Stress assessment based on ECG using neural networks

This study describes a method of stress assessment based on electrocardiogram (ECG) using neural networks. Twelve male rats were exposed to daily six hours restraint and water-immersion stress for four weeks, another thirteen rats were used as a control group. In the fourth week, ECG was measured under non-anesthetic condition. Then the adrenals and thymus were weighed soon after sacrifice. An objective index of stress was derived from the weights using a five-layer neural network. The indices of the stressed rats were significantly different from those of the normal rats. The ECG data of seventeen rats were used for training of a three-layer neural network which was trained to estimate the index based on EGG. The other data were used for testing the trained network. The network could estimate the index accurately. The results of stress assessment using the network suggested the usefulness of the method.