전기화학적 바이오센싱, 다항목 수질 모니터링 및 기계학습 기반 Microcystin-LR 의 신속 현장 검출 = On-Site Detection of Microcystin-LR Using Integrated Electrochemical Biosensing and Machine Learning based Multi-Parameter Water Quality Monitoring

다국어 초록 (Multilingual Abstract)

Microcystin-LR （MC-LR） is a potent hepatotoxic cyanotoxin produced during harmful algal blooms and is recognized as a global environmental concern because of its widespread occurrence in surface waters. Screen-printed carbon electrode （SPCE）-based biosensors have emerged as a rapid and cost-effective approach for MC-LR detection. However, their electrochemical responses are strongly affected by water quality conditions, including pH, electrical conductivity （EC）, and total dissolved solids （TDS）, often requiring frequent recalibration tailored to individual water matrices. To address this limitation, this study developed an integrated framework that combines SPCE-based antibody biosensing, multi-parameter water quality monitoring, and machine learning （ML） for the rapid on-site detection of MC-LR.
Field samples collected across Florida generated more than 200 observations integrating biosensor impedance responses with concurrent measurements of pH, EC, turbidity, and UV254. Among the evaluated ML models, Extreme Gradient Boosting （XGBoost） showed the highest predictive performance, with a Nash–Sutcliffe efficiency （NSE） of 0.89 and a root mean square error （RMSE） of 13.21, enabling the development of a unified model applicable across diverse water matrices. Shapley Additive Explanations （SHAP） analysis identified impedance （Z′） as the most influential predictive feature, followed by EC, pH, UV254, and turbidity, highlighting the importance of matrix-driven interfacial effects in determining biosensor response. Overall, this study establishes a robust data-driven framework for characterizing interactions between biosensor output and water matrix conditions and provides a practical strategy for faster and more accurate on-site measurement of MC-LR in complex environmental waters.

목차 (Table of Contents)

• Ⅰ. 서 론 1
• Ⅱ. 연 구 사 7
• Ⅲ. 재료 및 방법 17
• 1. 시료 채취 및 수질항목 분석 17
• 1.1. 시료 채취 및 조사지점 특성 17
• 1.2. 수질항목 측정 방법 19
• 2. 전기화학 바이오센서 제작 방법 21
• 2.1. MC-LR 전기화학적 바이오센서 제작 21
• 2.2. 시약 및 재료 25
• 2.3. MC-LR 바이오센서의 전기화학적 특성 평가 27
• 2.4. MC-LR 바이오센서 및 수질인자 데이터셋 구축 28
• 3. 머신러닝 모델링 및 MC-LR 예측 30
• 3.1. 모델 개발을 위한 입력변수 30
• 3.2. 예측모델 구축 31
• 3.3. 모델 성능 평가 36
• 3.4. Shapley additive explanations（SHAP） 분석 38
• Ⅳ. 결과 및 고찰 39
• 1. 센서 성능 검증 39
• 1.1. 다항목 수질센서의 보정 및 검증 39
• 1.2. MC-LR 바이오센서 검량선 평가 42
• 2. MC-LR 및 센서 데이터셋 구축 47
• 2.1. 통합 데이터셋 구축 및 특성 47
• 3. 머신러닝 모델 성능 평가 및 해석 50
• 3.1. 머신러닝 모델 성능 비교 50
• 3.2. SHAP 기반 모델 해석 52
• Ⅴ. 결 론 57
• 참 고 문 헌 60
• Abstract 64