Fig. 3. Offline data analysis using polynomial regression to remove the circadian rhythm from the pharmacological response. (A) The raw data are shown with a fourth order polynomial least-squares regression fitted to segments of data before the injection and after the treatment had worn off. The polynomial is subtracted from the data to produce the residual data (B). A second, higher order polynomial is fitted to the residual data. The peak response and 50% peak response points are calculated for this polynomial.