Prediction of blood glucose levels in diabetic and non-diabetic patients using Bi-RNN Transformers

Diabetes is a prevalent chronic metabolic disorder with increasing global prevalence, posing significant health challenges. This study explores innovative noninvasive approaches, such as saliva-based glucose monitoring and predictive mathematical models, to accurately estimate blood glucose levels. Leveraging machine learning and deep learning techniques, including Bi-Directional RNNs and Multi-Attention Transformer models, this research proposes a non-invasive method to predict glucose levels in the blood at 30-minute intervals using wearable devices and physiological sensors. By providing continuous and accurate predictions without invasive procedures, this approach aims to improve diabetes management and improve health outcomes for diabetic patients. The study highlights the transformative potential of AI-driven approaches in revolutionizing diabetes care and emphasizes the importance of continuous advancements in non-invasive glucose monitoring in the blood for personalized and effective diabetes management.