A method of predicting substance levels from EEG data, comprises the steps of: analysing EEG data to obtain the average power for each of a plurality of predetermined frequency bands calculating a value from the average powers derived for each frequency band, said value being calculated by combining the average powers for each frequency band by dividing and/or multiplying according to a predetermined order and obtaining an estimate of the substance level from the equation: Y=bX+C where Y is the substance level to be predicted, X is the value and b and C are constants, wherein the substance is selected from hormones, neuro transmitters, bio markers, cortisol or the like. The method / apparatus may be used to measure hormone levels, measure neuro-transmitters, measure bio markers, measure cortisol, or the like.
