Implementation of Modified K-Nearest Neighbor Algorithm in Electronic Nose System to Detect Gastroesophageal Reflux Disease

Abstract

This research was conducted to optimize the performance of the Electronic Nose (E-Nose) by applying the Modified K-Nearest Neighbor method. The E-Nose system was built using 3 gas sensors, namely Mq-7 (Carbon Dioxide), Mq-135 (Ammonia), Mq-136 (Hydrogen Sulfide and Carbon Dioxide), and ESP-32 as a microcontroller. The use of the E-Nose system aims to obtain a sensor response to the breathing air of GERD patients and healthy patients, then the sensor response results will be processed by the microcontroller and classified using the MK-NN method. This study used 120 training data (60 GERD patient data and 60 healthy patient data) and used 80 randomized trial data. The test results of the E-Nose system using the MK-NN method with a combination of 60% training data and 40% test data and using a value of K=3, K=5 and, K=7 show that this system can provide better performance than the K-NN method with an accuracy of 89.55 %, 88.00% precision for GERD patients, and 91.00% precision for healthy patients.

Access to Document

Other files and links

• Link to publication in Scopus

Fingerprint