In complex biological systems, abrupt transitions frequently occur as the system approaches critical states, resulting in shifts between distinct functional regimes. Identifying such critical transitions is therefore crucial for timely intervention and precision prevention. This review systematically synthesizes recent advances in critical point detection methods, with a particular emphasis on the methodological development of dynamic network biomarkers (DNBs) across diverse data settings, including multi-sample, single-sample, and single-cell datasets. We further review emerging approaches that integrate network dynamics with deep learning techniques to improve sensitivity in high-dimensional data. Representative applications in complex disease prediction, cell fate decision identification, and other dynamic biological process analyses are summarized, offering insights into ultra-early detection and mechanistic interpretation of critical transitions in complex biological systems.
Nie et al. (Thu,) studied this question.