Table 2.

Evaluation (F1 scores) of deep learning models for the classification of Health Belief Model constructs and speech acts.

Variable				BERTweet		BERTweet-Covid19		Ensemble (BERTweet+BERTweet-Covid19)
Health Belief Model constructs
	Per class performance
		Perceived barriers	0.64		0.64		0.77
		Perceived severity	0.76		0.80		0.83
	Overall model performance
		Macro average	0.71		0.74		0.81
Speech acts
	Per class performance
		Assertion	0.80		0.78		0.82
		Declaratives	0.92		0.83		0.89
		Directive	0.80		0.73		0.67
		Question	0.83		0.88		0.83
		Statement	0.67		0.59		0.69
	Overall model performance
		Macro average	0.80		0.76		0.78