Particulate matter (PM₂.₅ and PM₁₀) pollution in rapidly industrializing regions poses significant environmental and public health risks. This study presents an integrated assessment of spatiotemporal variability, satellite retrieval performance, forecasting, and health risks in the Savar–Gazipur industrial belt, Bangladesh (2019–2024). Satellite-derived analyses reveal persistent exceedances of Bangladesh National Ambient Air Quality Standards (BNAAQS) and WHO-guided values, with strong seasonal dynamics characterized by winter maxima (PM₂.₅ up to 233 μg/m3; PM₁₀ up to 368 μg/m3) and monsoon minima driven by wet deposition processes (r = −0.73 with rainfall). Validation against ground-based measurements (2022–2024) shows that satellite products effectively reproduce seasonal patterns, although exhibiting variable accuracy reflecting uncertainties associated with retrieval limitations in humid, heterogeneous urban-industrial environments. Forecasting using SARIMA, ETS, Holt–Winters, and ANN models (2024–2026) reveals pollutant-specific performance, with ANN best for PM₂.₅ (R2 ≈ 0.76) and Holt–Winters for PM₁₀ (R2 ≈ 0.68). Health risk assessment based on ground-measured concentrations revealed elevated non-carcinogenic and carcinogenic risks, with Hazard Quotient (HQ) values exceeding 2.4 and Excess Lifetime Cancer Risk (ELCR) values for all tested age groups well above the US EPA's acceptable threshold of 1 × 10−4. These findings indicate severe long-term exposure hazards for residents and workers in the industrial belt. This study uniquely integrates satellite remote sensing, hybrid forecasting models, and quantitative health risk assessment, providing a comprehensive framework for air quality evaluation in industrial regions. The findings emphasize the urgent need for mitigation while acknowledging inherent uncertainties in satellite-based PM estimation.