QSPR analysis of physico-chemical and pharmacological properties of medications for Parkinson’s treatment utilizing neighborhood degree-based topological descriptors

Topological indices are invariant quantitative metrics associated with a molecular graph, which characterize the bonding topology of a molecule. The main aim of analyzing topological indices is to summarize and transform chemical structural information, thus creating a mathematical relationship between structures and their physico-chemical properties, biological activities, and other experimental characteristics. Quantitative Structure-Property Relationships (QSPR) and Quantitative Structure-Activity Relationships (QSAR) utilize topological indices to correlate diverse molecular properties, including physico-chemical, thermodynamic, chemical, and biological activities, with their chemical structures. Parkinson's disease is marked by persistent psychosis due to cognitive deficits. Extended compliance with medication and therapy generally reduces symptoms. Factors such as solubility, metabolic stability, toxicity, permeability, and transporter interactions significantly impact the efficacy of drug design and are dependent on the physical and chemical properties involved. Computational tools for the discovery and development of medications for Parkinson's disease have recently gained prominence. Various methods evaluate therapeutic efficacy and adverse effects utilizing machine learning techniques. Further research has utilized computer simulations to explore the molecular mechanisms of the disease and to identify new therapeutic targets. This research investigates the predictive capacity of nine physico-chemical and thirteen pharmacokinetic parameters (ADMET) by utilizing both open and closed neighborhood degree-sum-based descriptors for twelve drugs used in the treatment of Parkinson's disease. The study employs linear, quadratic, cubic, and multiple linear regression models. A comparative analysis is conducted using several well-known degree-based indices alongside the selected open and closed neighborhood degree-sum-based indices within both univariate and multivariate regression methodologies.