Predictive modelling and ranking of physicochemicals in Aegle Marmelos through topological indices and multi-criteria decision-making techniques

JAGANATHAN B^*Aditya Narayan Bhandari

• Vellore Institute of Technology (VIT), Chennai, India

This study presents an integrated computational framework to investigate the quantitative structure–property relationship (QSPR) of fifteen major bio active phytochemicals from Aegle marmelos (Bael). A comprehensive set of sixteen discrete Adriatic topological indices (TIs) was computed to model six key physicochemical properties relevant to drug-likeness: Molecular Weight (MW), Topological Polar Surface Area (TPSA), Hydrogen Bond Donor and Acceptor Counts, Melting Point, and Rotatable Bond Count. Pairwise multiple linear regression (MLR) models were developed for each property, retaining only those with R2 ≥0.99 to ensure statistical robustness. The frequency of TI appearances in these high-performing models was used to derive objective weights for three multi-criteria decision-Making (MCDM) methods—VIKOR, SAW, and TOPSIS—thereby linking regression analysis with compound ranking. Among all descriptors, the Inverse Sum Indeg Index (ISI) emerged as the most influential. Model comparison revealed that MW had the highest predictability, with the top-performing model (MLSI, ISI) achieving R2 = 0.999. The three MCDM methods demonstrated strong concordance, consistently ranking Riboflavin, Ellagic Acid, and Kaempferol as the top bio actives. The proposed methodology uniquely combines QSPR modelling with MCDM ranking through frequency-based weighting and Bhandari et al. Predictive modelling in Aegle Marmelos demonstrates clear potential for virtual screening, compound prioritization, and early-stage drug discovery using plant-derived molecules.