A device of the disclosure determines to which of a plurality of sleep stages in a sleep state including a wake stage a sleep state of a subject belongs, based on a sleep state function that is calculated using as variables a respiratory motion feature and a body motion feature that are respectively extracted from respiratory motion index values and body motion index values of the subject measured in time series. The sleep state function is a function including coefficient parameters that are set based on learning data including sleep stage determination results by a measurement of a sleep state for calibration, and respiratory motion features and body motion features that are measured simultaneously with the measurement for calibration.
