A soft computing approach for the design of novel pharmaceuticals

Abstract

The pace of technological advancement in today's society has generated an enormous demand for methods facilitating the intelligent design of new pharmaceuticals, chemical compounds, and materials. The authors have developed computationally intelligent data mining and molecular modeling technologies for the automated design and understanding of complex molecular structures. The Transferable Atom Equivalent methodology for calculating derived molecular properties generates a large set of potential predictors for a set of molecules. Novel neural network based "data strip mining" techniques extract predictive models from this set. These models may be used to screen candidate pharmaceuticals prior to expensive and time consuming laboratory testing.