An artificial neural network (ANN) model for predicting HIV infection among adolescents in Adamawa State

Abstract

Global health concern acquired immunodeficiency syndrome (AIDS) is brought on by the retrovirus known as Human Immunodeficiency Virus (HIV), which compromises immunity. The study's objective is to create a model of an artificial neural network (ANN) that can anticipate adolescent HIV infection. The study identified specific features related to HIV infection among adolescents and used the Analytical Hierarchy Process (AHP) to rank the importance of these features to improve the accuracy of prediction. This is because the input features used in the previous studies did not achieve the desired accuracy. Therefore, the study makes use of 14 characteristics that the literature analysis found to be linked to HIV infection in adolescents. Professionals used a systematic Likert scale with six alternatives to determine the infection's severity. The most important factors influencing HIV infection in adolescents were found by using the data collected to rate the infections. In the study, the ANN was modeled using the Keras Sequential Classification model package. A confusion matrix was used to assess the model's performance after the ideal number of hidden layers and neurons for each hidden layer were established. The model is a feed-forward multi-layer perceptron that is optimized via the back propagation algorithm and is implemented using the supervised learning approach. Accordingly, three hidden layers with 128 neurons, 128 neurons, and 64 neurons each using rectified linear unit (ReLU) activation functions are produced by the model. The model's accuracy in predicting adolescent HIV infection was 95%. Consequently, this study examined the model that can accurately predict HIV infection in Yola North LGA, Adamawa State, Nigeria. For practical application, the research suggests enlarging the dataset and raising the number of training epochs to further increase accuracy.