Detection of Biolayer Formation Instant in SPR Biosensors Using Neural Networks

This work presents a method for detecting the biolayer formation instant in surface plasmon resonance (SPR) biosensors using signal processing and artificial intelligence. The proposed approach combines the wavelet transform (WT) with a multi-layer perceptron (MLP) neural network to analyze the asymmetry in the SPR response. Experimental data were collected from BSA adsorption cycles using an SPR biosensor and the VIR biochip. After preprocessing and feature extraction using the Daubechies db4 wavelet, the MLP was trained to recognize the characteristic pattern associated with monolayer formation. The model achieved over 99% accuracy, with an average temporal error of less than 3 seconds. The technique enables early detection of the biolayer formation, reducing reagent consumption and shortening the experimental duration by up to 474 seconds per cycle. These results demonstrate the potential of the method for automating functionalization steps in SPR-based biosensors, especially in cost-sensitive applications.