Fig. 8.

Predicting nanoparticle delivery to micrometastases based on physiological characteristics. (A) To generate a predictive model of nanoparticle delivery to micrometastases, a dataset was created via 3D imaging and analysis of physiological characteristics and nanoparticle delivery for individual micrometastases. The dataset was divided with 80% of the data being used to train and cross-validate potential models, and 20% of the data for prediction testing. Separate SVM models were generated for each nanoparticle delivery output. The optimal model generated from the training dataset was exported and run with the test dataset. The actual and the model predicted nanoparticle delivery values from the test dataset are shown for mean nanoparticle intensity (B), density of nanoparticle-positive cells (C), and number of nanoparticle-positive cells (D) per individual micrometastasis. (E) The actual and predicted number of nanoparticle-positive cells in each micrometastasis separated by time point. SA:vol, surface-area-to-volume ratio. NP, nanoparticle.