Harnessing artificial intelligence for dengue forecasting in climate-vulnerable regions: A narrative review with insights from Ba Ria-Vung Tau, Vietnam.

Dengue poses a mounting public-health threat in climate-sensitive, rapidly urbanizing settings. We conducted a narrative review with systematic elements to synthesize advances in artificial intelligence (AI) and machine learning (ML) for dengue forecasting and to appraise their operational relevance to Ba Ria-Vung Tau (BRVT), Vietnam. Structured searches of PubMed, Scopus, and Web of Science (2010 to 15 Dec 2024, English; peer-reviewed only) identified 20 eligible studies. For transparency, we report PRISMA 2020 counts; no meta-analysis was performed because of heterogeneity in populations, outcomes, and metrics. Across the evidence base, models typically used climate and environmental covariates (temperature, rainfall, humidity; earth-observation indices) with lagged effects. Deep-learning architectures-particularly long short-term memory (LSTM) networks and convolutional neural networks (CNNs)-and hybrid pipelines often outperformed classical statistical baselines on held-out tests, though performance reporting (e.g., root mean squared error (RMSE), mean absolute error (MAE), area under the receiver operating characteristic curve (AUC)) was inconsistent. External validation remained uncommon, uncertainty calibration was variably reported, and code/data availability was limited, constraining generalisability. Using BRVT as a worked example, we outline a practical pathway that prioritises aggregate, non-identifiable data feeds; temporal hold-out (and, where feasible, adjacent-province external validation); and lightweight, real-time forecasting components compatible with existing surveillance workflows. We recommend future studies (i) adopt stronger validation and reporting standards, (ii) quantify uncertainty for risk-based decisions, and (iii) couple technical innovation with feasibility considerations (cost, governance, infrastructure, workforce). This review offers a climate-aware roadmap for translating AI/ML forecasting into actionable dengue early warning in BRVT and similar at-risk provinces.