This thesis presents contributions on QCL-based spectroscopy and chemometric methods.  Additional investigation is done into signal-enhanced attenuated total reflection (ATR) spectroscopy, and a comprehensive study of multivariate analysis with convolutional neural networks and other chemometric methods is also included. The presented results show that fibre-coupled setups in transmission and ATR configurations are both suitable for measurements of glucose in peritoneal fluid. This system has a high potential for further testing in animal trials, and may be realized as a CGM device for humans.