Table 3.

Comparing predictive performance of machine learning models with and without sound as an input.

Model		SDNN		Normalized-HF
		RMSE	MAPE	RMSE	MAPE
Without Sound	NN	23.77	45.00	12.56	84.20
	CART	20.58	36.11	11.60	75.43
	MARS	19.85	35.04	11.04	73.57
	RF	17.08	28.78	9.77	57.35
	GBM	17.98	26.55	10.30	51.98
	Our model	17.50	27.80	9.42	50.27
With Sound	NN	23.77	45.00	12.56	84.20
	CART	20.58	36.11	11.60	75.43
	MARS	19.21	33.97	10.83	72.64
	RF	16.96	28.48	9.58	56.79
	GBM	17.05	26.07	9.24	51.35
	Our model	17.06	26.56	8.90	44.36