Prediksi Pasien Terkena Penyakit HIV Dengan Algoritma K-Nearest Neighbor (KNN) di RSUD Dr.Chasbullah Abdul Madjid Kota Bekasi

Abstract

HIV/AIDS is a global health concern present in almost every part of the world. The advancement of information system technology has helped solve various problems across different fields, one of which is through the application of data mining. The utilization of data mining is not only implemented in the healthcare sector but also in the technology industry. One method to predict patients who potentially have HIV/AIDS is by using Machine Learning (ML). ML aims to train models with algorithms capable of performing statistical analysis using Supervised Learning techniques to generate accurate predictions. Prediction is one of the most important statistical elements in the decision-making process. This research uses the K-Nearest Neighbor algorithm, which classifies data based on the majority class of K nearest neighbors. The algorithm is combined with the SMOTETomek technique as a resampling method to address data containing noise and class imbalance problems. The dataset used to train the K- Nearest Neighbor model comes from the Voluntary Counselling and Testing (VCT) unit with a total of 2,205 data points. The disease testing prediction results are then processed and visualized in a website format. Based on testing conducted using Confusion Matrix, the model’s performance measurement results show and Accuracy value of 97.96%, Precision of 78.61%, Recall of 98.88%, and f1-score of 84.45%. The results indicate that the use of machine learning is quite effective for implementation in HIV/AIDS disease prediction.