In early 2020, the team led by Wang Chao at Huazhong University of Science and Technology collected clinical lung sound data from COVID-19 patients at Wuhan Tongji Hospital. Building on this data, they developed the world's first dedicated AI processor chip for lung sound monitoring. The team proposed a two-stage hybrid neural network: it first performs coarse screening of normal lung sounds to skip redundant computation, and then conducts fine-grained diagnosis only on abnormal signals. They also designed a reconfigurable processing element array, a dual-mode systolic dataflow technique, and a multi-branch hybrid-driven memory architecture, significantly reducing power consumption. Fabricated in a 55-nanometer process, the chip has a core area of only 4.39 square millimeters and delivers excellent energy efficiency, making it capable of supporting long-term wearable lung sound monitoring and aiding early detection and intervention of lung diseases. Its low-power architecture can also be extended to monitoring scenarios for physiological signals such as heart sounds, ECG, and EEG.