This dissertation focuses on the development and improvement of the minimally invasive continuous glucose monitoring systems (CGMS). This dissertation gives emphasis in solving the problem of the accuracy and reliability of the blood glucose (BG) concentrations estimated by CGMS. The main novelty of this dissertation is that it takes into account the dynamic relationship between BG and interstitial glucose (IG) in the algorithms for BG estimation. Furthermore, pattern recognition and learning techniques are explored combining the information provided by the monitor itself and variables related to the patient's clinical condition, to detect correct and incorrect measurements, without requiring any reference measurement. The performance of the proposed methodological approaches has been validated with data from different clinical studies, in both diabetic and critically-ill patients
© 2001-2024 Fundación Dialnet · Todos los derechos reservados