Predictive modeling of influenza strain drugs using temperature-based topological indices and regression analysis via multi-criteria decision making techniques

Abstract The proposed study is an integrated graph-theoretical and statistical model of predictive modeling and ranking of influenza strain drugs based on temperature-based topological indices. Chemical graphs were used to model drug molecules and regression models that estimated important physicochemical properties were derived. The cubic models had the best predictive ability with coefficients of determination up to $$R^2=0.9791$$ of molar refractivity and polarity and $$R^2=0.9481$$ of molar volume and moderate correlation of boiling and flash points ( $$R^2\approx 0.72$$ ). Moreover, the multi-criteria decision-making methods (WSM and WPM) reported Azithromycin (81.25), Ritonavir (77.46), and Indinavir (72.82) as the best ranked ones. The presented solution will offer a cost-effective, interpretable, and reliable instrument of antiviral drug prioritization.