Diabetes is a global problem affecting almost all countries. Rates of diabetes have increased in many countries in recent years due to increasing rates of obesity as a leading risk factor. According to the World Health Organization (WHO), the Diabetes prevalence has been rising more rapidly in middle and low income countries. By 2045, this is expected to rise up to 700 million worldwide. In the said region, almost 58% of diabetes patients are not diagnosed. Lack of an understanding, expensive, relatively painful measuring & monitoring techniques are among the reasons for these alarming statistics. Further, nearly 4 million people die each year due to this disease. In Sri Lanka, there were 1,232,800 reported cases of diabetes among 14,109,200 adults in 2019, with a prevalence of 8.7% diabetes in adults.
DiabiTech is a research focused on finding a non-invasive solution, which is portable, practical, accurate, and cost-effective carried out at SLIIT. This solution uses a Near-Infrared (NIR) spectroscopy as its optical source as it possesses maximum penetration depth in tissues. Glucose has few light absorption peaks, and at 940nm wavelength, attenuation of the signal due to blood constituents is minimum. Moreover, a red LED was used to enhance the accuracy of the actual glucose concentration prediction. As a cost-effective alternative, DiabiTech uses multiple CJMCU OPT-101 photodiodes with built-in Trans-impedance amplifiers that are capable of changing a low-level photodiode current signal to a usable voltage to compensate with the accuracy loss. The architectural design of DiabiTech is depicted in Figure 1.
The DiabiTech was extensively tested under three main phases. The test subjects consisted volunteers of different ages. In all stages, the test results obtained from DiabiTech was validated against a commonly used commercially available invasive blood-based device. During the first, one test subject was used to test the principle of Beer–Lambert law used in the device under the following test procedure. At first, each test subject’s, blood sugar level was recorded from both DiabiTech and the other devices after being on fasting for 12 hours. Then, the test subject was given 75g of sugar and the blood sugar levels were recorded at every 30 minutes after consumption for 3 hours
In the second phase, a calibration system was introduced, and testing was done with the same test subjects. The study shed light on developing a subjective equation based on the calibrated values. Three readings were taken after 12 hours of fasting, then three readings were taken after 2 hours of meal consumption, and this calibration aids in identifying the range of the blood glucose level of a person on a normal day. The calibration results were used as the base values to calculate the blood glucose value after the calibration. The data analysis showed that there is around 90% accuracy in the readings where the commercial applications have a recorded error rate of 20%. As the third phase, a random test was performed with ten calibrated test subjects, and the results are as follows in Figure 2.
According to the results, the DiabiTech solution has a maximum of 10% error rate that is a 90% accuracy compared to the accuracy rate of a conventional blood glucose meter. The results confirm that DiabiTech is capable of replacing conventional methods to provide results that are present a high accuracy solution with a cost-effective and painless. Furthermore, DiabiTech is in the process of finding an investor for the purpose of commercialization. This research won the Overall Best paper award of 2019 International Conference on Advancements in Computing, Colombo, Sri Lanka and the Track Best paper award for Health Informatics in the same conference. Furthermore, Merit Award, Business Category at the National Best Quality Software Awards 2019 organized by British Computer Society of Sri Lanka and became the First Runner-up in Code with WIE 2019 organized by IEEE WIE Sri Lanka Section. The research was conducted by the undergraduates following the Information Systems Engineering specialization of BSc Honors degree in Information Technology offered by the Department of Computer Systems Engineering of Faculty of Computing at SLIIT.
The research team was led by S.W.I. Udara including the team members of U.V.D.M.A. Ananda, K.M.W.K. Silva and A.K. De Alwis and Ms. Chathurangika Kahandawaarachchi a Lecturer attached to the Department of Computer Systems Engineering of Faculty of Computing at SLIIT was the supervisor of this research project.
0 comments: