Rapid quantitative urinary chloride sensing with conductivity correction for cardiac patient management

We present a rapid, accurate and miniaturized electrochemical platform for urinary chloride measurement using chronopotentiometry coupled with conductivity measurement on screen-printed electrodes. Chronopotentiometry was employed to measure physiological chloride concentrations (10-200 mM) in urine mimicking electrolyte solutions (UMES) and urine samples obtained from research participants transitioning from high to low salt diets. The current density used in chronopotentiometry was systematically optimized between 60 and 960A/m² using UMES; however, no single current density was able to resolve the entire physiologically relevant chloride range. Therefore, two separate calibration curves were constructed using current densities of 120 A/m² and 600 A/m² for the 10-50 mM and 50-200 mM chloride concentration ranges, respectively. Additionally, our studies revealed that sample conductivity (3-42 mS/cm) introduces significant variability in the chloride measurement. Thus, the effect of conductivity on urinary chloride measurement was addressed using two independent correction strategies: a heuristic statistical approach and a machine learning approach. The heuristic statistical method improves the coefficient of determination (R²) from 0.81 (uncorrected) to 0.95 (corrected). The machine learning approach, based on the XGBoost algorithm, achieved R² values of 0.98 on the training set and 0.93 on the independent test set.