Novel computational approach integrating genetic algorithms with multi-objective optimization to develop efficient protocols for analysing Cd, Pb, Cr and Hg in Saccharum officinarum

Abstract

Background: Heavy-metal contamination in sugarcane (Saccharum officinarum) can compromise food safety and requires extraction protocols that are both efficient and analytically defensible across chemically heterogeneous matrices (juice and lignocellulosic tissues). Methods: A computational–experimental framework was developed that integrates sequential chemical extraction with ICP-OES quantification and a seven-objective NSGA-II optimizer to balance chemical performance (extraction efficiency, contamination resistance, and a phase-integrity/species-preservation proxy) against operational burden (time, reagent use, energy proxy, and cost). Experimental validation was conducted using n =5 independent biological replicates per tissue type per condition, with replicate instrumental measurements and QA/QC gating (certified reference materials plus sugarcane matrix spikes) to ensure analytical accuracy and identify/retain only qualified batches. Agreement between computational predictions and experimental outcomes was evaluated using Bland–Altman limits-of-agreement and uncertainty propagation, while protocol performance differences were tested using Kruskal–Wallis omnibus tests followed by Benjamini–Hochberg–adjusted pairwise Mann–Whitney tests, with 95% bootstrap confidence intervals reported for key metrics (see SI Tables S4A–S4D; SI Fig. S11). Results: Operational objectives exhibited rapid convergence, with 20–30% improvement by generation 15 relative to the initial population, followed by diminishing returns (SI Fig. S6–S9). Statistically significant differences among candidate protocols were observed for the primary objectives (omnibus tests p < 0.05; BH-adjusted pairwise comparisons reported in SI), and prediction–measurement agreement diagnostics supported the validity of experimentally qualified recommendations within the tested design space. Significance: The framework provides a reproducible, uncertainty-aware pathway for optimizing sequential extraction protocols in complex sugarcane matrices while explicitly reporting replication, statistical support, and prediction–measurement agreement needed for defensible method development.