SYNcronized Agentic Prescriptive System for Improving Sequentially HIV Care Continuum (SYNAPSIS-HIV-CC)

While useful, traditional machine learning and epidemiological prediction models for HIV health outcomes are limited in their ability to capture the complexities of HIV care, typically focusing on single outcomes, and not fully considering the nuisances and multiple health challenges faced by people living with HIV (PLWH). Furthermore, while powerful, traditional machine learning models often times fall short in providing human-interpretable explanations for their choices and predictions. To address this shortfall, we propose the SYNchronized Agentic Prescriptive System for Improving Sequentially HIV Care Continuum (SYNAPSIS-HIV-CC), a comprehensive system to optimize the HIV clinical care continuum. Our approach combines artificial intelligence (AI) agents, large language models, and knowledge graphs to provide individualized healthcare recommendations for multiple endpoints, considering multiple crucial HIV health outcomes: Loss of care (patient dropout from care); viral suppression (undetected viral load); CD4 count levels; a key health monitoring panel; development of new acute events; and development of new comorbidities. The approach also includes the development of an antiretroviral regimen optimization system based on causal AI. SYNAPSIS-HIV-CC outputs will be designed to be interpretable by humans, and supported by causal reasoning. Healthcare providers, such as infectious disease specialists, as well as community healthcare experts will be recruited to help design SYNAPSIS-HIV-CC clinical applications. We will obtain our goals via the following Specific Aims: Aim 1: Agentic AI for multimodal prediction of multiple HIV outcomes Aim 2: A multi-endpoint antiretroviral regimen optimization system via causal AI Aim3: Implementation strategy through qualitative expert assessment By leveraging a multimodal dataset including longitudinal electronic health records, clinical notes, social- behavioral determinants of health, and medical imaging, SYNAPSIS-HIV-CC is designed to help improve PLWH care. Taken together, our three aims will produce an agentic, multimodal, causal AI system for both integrated HIV multi-outcome prediction (Aim 1) and treatment optimization (Aim 2), complete with an implementation plan for real-world applications (Aim 3). Project Number: 1R01MH143270-01 | Fiscal Year: 2026 | NIH Institute/Center: National Institute of Mental Health (NIMH) | Principal Investigator: Simone Marini (+1 co-PI) | Institution: UNIVERSITY OF FLORIDA, GAINESVILLE, FL | Award Amount: $1,130,348 | Activity Code: R01 | Study Section: Special Emphasis Panel[ZRG1 NV-J (53)] View on NIH RePORTER: https://reporter.nih.gov/project-details/11308941