Scholarlink

<p>A hybrid plasmonic-GNN biosensing system was designed to allow rapid, precise and reliable quantification of glucose levels with a nano antenna optical biosensor. The paper is aimed at overcoming shortcomings of conventional plasmonic sensors, including low-concentration sensitivity, variability due to fabrication and spectral noise. It aimed to create a very sensitive nanoantenna system and design the best functional layers to bind the glucose molecules and couple the optical output with a powerful graph neural network that can derive the complex spectral-spatial patterns. The methodology combines several types of optimized gold nanoantennas (FDTD), 65&#xa0;nm boronic-acid functional layer (hydrogel), CCD-based spectral interrogation, and preprocessing to advanced processing methods before GNN regression. The data showed a high correspondence between the simulated and experimental resonance performance that resonance at 780&#xa0;nm (1.17% error), Q- factor of 41.3, and field enhancement of 27.9 times. The response of functionalized sensors was predictable and reversible with a 65&#xa0;nm coating shift of 7.4&#xa0;nm and a repeatability of 97.9% of the cycle. The GNN had high predictive accuracy, MAE of 0.12 mM, RMSE of 0.17 mM, and R<sup>2</sup> of 0.994. The cross-validation GOx assays exhibited R<sup>2</sup> = 0.992, which indicates clinical grade agreement and proves the platform to be well adapted to both continuous and point-of-care glucose monitoring.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Edge-Plasmonic Nanoantenna Network for High-Throughput Glucose Profiling via Graph Neural Models

  • 期刊论文

  • 发表日期：

    2026年3月20日

查看全文
  • D. Poornachandra Reddy,
  • C. Arvind,
  • Srihari K.

摘要

A hybrid plasmonic-GNN biosensing system was designed to allow rapid, precise and reliable quantification of glucose levels with a nano antenna optical biosensor. The paper is aimed at overcoming shortcomings of conventional plasmonic sensors, including low-concentration sensitivity, variability due to fabrication and spectral noise. It aimed to create a very sensitive nanoantenna system and design the best functional layers to bind the glucose molecules and couple the optical output with a powerful graph neural network that can derive the complex spectral-spatial patterns. The methodology combines several types of optimized gold nanoantennas (FDTD), 65 nm boronic-acid functional layer (hydrogel), CCD-based spectral interrogation, and preprocessing to advanced processing methods before GNN regression. The data showed a high correspondence between the simulated and experimental resonance performance that resonance at 780 nm (1.17% error), Q- factor of 41.3, and field enhancement of 27.9 times. The response of functionalized sensors was predictable and reversible with a 65 nm coating shift of 7.4 nm and a repeatability of 97.9% of the cycle. The GNN had high predictive accuracy, MAE of 0.12 mM, RMSE of 0.17 mM, and R2 of 0.994. The cross-validation GOx assays exhibited R2 = 0.992, which indicates clinical grade agreement and proves the platform to be well adapted to both continuous and point-of-care glucose monitoring.