Answers are grounded in this paper and cite their sources.
Surface plasmon resonance (SPR) biosensors enable label-free, real-time monitoring of biomolecular interactions with high sensitivity and minimal sample preparation. This review critically examines SPR biosensing studies reported between 2015 and 2026, covering plasmonic materials (noble metals, transition-metal nitrides, two-dimensional materials, hybrid nanocomposites and metasurfaces), fabrication routes, surface functionalisation and antifouling chemistry, optical and computational performance enhancement, and applications in clinical diagnostics, environmental monitoring, food safety and drug discovery. Beyond summarising reported figures, we compare material classes in terms of reproducibility, stability, cost and scalability; reconcile sensitivity, figure-of-merit and detection-limit values reported in different units and configurations; grade representative studies by level of evidence, from numerical simulation to clinical validation; and appraise how machine-learning-assisted SPR has been validated. Machine learning enables automated spectral interpretation, noise reduction, resonance-peak detection and biomarker classification, while data-driven optimisation can improve plasmonic nanostructure design. However, limited dataset sizes, risks of overfitting and insufficient independent validation across instruments and laboratories remain major challenges to the reliable deployment of ML-assisted SPR biosensors. The analysis shows that most reported performance gains remain numerical or buffer-based, and that the biointerface, manufacturability and validation, rather than the optical transducer, now limit translation. Technology-readiness levels and technical barriers are assessed for portable, wearable, physics-informed, terahertz and quantum-plasmonic approaches, and six research priorities are proposed for developing reliable, affordable and commercially viable SPR biosensors.
No takes yet. Share an insight, caveat, or question.
Upadhyaya et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: