上海科技大学知识管理系统

Real-time water pollution monitoring is crucial as global water pollution has become an urgent issue endangering the health of humanity. Microbial taste chips are promising for water pollution monitoring due to the advantages of short response time and real-time monitoring capability. However, although more than 200 journal research articles on microbial taste chips have been reported to date, sensor selectivity, which is the foremost critical parameter, remains an unsolved challenge even after utilizing gene-editing techniques. In addition, the response time is long and takes at least 3 min. Herein, we report a breakthrough to solve the selectivity challenge by a bioinspired wireless microfluidic microbial taste chip with artificial-intelligence(AI)-enabled high selectivity. Utilizing gated recurrent unit(GRU)-based deep learning algorithms, we demonstrate a classification accuracy of 98.9% for Cu2+, Pb2+, and Cr6+ by harnessing the different temporal output current patterns of the chips to different pollutants. A shortest 48-s response time is achieved, 3.75 times shorter than the fastest previously reported counterpart. The chip enables real-time sensing of Cu2+, Pb2+, and Cr6+ with high accuracy and linearity. Combined with a small footprint and wireless connectivity, the chip may find applications in real-time quantitative heavy metal ions in water monitoring and contribute to global efforts in fighting water pollution.