In the digital age, social media platforms, especially Twitter, have become important tools for disaster management by providing real-time information and facilitating communication in the event of an accident. This comprehensive review focuses on the use of machine learning and deep learning models to analyze Twitter data to obtain better results. Different machine learning models and deep learning techniques have been examined in the study, and their applications are highlighted in classifying disastrous tweets to improve the awareness in critical situations. Major issues related to data preprocessing and feature extraction are discussed, emphasizing the importance of developing robust models that can reduce the complexity of unstructured data. This study highlights the importance of machine learning and deep learning to improve the strategies related to crisis management and gives insights for the future scope in this area of study.