Diabetes Mellitus is an incurable disease related to the metabolism of insulin This disease has the potential to lead one to complication such as cardiovascular diseases or kidney diseases without early precaution. There are two type of Diabetes Mellitus, with type two considered to have higher potential to lead to danger due to the often mild or even non-existent symptoms until one reaches chronic point. Currently, the golden standard to detect Diabetes Mellitus is done through HbA1C tests, which calculates the average blood sugar level among the lifecycle of a red blood cell. Beside HbA1C test, home sugar level test system are also used to monitor blood sugar level during the testing time. Both approach requires the subject to harm themselves in order to obtain the data, hence making this an invasive approach. This research is done in order to develop an alternative system capable of detecting type two Diabetes Mellitus with non-invasive approach, using Photoplethysmography sensor GY-MAX30100, Arduino Nano v3, and K-Nearest Neighbor Algorithm. It is shown that using 10 data for testing, result shows an accuracy of 80% using K=9, Augmentation Index, and Ratio of Amplitude Systolic toward Pulse Interval features. It is also shown that system returns precision of 71%, sensitivity 100%, and F1 score of 83% for healthy class, while system returns precision of 100%, sensitivity 60%, and F1 score of 75% for type two Diabetes Mellitus class.
Copyrights © 2022