AI-Driven Cross-Continental Forecasting of Dust-Induced Influenza Peaks: A Geospatial Deep Learning Framework Linking Tehran with the U.S. Southwest

Core Problem:Seasonal influenza poses a significant global health challenge, particularly in regions impacted by severe dust storms and elevated particulate matter (PM) concentrations. Existing studies rarely integrate advanced artificial intelligence (AI) methods to forecast influenza peaks driven by dust and PM exposure, especially across diverse climatic and urban environments. This knowledge gap limits proactive public health interventions and hospital preparedness.Approach:We developed an AI-driven forecasting framework combining machine learning (ML) and deep learning (DL) techniques to assess influenza peak activity under dust and PM influence. Multi-source datasets—including satellite aerosol indices, ground-based PM measurements, and clinical influenza records—were collected for two distinct regions: Tehran, Iran, and the U.S. Southwest. Random Forest, XGBoost, LSTM, Bi-LSTM, and CNN-LSTM models were trained and validated to capture nonlinear, lagged, and spatiotemporal dependencies. Dust and PM effects were analyzed separately and jointly to determine region-specific drivers.Key Findings:DL models, particularly Bi-LSTM and CNN-LSTM, outperformed traditional ML methods, accurately predicting both timing and magnitude of influenza peaks. Dust storms increased peak probability by 20–35%, while PM2.5. showed the strongest short-term effect within 3–5 days. Comparative analysis revealed that Tehran’s urban PM dominates short-term influenza surges, whereas dust-driven events exert a stronger influence in the U.S. Southwest.Impact:This cross-regional AI framework provides a scalable, evidence-based decision-support tool for public health planning. It enables early detection of influenza peaks, proactive interventions, optimized hospital preparedness, and improved respiratory risk management under worsening climate variability and air-quality extremes. The study demonstrates a novel, globally relevant approach linking AI, environmental exposure, and infectious disease forecasting.