The accurate prediction of patient prognosis is a critical challenge in clinical practice. With the availability of various patient information, physicians can optimize medical care by closely monitoring disease progression and therapy responses. To enable better individualized treatment, dynamic prediction models are required to continuously update survival probability predictions as new information becomes available. This article aims to offer a comprehensive survey of current methods in dynamic survival analysis, encompassing both classical statistical approaches and deep learning techniques. Additionally, it will also discuss the limitations of existing methods and the prospects for future advancements in this field.

翻译：患者预后的精准预测是临床实践中的关键挑战。借助多种患者信息的可用性，医生可通过密切监测疾病进展和治疗反应来优化医疗护理。为实现更优的个体化治疗，需要动态预测模型以在新信息出现时持续更新生存概率预测。本文旨在系统综述动态生存分析的现有方法，涵盖经典统计学方法与深度学习技术。此外，还将讨论现有方法的局限性及该领域未来发展的前景。