(380w) Machine Learning Determination of Resistant and Tolerant Bacteria Growth Modes Induced By Nanoparticles | AIChE

(380w) Machine Learning Determination of Resistant and Tolerant Bacteria Growth Modes Induced By Nanoparticles


Jones, A. A. - Presenter, Northeastern University
Mi, G., Northeastern University
Bartomeu Garcia, C., Northeastern University
Medina, D., Northeastern University
Kim, N. Y., Northeastern University
Baranda Pellejero, L., Universitat Rovira I Virgili
Bassous, N., Northeastern University
Webster, T. J., Northeastern University
Nanoparticles are promising alternatives to antibiotics since nanoparticles are easy to manufacture, non-toxic, and do not promote resistance. Nanoparticles act via physical disruption of the bacterial membrane and/or the generation of high concentrations of reactive-oxygen species locally. Potential for physical disruption of the bacterial membrane may be quantified by free energy methods, such as the extended Derjuan-Landau-Verwey-Overbeek theory, which predicts the initial surface-material interactions. The generation of reactive-oxygen species may be quantified using enthalpies of formation to predict minimum inhibitory concentrations. Neither of these two quantitative structure- activity values describes the dynamic, in situ behavioral changes in the bacteria's struggle to survive. In this paper, borrowing parameters from logistic, oscillatory, and diauxic growth models, we use principal component analysis and agglomerative hierarchical clustering to classify survival modes across nanoparticle types and concentrations. We compare the growth parameters of 170 experimental interactions between nanoparticles and bacteria. The bacteria studied include Escherichia coli, Staphylococcus aureus, Methicillin-Resistant Staphylococcus aureus, Staphylococcus epidermidis, Pseudomonas aeruginosa, and Helicobacter pylori, and were tested across multiple concentrations of liposomal drug delivery systems, amphiphilic peptide, and silver and selenium nanoparticles. Clustering reveals specific pairs of bacteria and nanoparticles where the nanoparticle induced growth dynamics could potentially spread the infection through the development of resistance and tolerance. This rapid screening also shows that bacteria generated nanoparticles do not induce growth modes indicative of the development of resistance. This methodology can be used to rapidly screen for novel therapeutics that do not induce resistance before using more robust intracellular content screening. This methodology can also be used as a quality check on batch manufactured nanoparticles.