Scholarlink

<p>Opisthorchiasis, a major foodborne parasitic zoonotic disease in Thailand and neighboring countries, is caused by the carcinogenic liver fluke <i>Opisthorchis viverrini</i> (OV). Accurate classification of OV infection is critical for timely intervention and public health management. In this study, we propose a reliable machine learning (ML) classification model based on peak current data from an electrochemical immunosensor and additional patient condition features, which can facilitate intuitive decision-making without the need for expert personnel. This is the first report to classify OV infection using a ML algorithm integrated with electrochemical biosensor data. A total of 531 urine samples from both OV-positive and OV-negative individuals in endemic areas were analyzed using the immunosensor. We evaluated the effectiveness of six different ML models through cross-validation. Among these models, the decision tree and AdaBoost classifiers demonstrated outstanding performance, each achieving the highest accuracy of 90.65% (95% CI 0.89–0.91). The decision tree model yielded an F1 score of 91%, sensitivity of 95%, and specificity of 83%, while the AdaBoost model achieved an F1 score of 90%, sensitivity of 94%, and a higher specificity of 86%. The neural network model also performed excellently, with an accuracy of 89.72% (95% CI 0.84–0.93), and an F1 score of 89%. The statistical comparison of the model’s performance highlighted the significant difference between the top-performing models and the rest. These results underscore the significance of incorporating sensor data and ML to accurately classify OV infections and enable early diagnosis and intervention. By using this ML model, the status of OV infection can be detected by interpreting sophisticated raw electrochemical data. This implies that patients or medical staff with no prior experience with electrochemical sensors can nevertheless comprehend the disease condition with confidence. The proposed ML models hold promise for enhancing disease surveillance and control strategies in endemic regions and could thus assist medical professionals in the decision-making process and in addressing the burden of opisthorchiasis infection.</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

Machine learning approach using electrochemical immunosensor data for precise classification of Opisthorchis viverrini infection

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月3日

查看全文
  • Nang Noon Shean Aye,
  • Sakda Daduang,
  • Patcharaporn Tippayawat,
  • Anchalee Techasen,
  • Pornsuda Maraming,
  • Paiboon Sithithaworn,
  • Rungrueang Phatthanakun,
  • Jureerut Daduang

摘要

Opisthorchiasis, a major foodborne parasitic zoonotic disease in Thailand and neighboring countries, is caused by the carcinogenic liver fluke Opisthorchis viverrini (OV). Accurate classification of OV infection is critical for timely intervention and public health management. In this study, we propose a reliable machine learning (ML) classification model based on peak current data from an electrochemical immunosensor and additional patient condition features, which can facilitate intuitive decision-making without the need for expert personnel. This is the first report to classify OV infection using a ML algorithm integrated with electrochemical biosensor data. A total of 531 urine samples from both OV-positive and OV-negative individuals in endemic areas were analyzed using the immunosensor. We evaluated the effectiveness of six different ML models through cross-validation. Among these models, the decision tree and AdaBoost classifiers demonstrated outstanding performance, each achieving the highest accuracy of 90.65% (95% CI 0.89–0.91). The decision tree model yielded an F1 score of 91%, sensitivity of 95%, and specificity of 83%, while the AdaBoost model achieved an F1 score of 90%, sensitivity of 94%, and a higher specificity of 86%. The neural network model also performed excellently, with an accuracy of 89.72% (95% CI 0.84–0.93), and an F1 score of 89%. The statistical comparison of the model’s performance highlighted the significant difference between the top-performing models and the rest. These results underscore the significance of incorporating sensor data and ML to accurately classify OV infections and enable early diagnosis and intervention. By using this ML model, the status of OV infection can be detected by interpreting sophisticated raw electrochemical data. This implies that patients or medical staff with no prior experience with electrochemical sensors can nevertheless comprehend the disease condition with confidence. The proposed ML models hold promise for enhancing disease surveillance and control strategies in endemic regions and could thus assist medical professionals in the decision-making process and in addressing the burden of opisthorchiasis infection.