Calibration algorithms for continuous glucose monitoring from user calibration to intelligent adaptive no-user calibration

Continuous glucose monitoring (CGM) has evolved from an adjunctive display technology into a source of data for therapeutic decision-making, hypoglycemia alerts and automated insulin delivery. This transition has made calibration a safety-critical function. Factory-calibrated CGM removes routine user calibration but does not eliminate calibration itself; responsibility shifts to manufacturing control, sensor coding, embedded estimation, validation evidence and runtime safety governance. This Review examines the technical basis of calibration, the progression from user-calibrated to factory-calibrated CGM and the future development of intelligent adaptive no-user-calibration systems. We distinguish technically supported component methods from their developmental integration into a system that maintains calibration state during use. Evaluation of such systems will need to extend beyond mean absolute relative difference (MARD) to include performance in high-risk scenarios, consistency across algorithm versions, confidence output and regulatory evidence linking analytical performance, clinical safety and real-world use.