Achieving Selective Sensing in Microbial Fuel Cell Biosensors by Deep-Learning-Based Neural Networks for Detection of Tetracycline and Penicillin in Water

Microbial fuel cell (MFC) biosensors have been a hot research topic due to their advantages of low-cost and real-time environmental monitoring capability. However, although it has been studied for more than one decade, there are few studies reporting selectivity in MFC biosensors. For the first time, we report achieving selective sensing in recognizing the type of antibiotics by an MFC biosensor through utilizing deep learning algorithms to analyze the current versus time data after the biosensor is exposed to two types of antibiotics. By employing a deep learning algorithm based on gated recurrent units (GRUs), we achieve a 97.9% classification selectivity for tetracycline (TC) and penicillin (PE). Additionally, the system incorporates wireless data transmission for real-time monitoring over practical distances. This study highlights the potential of combining MFC biosensors with deep learning to achieve selective sensing.