A hybrid QSPR-MCDM framework for ranking organic acids using topological indices

Organic acids are industrially and pharmaceutically important compounds, yet selecting the most suitable acid for a given application is complicated by the need to reconcile multiple, often conflicting physicochemical properties. This study develops and validates a hybrid framework that integrates Quantitative Structure–Property Relationship (QSPR) modeling with Multi-Criteria Decision-Making (MCDM) to systematically rank twenty organic acids. Thirteen degree-based topological indices were computed from the hydrogen-suppressed molecular graphs of the acids and used as structural descriptors to build linear QSPR models for five key physicochemical properties: boiling point, molecular weight, enthalpy of vaporization, flash point, and density. All five models were statistically significant ( \(p < 0.001\) ), with leave-one-out cross-validated \(Q^2\) values ranging from 0.52 to 0.77, confirming genuine predictive capability. The QSPR-estimated properties then constituted the decision matrix for the MCDM stage. Three MCDM methods (TOPSIS, VIKOR, and SAW) were combined with four weight-allocation approaches (Point Allocation, Mean Weight, Standard Deviation, and Entropy), yielding twelve distinct ranking scenarios that were subsequently integrated into a single composite ordering via rank-average aggregation. Unlike earlier applications of MCDM in chemical graph theory, where topological indices themselves served as criteria, the present framework employs indices exclusively as structure descriptors within the QSPR stage, ensuring that the final ranking reflects predicted chemical behavior rather than abstract graph-theoretic quantities. The results demonstrate strong robustness at the extremes of the ranking: Ascorbic acid and Gallic acid consistently occupy the top two positions across all twelve scenarios, while Tetraacetic acid is uniformly ranked last. In contrast, mid-tier acids such as Citric acid, Tartaric acid, Phthalic acid, and EDAHA display considerable rank variability, reflecting more balanced property profiles that are sensitive to weighting choices. Spearman rank correlation between individual scenarios and the composite ranking ranged from \(\rho \approx 0.87\) to \(\rho \approx 0.99\) , with differentiated weighting schemes (PA, SD, E) agreeing more closely with the composite than the equal-weight scheme. The proposed QSPR-MCDM framework offers a robust, objective, and adaptable tool for holistic compound assessment and rational selection in chemical, material, and pharmaceutical applications, moving beyond single-property prediction toward informed multi-criteria decision-making.