AI integration into wavelength-based SPR biosensing: Advancements in spectroscopic analysis and detection

Background: In recent years, employing deep learning methods in the biosensing area has significantly reduced data analysis time and enhanced data interpretation and prediction accuracy. In some SPR fields, research teams have further enhanced detection capabilities using deep learning techniques. However, the application of deep learning to spectroscopic surface plasmon resonance (SPR) biosensors has not been reported. This study addresses the integration of AI methods to improve the signal-to-noise ratio (SNR) and detection accuracy of wavelength-based portable SPR biosensors. Results: We designed a deep neural network integrated with the spectral subtraction method to extract SPR responses from the proposed portable SPR biosensor. Using difference spectra as the model input, our AI model provided superior noise reduction and enhanced detection capabilities, outperforming traditional spectral feature extraction methods like dip or centroid positioning. Our study achieved a significantly amplified SNR and improved detection resolution to an impressive 10⁻⁷ RIU level. In addition, we employ Shapley Additive Explanations (SHAP) analysis to determine which parts of the input the AI model considers most important when extracting SPR response, thereby increasing the interpretability and transparency of the AI model. The results indicate that the wavelength regions considered most important by our proposed AI model are very close to the full width at half maximum (FWHM) range. This region is also recognized by traditional theory as having a significant impact on the sensitivity of SPR sensing. Significance: Integrating AI into wavelength-based portable SPR biosensing represents a significant advancement in on-site detection technologies, driving potential applications across various monitoring scenarios. Our findings highlight the AI model's effectiveness in reducing noise and enhancing detection accuracy, particularly in measurements involving low-concentration analytes. This innovation holds great promise for fields that demand real-time, high-precision, on-site detection, such as biomedical diagnostics, environmental monitoring, and biochemical analysis, setting the stage for transformative shifts in these critical areas.