SIMULATION OF PRE-PROCESSING SYSTEM FOR NON-INVASIVE HBA1C MONITOR

• Main Author: Alfianto, Vincent
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/49504
• Institution: Institut Teknologi Bandung

Diabetes mellitus (DM) is a disease that is characterized by the imperfect regulation of blood glucose. This disease has a prevalence of 10.9% in Indonesian people 15 years of age and above. Effective blood glucose control is important to avoid serious secondary complications from DM such as microvascular and macrovascular changes that can lead to neuropathy, nephropathy, and diabetic retinopathy. From the various method available for DM patients, monitoring of glycolytic Hemoglobin (HbA1c) is the standard choice for long-term observation of blood glucose. HbA1c is formed from the non-enzymatic glycosylation process of hemoglobin that is exposed to blood glucose. From this fact, we can reaffirm that there is a correlation between blood sugar concentration and HbA1c. At present, the examination of HbA1c levels is usually done in clinical laboratories in an invasive way, namely by conducting blood tests in a clinical laboratory. However, this examination method has shortcomings such as a long testing time and relatively expensive laboratory test costs. To overcome these shortcomings, in this final project, a device is designed to detect HbA1c levels non-invasively by using the Spectroscopy method. Absorbance spectroscopy at specific wavelength regions can analyze vibrational transitions from the HbA1c building molecules and can be used to quantify HbA1c levels indirectly. This final project is done by 3 people, this final year project especially focuses on designing the pre-processing system of this non-invasive HbA1c monitor. The design starts with data synthesis and analog signal processing. First, data synthesis is done by searching for photoplethysmography (PPG) dataset that has various subject conditions. The electrical signals have a small amplitude and contain significant noise, so analog signal processing is needed in the form of amplification and filtering processes. Signals that have been amplified and through the filtering process are outputs from the pre-processing system. The output of this system is connected with a digital processing system, which is a system that converts to digital signals and further processing of statistical signal parameters. The change of signal from analog to digital is intended to facilitate easier data processing. Then to make a more accurate data conversion system from a digital signal to HbA1c levels, a machine learning oriented algorithm is used. Due to limitations caused by the COVID-19 pandemic, the implementation and testing for this final year project are still in the form of circuit simulation. Further work needed for the physical realization and testing of the design.