Rapid energy expenditure estimation for ankle assisted and inclined loaded walking

Table 2 Comparison of linear regression and neural network energy expenditure estimates made per gait cycle for different use cases during inclined loaded walking

Model	Metric	Novel Condition	Novel Subject	Both-Novel	Subject Vertical Force^a	Raw Subject^b
Linear Regression	RMSE^c\(\left (\frac {\mathrm {W}}{\text {kg}}\right)\)	0.62	0.94	0.95	0.98	1.39
	Error	6.7%	12.1%	13.7%	12.3%	16.5%
Neural Network	RMSE \(\left (\frac {\mathrm {W}}{\text {kg}}\right)\)	0.56	0.83	0.78	0.86	0.88
	Error	6.1%	9.7%	11.7%	10.0%	11.2%

^aThe subject vertical force use case was the novel subject use case with inputs restricted to vertical ground reaction forces and EMG signals
^bThe raw subject use case was the novel subject use case without any data preprocessing other than rectifying the EMG signals
^cRMSE is the root mean squared error normalized by the average subject mass

ISSN: 1743-0003