Multi criteria decision analysis based ranking and regression driven prediction using graph topological indices for Urinary tract infections therapeutics.

Chemical graph theory has provided a rigorous framework to represent molecular structure with graphs and hence, topological indices could be derived in a systematic way and bear enough predictive power. In fact, these indices are critical in the studies of Quantitative Structure Property Relationship (QSPR), as they relate molecular structure to measurable physicochemical properties. Present research focuses on pharmaceuticals for Urinary Tract Infections (UTIs) and aims to outline the power of topological indices in modelling the physicochemical properties of the commonly used 10 UTIs drugs. Topological indices are first calculated, and then quadratic and cubic polynomial regression models were formulated and compared. Models are evaluated based on statistical criteria to decide on the best-fitted predictive model. Since there is a limited number of observations, Leave-One-Out Cross-Validation (LOOCV) was conducted to confirm the reliability/robustness of the model. Additionally, drugs have been ranked using Simple Additive Weighting (SAW) and Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) approaches based on combined values of their structural and property descriptors. The results show that the quadratic regression was validated as the optimal model and consistent SAW and TOPSIS rankings. This paper also showcases the relevance of topological descriptors in QSPR based drug analysis.