BPNN-assisted sensor based on micro-nano fiber coupler for human blood lead detection

In this study, a micro-nano fiber coupler (MFC) sensor modified by chitosan-complexed glutathione (CS@GSH) was demonstrated to detect trace heavy metal ions in the blood. The sensor selectively responded to Pb2+ in an aqueous environment with a detection limit (LOD) of 0.02 μg/dL. Further experiments demonstrated the application of the sensor in the detection of blood lead concentration. The results showed that the LOD of the sensor for lead in deproteinized blood was 0.025 μg/dL, which is lower than the minimum calibrated in the WHO analytical method for determination of blood lead level exposure standard (5 μg/dL). In addition, a machine learning (ML) model based on a back-propagation neural network (BPNN) was proposed to assist the Pb2+ detection of the sensor, which reduced the number of sampling points of the spectra by model training and performed the dual covariate interrogation of the Pb2+ concentration and temperature within the acceptable error range.