Assessment of autonomic function and mental condition on cardio-respiratory variability and thermal regulation by using neural network

In this study, we could evaluate the relationship between autonomic function and mental condition by analysing constructed neural network. Subjective mental condition was investigated by POMS, and 3 factors called vigor, negative emotion and fatigue were used as output data for neural network. Then 19 autonomic function indicators were acquired to use as input data. Using these data sets, the feedforward neural networks were constructed on each subject. Moreover, each unit's output function of constructed networks was replaced by its linear approximate formula. By substituting the weight of constructed network for the formula, informational weights of autonomic function indicators to each mental factor were calculated. As the result, it was indicated that the sympathetic activity was dominant under vigorous condition, and autonomic nervous activity declined under fatigue condition.