Predictive Behavior of a Computational Foot/Ankle Model through Artificial Neural Networks

Table 2

Performance of optimized networks.

	MSE
	Validation	Test
FFN	0.032	0.086	0.986
RBFN	0.053	0.054	0.980

Performance for optimized feedforward (FFN) and radial basis function (RBFN) networks as assessed by mean square error (MSE) and correlation (); both compared network-predicted stiffness to target values. MSE was calculated on the validation and test subsets, while was calculated for the entire dataset ( = 163). (Note: MSE was calculated using standardized input-output data, which fell within the range of .)