Machine learning-based models for the qualitative classification of potassium ferrocyanide using electrochemical...

Area of Science:

• Electrochemistry
• Analytical Chemistry
• Machine Learning

Background:

• Iron is a crucial trace element for the human immune system, particularly against SARS-CoV-2 variants.
• Electrochemical methods offer convenient and sensitive detection of analytes.
• Square wave voltammetry (SQWV) and differential pulse voltammetry (DPV) are effective electrochemical techniques.

Purpose of the Study:

• To improve machine learning models for classifying analyte concentrations directly from voltammograms.
• To quantify ferrous ions (Fe2+) in potassium ferrocyanide (K4[Fe(CN)6]) using SQWV and DPV.
• To validate the electrochemical data using advanced machine learning classification algorithms.

Main Methods:

• Utilized Square Wave Voltammetry (SQWV) and Differential Pulse Voltammetry (DPV) for electrochemical measurements.
• Employed machine learning algorithms including Backpropagation Neural Networks, Gaussian Naive Bayes, Logistic Regression, K-Nearest Neighbors, K-Means clustering, and Random Forest.
• Trained and validated models using datasets derived from measured electrochemical signals.

Main Results:

• Achieved high accuracy in classifying analyte concentrations based solely on voltammogram data.
• Demonstrated a maximum accuracy of 100% for each analyte within a 25-second timeframe.
• Outperformed previously used algorithms in data classification accuracy.

Conclusions:

• Machine learning models can effectively classify analyte concentrations from electrochemical data with high accuracy.
• The developed method provides a rapid and sensitive approach for quantifying ferrous ions.
• This study highlights the potential of integrating electrochemistry with machine learning for advanced analytical applications.