Electrical Engineering Doctoral Program student of the Faculty of Engineering (FT) of Universitas Indonesia (UI), Dr. Ernia Susana, developed a new method to monitor blood sugar levels without requiring needles. This method utilizes light technology called Photoplethysmography (PPG) signals and Artificial Intelligence (AI) which is expected to be able to monitor diabetes patients’ blood sugar levels more easily, comfortably, and affordably.
The PPG technique gauges blood volume in the blood vessels using light. The biggest challenge of this method is signal disruption due to movement and other factors. To deal with this problem, Dr. Ernia utilized the time-frequency analysis technique (TFA), which is based on the short-time Fourier transform (STFT) to improve the signal quality.
Dr. Ernia did three steps in this research; monitoring system development, TFA technique implementation, and secondary data testing. In monitoring system development, Dr. Ernia fused electronic filters and AI to create a more accurate blood sugar level monitoring system. In this step, data was taken from 80 adults during the Covid-19 pandemic. The best model found was Ensemble Bagged Trees (EBTA) with a 97.8% accuracy.
The TFA technique was then used to improve signal quality inputted to the artificial intelligence model. The research shows that the Support Vector Machine (SVM) model could reach a 91.3% accuracy with a training time of 9.25 seconds, while the Bidirectional Long Short Term Memory (BLSTM) reached an accuracy of 87% with a training time of 15 seconds.
Dr. Ernia explained that the further development of this research utilizes recommendations for using BLSTM-based deep learning algorithms with optimization techniques which can improve accuracy and decrease training time. In addition, this research also suggests Android-based BGL monitoring application development to process data faster and more responsive.
The Dean of FTUI, Prof. Dr. Ir. Heri Hermansyah, S.T., M.Eng., IPU., conveyed, “The development of the non-invasive BGL monitoring technique offers a potential solution to increase patient’s obedience in monitoring blood glucose levels routinely. With a higher accuracy and a more efficient training time, this technology can become an important tool in early detection and diabetes management in the future. Further research is hoped to be able to optimize this technology through algorithm combination and mobile device-based application development.”
Dr. Ernia’s research was poured into her dissertation, “Pengembangan model Kecerdasan Buatan pada Klasifikasi Non-Invasive Kadar Glukosa Darah Berbasis Sinyal Photoplethysmography dengan Penguatan Ekstraksi Fitur Time Frequency Analysis untuk Deteksi Dini Diabetes.” The Open Doctoral Promotion session was held on July 3, 2024, at the Smart Meeting Room of the FTUI Dean Building.
Owing to this research, Dr. Ernia managed to defend her dissertation and obtained a Doctorate in the Electrical Engineering field with a Cum laude predicate and a GPA of 3.98. Dr. Ernia has become the 173rd doctor in the Electrical Engineering program and the 555th doctor in FTUI.
