforcement Learning-Based Myocarditis Diagnosis Combined with a Population-Based Algorithm for Pretraining Weights

Table 3

5-CV classification performances (F-measure, specificity, and G-means) obtained for automated myocarditis detection using various combinations of methods with the Z-Alizadeh Sani myocarditis dataset.

	F-measure					Specificity					G-means
Method	Min	Median	Max	Mean	Std.dev.	Min	Median	Max	Mean	Std.dev.	Min	Median	Max	Mean	Std.dev.
CNN-KCL [3]	0.718	0.746	0.798	0.748	0.031	0.814	0.852	0.870	0.845	0.022	0.772	0.795	0.838	0.797	0.025
CNN + random weight	0.681	0.702	0.746	0.704	0.026	0.788	0.800	0.838	0.806	0.020	0.742	0.759	0.795	0.760	0.021
CNN + ABC	0.735	0.745	0.798	0.758	0.026	0.826	0.835	0.864	0.842	0.018	0.787	0.795	0.839	0.806	0.021
CNN + RL	0.767	0.777	0.827	0.790	0.026	0.836	0.864	0.889	0.863	0.020	0.811	0.821	0.862	0.831	0.022
RLMD-PA (CNN + ABC + RL)	0.820	0.847	0.882	0.851	0.024	0.877	0.900	0.932	0.901	0.024	0.857	0.879	0.905	0.882	0.019