AI-Driven Analytical Modeling for Differential Detection and Quantification of Bio-Relevant Alcohols in Electrochemical Diagnostic Systems

The integration of artificial intelligence with electrochemical diagnostic platforms provides an advanced analytical strategy for resolving structurally similar bio-relevant analytes within complex matrices. This study presents an AI-driven modeling framework designed to achieve differential detection and quantitative assessment of alcohol-based biomarkers using electrochemical signal fingerprints as predictive inputs. Experimental datasets, collected under controlled analytical conditions, were subjected to standardized preprocessing and feature-engineering procedures to derive kinetic, impedance-related, and current–potential descriptors. These descriptors were subsequently used to train supervised machine-learning models capable of both binary discrimination and concentration regression across clinically relevant ranges. Model performance was evaluated through stratified cross-validation and independent testing, demonstrating stable discriminatory capability and reliable quantitative predictions. Interpretability analyses revealed the dominant electrochemical features contributing to prediction accuracy, allowing mechanistic insight without revealing proprietary sensing configurations or operational parameters. The proposed framework illustrates a scalable and scientifically grounded pathway for translating electrochemical patterns into clinically meaningful diagnostic outputs. This work highlights the potential of AI-assisted electrochemical systems to support the future development of safe, robust, and compact point-of-care diagnostic kits.

کلیدواژه ها:

Artificial intelligence ، Electrochemical diagnostics ، Machine learning ، Feature engineering ، Bio-relevant alcohols ، Differential detection ، Signal processing ، SHAP interpretability