(P08-015-25) A Paper-Based Assay Interfaced With AI for the Quantification of Iron EDTA in Fortified Staple Flours

Objectives: Reliable quantification of micronutrients in fortified foods is critical to addressing micronutrient deficiencies worldwide. This study presents a validated paper-based, colorimetric sensor for cost-effective detection of sodium iron EDTA (NaFeEDTA), an iron form extensively used in fortification, offering a feasible solution for low-resource settings.

Methods: Utilizing a colorimetric method with Ferrozine reagent for ferrous iron detection, the assay employs an origami-inspired design to separate reagents on two types of paper (hydrophobic and hydrophilic), facilitating the dissociation and reaction of iron with Ferrozine. Sensor optimization involved refining reagent integration, sample processing, and physical setup to enhance efficiency, sensitivity, and affordability. A built-in house image analyzer using convolutional neural networks was designed to analyze color development in real time. As a control, color intensity (i.e., green pixel intensity) was captured manually using digital photos and translated into iron concentration using a calibrated algorithm. Validation involved comparing the assay's results with atomic spectroscopy using iron-fortified (20 ppm) maize samples from commercial fortification mills in Tanzania. Assessment criteria included a comparison of accuracy and precision parameters against Tanzania's Fortification Policy (10 mg/kg ≤ pass ≤ 30 mg/kg, otherwise reject) using a confusion matrix.

Results: Figures of interest included the limit of detection, the limit of quantification, and the linear range (1.60, 3.89, 3.89-60 µg Fe/mL, R²=0.99, respectively). Mean iron content (18.4 mg/kg corn) was the same between the two methods. Bias and total error were -1.79± <1.18 and -9.84-8.49 mg Fe/kg corn, respectively. Six-Sigma analysis revealed that the sensor's performance improved from good to excellent when transitioning from manual analysis to computer vision (lower random and systematic errors). The proportion of false positives and the error rate were 6.25 and 14%, respectively.

Conclusions: The paper-based assay offers an affordable, accurate and precise method for measuring NaFeEDTA in fortified flours, demonstrating significant potential for broader application in micronutrient monitoring within small-scale milling operations.

Funding Sources: Funded in part by Sanku-Project Healthy Children