Abstract: Life prediction and health management are core technologies in the reliability engineering field and have been widely used in the past decade. Since it is difficult to obtain the physical failure mechanisms of complex equipment with high reliability, data-driven prediction methods have become mainstream techniques in recent years. In data-driven prediction technology, lack of data, uncertainty of equipment failure processes, and other issues have been identified as important factors that influence prediction accuracy. In view of these problems and the deficiencies of previous studies, a large number of authors have proposed improvement measures. In this paper, under the classifications of failure data-based, degradation data-based, and multi-source data fusion, we review life prediction technology with an emphasis on degradation data-based prediction methods. We list the problems that have been solved and their corresponding improvement measures. In addition, we introduce the three hotspot issues of condition monitoring, maintenance policy, and spare ordering, and summarize new advances in health management techniques based on life prediction results. Finally, we highlight the urgent problems and promising research directions that deserve further research.