Molecular-scale properties of biomaterials relevant to protein adsorption and cell growth using data mining of combinatorial libraries of polymers

Molecular-scale properties of biomaterials relevant to protein adsorption and cell growth were analyzed using machine learning, data mining and molecular modeling. A total of 104 Quantitative Structure Activity Relationship (QSAR) descriptors were generated for each polymer in order to correlate experimental measurement with molecular-scale polymer properties. A sequence of 500,000 computer-based experiments was performed varying the value of fibrinogen adsorption and cell response randomly, but within a normal distribution defined by the experimental standard deviation. The results represent a substantial advance in the development of computational methodologies for the modeling of biological phenomena.