A systematic beginners guideline for choosing suitable tuning parameters, particularly process noise covariance, has been proposed for the estimation of glucose. A modified Bergman’s minimal human glucose homeostasis model has been considered. It replicates Type-I Diabetic patient model with the inclusion of absorption of meal intake as well as subcutaneous glucose concentration. Extended Kalman Filter has been implemented and the choice of a suitable set of tuning parameters has been decided with the help of robustness and sensitivity metrics. A theoretical lower bound, called Cramer Rao Lower Bound, also has been evaluated and compared with those particular choices of process noise covariance Q. The results have been validated with Root Mean Squared plots and calculated Integral Time Squared Errors over 5000 Monte Carlo Runs.