PROBLEM TO BE SOLVED: To select a proper model according to a state of an estimation object.SOLUTION: When a work awakening degree is estimated, determination feature amounts CSI, SDNN are calculated from RRI time series data, for determining a correlation factor value of the feature amounts, then the correlation factor value is compared with a threshold for determining magnitude of variation of a respiratory cycle. Then when the variation of the respiratory cycle is large, a first estimation model based on a time area feature amount, is selected, then, the work awakening degree is estimated, based on the first estimation model, and time area feature amounts Mean, pNN50, CSI, CVI calculated from the RRI time series data. On the other hand, when the variation of the respiration cycle is small, a second estimation model based on a frequency area feature amount is selected, then, the work awakening degree is estimated based on the second estimation model, and frequency area feature amounts LF, HF, log(LF), log(HF) calculated from the RRI time series data.SELECTED DRAWING: Figure 1
