Table 3.

Automated transcription and documentation performance by different settings.

Setting	ASRa transcription output			Automated TCCCb documentation
	Medical word error rate (%)	P valuec	Precision		Recall	F1 score
Setting 1: baseline	69.9	N/Ad	0.484		0.828	0.611
Setting 2: baseline + multi-style training	43.0	Setting 1 vs 2: <.001	0.634		0.824	0.717
Setting 3: baseline + multi-style training + updated language model	33.3	Setting 2 vs 3: <.001	0.803		0.824	0.813
Setting 4: baseline + multi-style training + updated language model + speech enhancement	46.3	Setting 3 vs 4: <.001	0.747		0.819	0.781

^aASR: automatic speech recognition.

^bTCCC: Tactical Combat Casualty Care.

^cMcNemar test with Bonferroni correction was used to calculate the statistical significance.

^dN/A: not applicable.