AI for prognosis among people living with HIV (PLHIV): a systematic review and meta-analysis protocol (Preprint)

BACKGROUND Advanced HIV Disease (AHD), caused by the human immunodeficiency virus (HIV), remains a significant global health concern, with nearly 40.8 million people living with HIV (PLHIV) as of 2024. Antiretroviral therapy (ART) has improved outcomes, yet its success depends on timely interventions, adherence, and retention in care. Artificial intelligence (AI), including machine learning and deep learning algorithms, offers promising tools for prognostic modeling in HIV care, supporting clinical decision-making and personalized treatment. Recent evidence has reported a wide range of AI applications across the HIV care continuum. However, existing syntheses have largely adopted broad narrative approaches and have not focused specifically on AI-based prognostic prediction models or systematically evaluated their performance, risk of bias, reporting quality, and clinical readiness. Consequently, a critical gap remains in understanding the validity, robustness, and applicability of AI-driven prognostic models for PLHIV. OBJECTIVE The present study aims to conduct a systematic review and meta-analysis of AI-based prognostic models predicting treatment and disease outcomes among PLHIV, with a focused assessment of predictive performance, methodological rigor, reporting transparency, and potential for clinical implementation. METHODS A comprehensive literature search will be performed across six databases – PubMed, Embase, Scopus, Web of Science, IEEE Xplore, and ACM Digital Library, covering studies published from January 2015 to December 2025. Eligible studies included original research using AI to predict individual-level outcomes among PLHIV. The study is registered with PROSPERO (Registration number: CRD420251034551) and follows TRIPOD-SRMA guidelines. Data extraction will follow a standardized form based on the CHARMS Checklist and extended with elements from PROBAST, TRIPOD-AI, DECIDE-AI, and the NeurIPS Paper Checklists. Risk of bias, reporting transparency, implementation, and reproducibility will be assessed using these tools. When feasible, prognostic accuracy metrics (e.g., AUC, sensitivity, specificity) will be synthesized using random-effects meta-analytic models, including bivariate analysis and hierarchical summary receiver operating characteristic (HSROC) curves. Heterogeneity will be assessed and explored through subgroup analysis and meta-regression. The strength of evidence will be graded using an adapted GRADE framework that incorporated AI-specific quality dimensions. RESULTS Not applicable CONCLUSIONS This review will provide a focused and methodologically rigorous synthesis of AI-based prognostic models in HIV care, identifying models with robust predictive performance and highlighting critical gaps in validation, reporting, and clinical readiness to inform best practices for future development and implementation.