In recent years, the application of artificial intelligence (AI) technology in the field of biology has made rapid progress, with the most notable achievements emerging in protein structure prediction and design, which was awarded the Nobel Prize in Chemistry in 2024. It is foreseeable that the accurate prediction of various physical and chemical properties of proteins will be an important direction for the development of the protein prediction field in the next phase. Protein thermodynamic stability is of great significance in deepening the understanding of life mechanisms, drug development, disease diagnosis and treatment, and in the biotechnology industry for the production of enzyme preparations, the development of biosensors, and the preparation of protein drugs. Accurate prediction of Protein thermodynamic stability by artificial intelligence technology will greatly improve the research capabilities and industrial development efficiency of protein. This article reviews the development of protein thermodynamic stability prediction technology, from biological experimental determination methods, traditional energy function calculation methods to modern machine learning prediction methods. The focus is on the prediction models based on machine learning, especially the breakthroughs of cutting-edge algorithms such as deep neural networks, graph neural networks and attention mechanisms in protein thermodynamic stability prediction. The core challenges of mutation stability prediction, such as the imbalance between the quality and quantity of data sets, model overfitting and modeling of protein dynamics, are discussed in depth.This paper aims to provide researchers with a comprehensive reference framework to facilitate the development of mutation-based Protein thermodynamic stability prediction technology.