Regresi Elastic Net dengan Peringkasan Luas untuk Mengukur Keakuratan Alat Non-Invasive Produk Tahun 2017 dan 2019

Abstract

Diabetes melitus is one of dangerous disease because it’s hard to be cured. This is shows it’s important for everyone to always control and checking their blood glucose levels to prevent make the diabetes melitus is getting worse. Non-invasive biomarking team from IPB currently developing blood glucose device measurement with non-invasive method. Now, the non-invasive biomarking team from IPB already created 2 products, design product for 2017’s and 2019’s with the output in the form of a residual intensity spectrum with respect to the time-domain. Therefore, calibration modeling is needed to predict blood glucose level. The best calibration modeling method for 2017’s device discovered by Herianti (2020) with elastic net regression and DDC algorithm for resolve the outlier. In 2019, measuring the blood glucose level were using different tools. This research aims to determine a more stable tool for measuring the blood glucose level with non-invasive method from 2 available tools, and to determine a more accurate summarization method of the intensity residual spectrum. More stable tool for measuring the blood glucose level is a 2017’s device. The summarization method in this research uses a trapezoidal area and 3 digit summarization approach. The result showed that the 2 summarization method didn’t have a significant different in accuracy.