Multiple machine learning algorithms assisted QSPR models for aqueous solubility: Comprehensive assessment with CRITIC-TOPSIS

Machine learning-assisted data-driven optimization and understanding of the multiple stage process for extraction of polysaccharides and secondary metabolites from natural products

“…Herein, CRITIC-TOPSIS was introduced to determine the weights of the two-stage extraction process, and find the unique optimal solution from numerous Pareto solutions from NSGA-II. 56,57 The medium optional solutions, which refer to the ones with medium parameter values, of the two extraction processes from NSGA-II were selected to be elements to select for the best extraction process. Random assortment was adopted to form the pairs to participate in the comparison.…”

On the Development of Descriptor-Based Machine Learning Models for Thermodynamic Properties: Part 1—From Data Collection to Model Construction: Understanding of the Methods and Their Effects

“…Quantitative Structure Property/Activity Relationship (QSPR/QSAR) models have been widely employed for several decades in chemistry-related fields to predict various endpoints of molecules (i.e., physico-chemical properties and biological activities, respectively) on the basis of their structure (e.g., descriptors, fingerprints, graphs), via mathematical methods. Successful QSPR/QSAR applications include very different endpoints such as critical temperature and pressure [1], normal boiling point [2], heat capacity [3], enthalpy of solvation [4]/vaporization [5,6], blood-brain barrier permeability [7], physico-chemical properties of polymers/fuels/ionic liquids [8][9][10][11][12][13][14][15], solubility [16][17][18][19][20][21], minimum ignition energy of combustible dusts [22] or antibacterial/antiviral properties [23,24].…”