This paper observes advancements in the field of non-invasive blood glucose monitoring via a saliva-based tongue depressor biomarker sensor, augmented with a customised variation of the XGBoost algorithm. The research spotlights developing a method that accurately measures blood glucose concentration in saliva, meeting the severe accuracy requirements vital for effective diabetes management. By utilising prior research and dataset analyses, several sensors and algorithms are levied according to their precision, and responsiveness to the smallest variations in biomarker levels. Furthermore, the paper unveils an improved version of the XGBoost algorithm, moulded to enhance the overall accuracy and reliability of blood glucose readings from saliva samples. The overarching objective is to establish a practical, precise and reliable alternative to traditional invasive blood tests like finger-pricking, offering diabetic persons a more comfortable and adept monitoring solution.