Artificial intelligence in HIV research: a structured review and task-oriented clinical framework

Le résumé fourni par la source

Background: Human Immunodeficiency Virus (HIV) poses a global health challenge despite the success of the Antiretroviral Treatment (ART), which allows the disease to be potentially controlled. The growing availability of heterogeneous data has impulsed the use of Artificial Intelligence (AI) to address the different clinical domains of HIV, facilitating the creation of decision making tools to assist clinical professionals. Objective: This study aims to propose a task-oriented framework to support clinicians and researchers that want to apply AI in the management of HIV, linking clinical tasks with appropriate AI approaches. Methods: Following a structured review of over 50 relevant studies in this area since 2017 to November 2025, literature was analysed considering four complementary dimensions: (1) the scope of clinical application in the context of HIV, (2) the data types employed, (3) databases and data sources, and (4) AI techniques used, ranging from statistical models to approaches based on neural networks and natural language. Results: Building on this analysis, the proposed framework associates the main clinical tasks of the different clinical domains of HIV with the most appropriate AI approaches, providing recommendations of algorithms and techniques that can be used in different scenarios. Ultimately, this study tackles the use of AI as a key tool for HIV management in different phases of the disease, taking into account the type of available data. Conclusions: AI represents a key tool for enhancing HIV management. This framework provides a structured basis for future research, though it will be necessary to continuously redefine and update this framework as new medical challenges and technical methods arise.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.