IMPLEMENTATION OF KALMAN FILTER, RECURRENT NEURAL NETWORK, AND DECISION TREE METHOD TO FORECAST HIV CASES IN EAST JAVA

Abstract

HIV (Human Immunodeficiency Virus) is a virus that infects cells in the body and weakens the human immune system, making it more susceptible to various diseases. Meanwhile, the symptoms of the disease arising from HIV itself are referred to as AIDS (Acquired Immune Deficiency Syndrome). Approximately 50% of people with AIDS in Indonesia are adolescents. Until now, HIV/AIDS has ranked second in East Java province. HIV/AIDS is classified as a dangerous disease because of the risk of death. Unfortunately, there is no treatment method or vaccine that could prevent this disease. This monitoring program to prevent the development of dangerous health cases such as HIV/AIDS is very helpful for local governments. Along with the development of information technology, the emergence rate of new HIV/AIDS cases can now be forecasted using machine learning as a monitoring tool to support. This machine learning-based monitoring program works with past data for statistical analysis. In this study, the methods used are Kalman Filter, Recurrent Neural Network, and Decision Tree. The Kalman Filter is a type of filter method that is used to predict the state of a dynamic, stochastic, linear, discrete system. A Recurrent Neural Network (RNN) is a development of a Neural Network. RNN deals with input sequence/time-series data by individual sector at each step and preserves the information it has captured at previous time steps in a hidden state. A Decision Tree is one of the classic tree-based prediction methods. The best error value (RMSE) achieved by each method is 0.0885 for the Kalman Filter, then for the Recurrent Neural Network method achieved 0.198, and the Decision Tree method successfully achieved 0.0287.