Abstract: Forest fires are a serious threat to global ecosystems, economies, and human safety. Predicting fire risk and spread with accuracy is therefore essential for effective prevention, mitigation, resource allocation, and emergency response. Following a data-model-key trend sequence, this paper provides a comprehensive review of the data sources, systems, and models commonly used for forest fire risk and spread prediction. In practice, global fire prediction systems are evolving along a path from regional and national to global scales, while also exhibiting trends of deepening multi-source data fusion and strengthening international cooperation. The paper highlights the core role of satellite remote sensing in active fire detection and fuel mapping, and provides an in-depth analysis of the potential and challenges of Artificial Intelligence (AI) and Machine Learning (ML) in integrating heterogeneous data, identifying complex patterns, and enhancing predictive accuracy. The final section of this paper summarizes current challenges at the data, model, and systemic levels and discusses future directions, including multi-source data fusion and Physics-Informed Machine Learning (PIML), aiming to offer a reference framework for the development of efficient and intelligent forest fire risk and spread prediction system in China.