Machine learning surface plasmon resonance biosensor for simultaneous detection of glucose and Plasmodium falciparum biomarkers.

In low-resource clinical settings, Plasmodium falciparum infection and hypoglycemia frequently co-present yet require discrete diagnostic platforms, creating a practical barrier to timely patient management. This study describes a theoretically designed multilayer surface plasmon resonance (SPR) biosensor for label-free, simultaneous quantification of blood glucose and Plasmodium falciparum histidine-rich protein II (HRP-II) from a single analyte sample.The sensor stack comprises Ag/LiB3O5/BiB3O6/Cu/FASnI3/black phosphorus layers deposited on a BK7 glass prism in the Kretschmann attenuated total reflection configuration, operating at 633 nm. Each layer is selected for a specific optical function: phase-matching at the Ag/LiB3O5 interface, evanescent field confinement within the BiB3O6 and Cu layers, and plasmonic loss reduction via the FASnI3/black phosphorus termination. The combined effect yields a 1.82-fold sensitivity gain over a reference single-layer silver SPR sensor.Layer thicknesses and refractive index conditions were optimized using the Transfer Matrix Method. A polynomial regression model trained on 18,000 simulated data points was then used to map structural parameters to sensor response, achieving R2 = 0.997 for glucose and R2 = 0.994 for HRP-II, with computational cost low enough for embedded point-of-care hardware.The optimized configuration produces angular sensitivities of 221.3 deg/RIU for glucose and 212.7 deg/RIU for HRP-II, with detection limits of 0.15 mg/dL and 0.8 ng/mL respectively. Simulated fabrication tolerance analysis gives a performance variation of 3.30% across layer thickness deviations consistent with standard thin-film deposition processes. Per-chip material cost is estimated at USD 1.50 to 4.00.