Non Invasive Blood Sugar Detection Using The Extraction Method Of Principal Component Analysis

Diabetes Mellitus (DM) is a non communicable disease but has a high level of danger because it can cause complications such as heart disease, stroke, blood vessels, and many more. In 2000, the population of people with DM in Indonesia was ranked 4th in the world with a total of 8.4 million people. To reduce the level of DM, early detection is needed, but the existing tools are usually invasive which is using blood on the fingertips, this certainly reduces public interest. Therefore, a noninvasive (without injuring the body) detection tool is needed to help increase public interest in routine checks for diabetes. The processing of detection tools requires the right method so that the tool can have a good accuracy. One method to perform signal processing is Principal Component Analysis (PCA). The advantage of the PCA method is that the learning method is quite easy, in addition, according to previous researchers, PCA has a higher accuracy than the other method. The results of the PCA extraction are then classified using a Support Vector Machine (SVM) with a linear line to separate DM and non-DM data. The results of the classification of this study obtained an accuracy of 83.333%.