Scholarlink

<p>The prompt and accurate identification of pathogenic bacteria is crucial for mitigating the transmission of infections. Conventional detection methods face limitations, including lengthy processing, complex sample pretreatment, high instrumentation costs, and insufficient sensitivity for rapid on-site screening. To address these challenges, an aptamer (Apt)-sensor based on functionalized magnetic nanoparticles (MNPs) was developed for detecting <i>Escherichia coli</i>. Fe<sub>3</sub>O<sub>4</sub>@Au nanoparticles were synthesized by stepwise modification, followed by Apt conjugation via Au–S bonds to form Fe<sub>3</sub>O<sub>4</sub>@Au@Apt. Subsequently, efficient capture and separation of target bacteria was achieved by combining the specific Apt-<i>E. coli</i> recognition sites with magnetic solid-phase extraction. A time–frequency domain feature-assisted XGBoost model was constructed for the sensor to achieve accurate prediction of bacterial concentration. Six equivalent-circuit frequency domain and six time domain characteristic parameters were extracted from the equivalent circuit model (ECM) and the distribution of relaxation times (DRT), respectively, and Bayesian optimization (BO) was subsequently adopted for automatic hyperparameter search to reduce prediction errors. Furthermore, the SHapley Additive exPlanations (SHAP) analysis demonstrated the necessity of time–frequency feature fusion for enhancing prediction accuracy. The experimental results indicated that the Fe<sub>3</sub>O<sub>4</sub>@Au@Apt-modified magnetic glassy carbon electrode (MGCE) can achieve quantitative detection of <i>E. coli</i> in a concentration range of 10<sup>0</sup>–10<sup>7</sup>&#xa0;CFU/mL, with a detection Limit down to 1&#xa0;CFU/mL. In addition, the intelligent detection framework based on BO-XGBoost exhibited excellent predictive performance, with an <i>R</i><sup>2</sup> value of 0.990, mean absolute error (MAE) of 0.087&#xa0;CFU/mL, and root mean square error (RMSE) of 0.158&#xa0;CFU/mL. This approach shows significant potential for future <i>E. coli</i> monitoring applications in food safety and environmental surveillance.</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

Ultrasensitive and specific electrochemical detection of Escherichia coli via functionalized magnetic nanoparticles by time–frequency analysis

  • 期刊论文

  • 发表日期：

    2025年9月11日

查看全文
  • Siqi Dong,
  • Xiaobin Zhang,
  • Shijuan Cao,
  • Chenpan Lei,
  • Hanyang Bao,
  • Ying Xu

摘要

The prompt and accurate identification of pathogenic bacteria is crucial for mitigating the transmission of infections. Conventional detection methods face limitations, including lengthy processing, complex sample pretreatment, high instrumentation costs, and insufficient sensitivity for rapid on-site screening. To address these challenges, an aptamer (Apt)-sensor based on functionalized magnetic nanoparticles (MNPs) was developed for detecting Escherichia coli. Fe3O4@Au nanoparticles were synthesized by stepwise modification, followed by Apt conjugation via Au–S bonds to form Fe3O4@Au@Apt. Subsequently, efficient capture and separation of target bacteria was achieved by combining the specific Apt-E. coli recognition sites with magnetic solid-phase extraction. A time–frequency domain feature-assisted XGBoost model was constructed for the sensor to achieve accurate prediction of bacterial concentration. Six equivalent-circuit frequency domain and six time domain characteristic parameters were extracted from the equivalent circuit model (ECM) and the distribution of relaxation times (DRT), respectively, and Bayesian optimization (BO) was subsequently adopted for automatic hyperparameter search to reduce prediction errors. Furthermore, the SHapley Additive exPlanations (SHAP) analysis demonstrated the necessity of time–frequency feature fusion for enhancing prediction accuracy. The experimental results indicated that the Fe3O4@Au@Apt-modified magnetic glassy carbon electrode (MGCE) can achieve quantitative detection of E. coli in a concentration range of 100–107 CFU/mL, with a detection Limit down to 1 CFU/mL. In addition, the intelligent detection framework based on BO-XGBoost exhibited excellent predictive performance, with an R2 value of 0.990, mean absolute error (MAE) of 0.087 CFU/mL, and root mean square error (RMSE) of 0.158 CFU/mL. This approach shows significant potential for future E. coli monitoring applications in food safety and environmental surveillance.