In an implantable medical device for monitoring blood-glucose concentration in the blood, metabolic oxygen consumption is derived by measuring physiological metrics related to mixed venous oxygen concentration. Blood-glucose concentration is determined using correlations of blood-glucose concentration with measures of metabolic oxygen consumption including oxymetric, temperature, and electrocardiographic data. Additional physiological sensor measurements may be used to enhance the accuracy of the analysis of blood-glucose concentration. By using a combination of oxymetric and other physiological metrics, blood-glucose concentration can be reliably calculated over a wide range. The device compares the blood-glucose concentration with upper and lower acceptable bounds and generates appropriate warning signals if the concentration falls outside the bounds. The device may also control a therapeutic device to maintain blood-glucose concentration within an acceptable range. A calibration technique determines patient-specific parameters for use in the calculation of blood-glucose concentration.
