Recent years have brought unprecedented access to diverse solar and heliospheric observations from both ground-based facilities and space missions. Concurrently, advances in numerical modeling, data assimilation, and machine learning are opening new pathways for forecasting solar magnetic activity across a wide range of temporal and spatial scales. This session focuses on emerging approaches that integrate observations, physics-based models, and data-driven methods to improve predictive capabilities. Topics of interest include, but are not limited to: solar cycle prediction, active region emergence/evolution, flare and coronal mass ejection forecasting, and multi-scale-coupling from the solar interior to the heliosphere. Contributions highlighting the synergy between data, theory, and machine learning are especially encouraged, including efforts that combine realistic simulations with observations, develop data assimilation frameworks, or leverage AI/ML for feature detection and prediction. The session aims to showcase recent progress and identify key challenges and opportunities for advancing robust, physics-informed forecasting of solar magnetic activity.