A dual-matrix analytical framework integrating bio-inspired machine learning and Sobol-optimized UV spectrophotometry for five-analyte quantification: comprehensive sustainability assessment.

The development of environmentally sustainable analytical methodologies capable of resolving highly overlapped multicomponent systems remains an important challenge in modern analytical chemistry. UV spectrophotometry offers several practical advantages, including minimal solvent-consumption, rapid analysis, low operational cost, and instrumental simplicity; however, its quantitative application is often constrained by severe spectral-overlap among analytes. In the present study, a dual-matrix chemometric framework was developed for the simultaneous determination of five spectrally overlapping analytes-hydrochlorothiazide (HCD), losartan potassium (LOS), ramipril (RAP), N-nitrosodiethylamine (NDA), and toluene (TLN)-in pharmaceutical formulations and fortified human plasma samples. The proposed methodology integrates a five-factor, five-level multilevel experimental design for calibration, Sobol quasi-random sampling for external validation, and Firefly Algorithm-optimized Partial Least Squares (FA-PLS) modeling for adaptive wavelength selection and multivariate calibration. The developed models demonstrated excellent analytical performance. For pharmaceutical formulations, mean recoveries ranged from 99.68 to 100.57%, with RMSEP values between 0.037 and 0.094 µg/mL. For fortified human plasma samples, mean recoveries ranged from 98.18 to 98.97%, with RMSEP values between 0.059 and 0.146 µg/mL, demonstrating the feasibility of the proposed approach under biologically relevant matrix conditions. The environmental profile of the proposed methodology was comprehensively evaluated using multiple complementary tools, including NEMI, ComplexGAPI, AGREE, the Multicolor Assessment Tool, carbon footprint estimation, and the Need-Quality-Sustainability index. Collectively, these assessments confirmed the favorable greenness and overall sustainability characteristics of the developed spectrophotometric approach. To the best of our knowledge, no previous UV spectrophotometric chemometric method has been reported for the simultaneous determination of LOS, RAP, and HCD in the presence of both NDA and TLN within a single analytical workflow without a prior separation step. The proposed dual-matrix FA-PLS framework demonstrates that UV spectrophotometry, when coupled with advanced chemometric optimization and statistically designed calibration strategies, can provide an accurate, sensitive, and environmentally sustainable platform for multicomponent pharmaceutical analysis and proof-of-concept application to biological matrices.