Maximizing Glucose Detection Accuracy with Sensors Based Optimized Deep Learning Algorithm
Dr. Pooja SharmaCentre of Research Impact and Outcome, Chitkara University, Rajpura, Punjab, India. pooja.sharma.orp@chitkara.edu.in0009-0006-8507-0543
Dr. Honganur Raju ManjunathDepartment of Physics, Faculty of Engineering and Technology, JAIN (Deemed-to-be University), Bangalore, Karnataka, India. hr.manjunath@jainuniversity.ac.in0000-0002-1764-0683
Dr. Geetika M. PatelDepartment of Community Medicine, Parul University, Limda, Waghodia, Vadodara, Gujarat, India. vicepresident_86@paruluniversity.ac.in0000-0003-3789-184X
Dr. Sudhanshu GongeDepartment of Computer Science and Engineering, Symbiosis Institute of Technology, Symbiosis International (Deemed University), Pune, India. sudhanshu.gonge@sitpune.edu.in0000-0002-4092-6874
Dr. Awakash MishraProfessor, Maharishi School of Engineering & Technology, Maharishi University of Information Technology, Uttar Pradesh, India. awakash.mishra@muit.in0009-0009-8318-950X
Dr. Hitesh KalraChitkara Centre for Research and Development, Chitkara University, Himachal Pradesh, India. hitesh.kalra.orp@chitkara.edu.in0009-0000-5064-2165
Keywords: Diabetes, Data Augmentation, Accuracy, Non-Invasive, Multi-Layered, Hyperparameter.
Abstract
Non-Invasive Blood Glucose Measurement: Now, the most common way of testing blood sugars is an invasive route - blood-based detection can be quite an awkward and painful test for patients. Safe glucose sensors are appearing on the market as a replacement for the pain of capillary glucose measuring; however, they are often not accurate. Addressing these issues, this work presents a deep learning-based algorithm specifically designed for non-invasive glucose sensors, aiming to improve accuracy. An algorithm using a multi-layer neural network for extracting features from non-invasive glucose sensors the fact network has been trained on the non-invasive sensor data measures and the blood analysis measures. If all else fails, optimize the algorithm to work better through methods such as hyperparameter tuning and data augmentation.