AI-DRIVEN OPTIMIZATION OF A SMARTPHONE-BASED COLORIMETRIC METHOD FOR SIMULTANEOUS DETERMINATION OF MULTIPLE HEAVY METAL IONS IN ENVIRONMENTAL WATER OF BABIL GOVERNORATE, IRAQ

Copper, lead, cadmium, mercury, chromium, zinc and nickel reach rivers, irrigation canals and groundwater from industrial effluents, agricultural drainage, urban wastewater and corroding distribution pipes, and several of them are harmful at microgram-per-litre levels. Their routine determination still depends on atomic absorption and inductively coupled plasma instruments, which are expensive, need trained operators and are rarely available near the sampling site. This research plan describes a low-cost method in which a six-reagent chromogenic array (4-(2-pyridylazo)resorcinol, Zincon, micellar dithizone, 1,5-diphenylcarbazide, dimethylglyoxime and xylenol orange) is prepared in a 96-well microplate and photographed with a smartphone inside a closed enclosure with controlled LED lighting. Color values will be extracted from every well in the RGB, HSV, HSL and CIELAB spaces after correction against a reference card. Reaction pH, reagent concentration, reaction time and sample volume will be screened with a Box–Behnken design and refined by Gaussian-process Bayesian optimization. Multiple linear regression, partial least squares, random forest, support vector regression and artificial neural network models will be trained on synthetic multi-metal mixtures to predict the seven concentrations at the same time, and they will be compared by nested cross-validation and an external test set using the root mean square error, mean absolute error and coefficient of determination. The selected procedure will be validated following ICH Q2(R2) and Eurachem guidance for linearity, detection and quantification limits, precision, recovery, selectivity, interference and robustness, and then applied to surface, well and tap water from Babil Governorate, Republic of Iraq, with ICP-MS as the reference method. The expected result is a portable procedure that screens seven metals in one run of about 25 minutes, at a small fraction of the cost of instrumental analysis.