A classification nanostructure-activity relationship (nano-SAR) was developed and validated based on a biological profile that depicts the activities of 51 nanoparticles (primary size 20-74 nm) with 64 different metrics (4 cell types � 4 assays � 4 concentrations). The 64-dimensional activity profile was statistically analyzed, which provided a biologically meaningful class definition that labels nanoparticles as biologically active versus inactive for nano-SAR development. Four nano-SAR descriptors were evaluated and the best performing model was based on zeta potential and spin-lattice relaxativity (R1). This model demonstrated good classification accuracy of 79.1% in 5-fold cross-validation and bootstrapping validation while also demonstrating robustness in Y-randomization. In addition, the posterior class probabilities, which were obtained in the nano-SAR development process, were utilized to assess the impact of different prediction acceptance levels of false negatives relative to false positives on regulatory decisions that could be based on nano-SAR predictions of nanoparticles toxicity.

Nano-SAR development with application to regulatory decision making

244th National Meeting (2012)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.