With the rapid advancement of cutting-edge technologies such as human-computer interaction, embodied intelligence, medical wearables, and soft robotics, flexible sensing has emerged as a critical medium for interaction between the physical and digital worlds, serving as the core hub that bridges real-world perception and intelligent computation. Traditional rigid sensors struggle to adapt to non-regular carriers like curved surfaces, human skin, and soft mechanisms, while flexible sensors, leveraging their unique advantages of deformability and conformal adhesion, break the morphological constraints of conventional electronic devices and usher in a paradigm shift in sensing technology.

After years of development, significant research achievements have been made in various branches such as material modification, micro-nano structure processing, hardware circuits, and intelligent algorithms. However, the field has long been plagued by a fragmented research paradigm: material science focuses on achieving the ultimate intrinsic sensing performance, structural research concentrates on improving micro-morphology, hardware teams prioritize circuit integration, and algorithm studies often treat sensors as black-box inputs, leaving each segment isolated from the others. This fragmented development model leads to widespread performance trade-offs, making it difficult to replicate the exceptional metrics of lab devices in complete systems. Breakthroughs in a single dimension are often offset by shortcomings at other levels, resulting in a substantial decline in performance when transitioning from device-level impressive parameters to practical systems.

In the face of the practical demands for multi-dimensional, high-reliability, and anti-interference perception in next-generation embodied intelligence and electronic skin, relying solely on single-point innovations in materials, structures, hardware, or algorithms has encountered significant bottlenecks. It is precisely under this academic and practical contradiction that a paradigm shift is urgently needed—breaking free from the conventional mindset of independent iteration within individual subfields and establishing a comprehensive, end-to-end systematic academic framework. This forms the foundational premise of the research presented in this review.

Flexible electronics represent a paradigm shift in modern electronic technology, with flexible sensors as their core components holding immense application potential in fields such as healthcare, robotics, and human-machine interaction. Existing research predominantly focuses on optimizing individual aspects—materials, structures, hardware, or algorithms—leading to inherent performance trade-offs: high sensitivity often comes at the expense of limited detection range, while high flexibility compromises stability. Multimodal integration further diminishes single-sensor performance, hindering the further advancement of flexible sensors.  To address this challenge, teams led by He Peng and Zhang Shuye from Harbin Institute of Technology, alongside Pang Jinbo from Jinan University, proposed a novel "material-structure-hardware-algorithm" holistic collaborative design paradigm. This approach eliminates reliance on breakthroughs in single components, emphasizing bidirectional feedback and mutual compensation among the four key elements to pursue globally optimal system-level performance. The review systematically compiles representative research progress across these four hierarchical levels, analyzes the intrinsic mechanisms of hierarchical collaboration, and constructs a forward-looking theoretical framework, providing a comprehensive roadmap for the development of next-generation flexible sensors.