Abstract
A method to estimate stress state in daily life is necessary for depression and life-style related diseases prevention. We focused on three states; relax state, mental arithmetic stress state, and monotonous work stress state. To estimate stress state, we extracted nine physiological indices of stress from physiological data collected during stress experiment, consisting in two types of PASAT assigned to 10 experiment participants. Because variation range of each index is different between individuals and types of stress, we divided estimation process into two steps: step I to discriminate relax state from stress state, and step 2 to discriminate mental arithmetic stress state from monotonous work stress state. For each step, we selected the most relevant set of indices that enable to estimate stress state regardless of individuals. Then we processed two steps, and estimated the stress state of 10 experiment participants in experiments by cross-validation. We could achieve high estimation accuracy (average: 75.3%, standard deviation: 18.4%), and Confirm this method can be an effective solution to estimate various types of stress state regardless of individuals.
