The complex dynamics of glucose metabolism depend on multiple individual factors such as chronobiology and meal patterns and are difficult to predict. This project aims to simulate glucose metabolism patterns in different fasting forms, using continuous glucose monitoring (CGM) data to model metabolic reactions to food ingestion and fasting. The thesis has two main objectives, one is the implementation of a generative adversarial network approach, that can synthesize daily glucose time series specific to different fasting forms. The other one is the development of new time-dependent biomarkers from continuous glucose monitoring originating from clinical trials (DIfE, Dr. Ramich and Charité, Dr. Pappe).