Artificial neuro-fuzzy intelligence for dynamic tracking and prediction of HIV prevalence and new infections among individuals aged 15 to 49 years in Uganda

Abstract

Human immunodeficiency virus (HIV) infection remains a major global health burden, and Uganda continues to experience rising new HIV incidence, making prevention and management increasingly challenging. Accurate estimation of HIV prevalence (PHIVP) and tracking new HIV incidence (PHIVN) require robust computational models to effectively inform public health interventions. This study aims to develop a computational tracking model based on a 2-level intelligent framework to dynamically monitor PHIVP and PHIVN in Uganda and to identify influential factors that contribute to HIV trends among individuals aged 15 to 49. Six datasets from 146 Ugandan districts for 2023: sex rate, antiretroviral therapy coverage, awareness, education, personal income, and historical HIV indicators (prevalence and new incidence) were analyzed. A 2-level intelligent technique was employed: level I, an upgraded clustering algorithm was used to identify optimal parameters influencing PHIVP and PHIVN and level II, a neuro-fuzzy model was trained on these parameters to construct the computational tracking model. Data were partitioned into 85% for model training and 15% for testing and validation. Model accuracy was evaluated using regression performance, error metrics, and statistical significance. The model predicted a decline in PHIVP from 5.1% to 4.92%, and accurately tracked 37,996 PHIVN cases out of 38,132 actual incidences. The computational framework demonstrated strong predictive performance, yielding high regression values of R = 0.832 to 0.987 for PHIVP and PHIVN across 3 scenarios. Statistical analysis confirmed model reliability with P < .001 and P = .001. The 2-level intelligent approach exhibits strong potential for dynamically monitoring HIV prevalence and new incidence in Uganda. The 6 analyzed datasets represent significant determinants of PHIVP reduction and PHIVN tracking among adults aged 15 to 49. These findings support the feasibility of computational modeling as a tool for enhancing HIV surveillance, guiding prevention strategies, and strengthening evidence-based public health decision-making in Uganda.