Closed-loop optimization of fast-charging protocols for batteries with machine learning.

Simultaneously optimizing many design parameters in time-consuming experiments causes bottlenecks in a broad range of scientific and engineering disciplines1,2. One such example is process and control optimization for lithium-ion batteries during materials selection, cell manufacturing and operation. A typical objective is to maximize battery lifetime; however, conducting even a single experiment to evaluate lifetime can take months to years3-5. Furthermore, both large parameter spaces and high sampling variability3,6,7 necessitate a large number of experiments. Hence, the key challenge is to reduce both the number and the duration of the experiments required. Here we develop and demonstrate a machine learning methodology  to efficiently optimize a parameter space specifying the current and voltage profiles of six-step, ten-minute fast-charging protocols for maximizing battery cycle life, which can alleviate range anxiety for electric-vehicle users8,9. We combine two key elements to reduce the optimization cost: an early-prediction model5, which reduces the time per experiment by predicting the final cycle life using data from the first few cycles, and a Bayesian optimization algorithm10,11, which reduces the number of experiments by balancing exploration and exploitation to efficiently probe the parameter space of charging protocols. Using this methodology, we rapidly identify high-cycle-life charging protocols among 224 candidates in 16 days (compared with over 500 days using exhaustive search without early prediction), and subsequently validate the accuracy and efficiency of our optimization approach. Our closed-loop methodology automatically incorporates feedback from past experiments to inform future decisions and can be generalized to other applications in battery design and, more broadly, other scientific domains that involve time-intensive experiments and multi-dimensional design spaces.

T315I-mutated myeloid sarcoma.

Myeloid Sarcoma (MS) is diagnosed by an extramedullary proliferation of immature granulocytic cells. Its association with chronic myeloid leukemia (CML) is rare. CML is characterized by BCR-ABL1 gene rearrangement and therapies with tyrosine kinase inhibitors (TKI) are very effective. However, TKI resistance may occur secondary to the development of ABL1 mutations. T315I is a common mutation that accounts for ∼20% clinical resistance to TKIs. We report the first case of a patient with T315I mutated myeloid sarcoma that occurred after complete cytogenetic response with dasatinib of a chronic phase CML. The patient was successfully treated with induction chemotherapy and ponatinib.

node2vec: Scalable Feature Learning for Networks.

Prediction tasks over nodes and edges in networks require careful effort in engineering features used by learning algorithms. Recent research in the broader field of representation learning has led to significant progress in automating prediction by learning the features themselves. However, present feature learning approaches are not expressive enough to capture the diversity of connectivity patterns observed in networks. Here we propose node2vec, an algorithmic framework for learning continuous feature representations for nodes in networks. In node2vec, we learn a mapping of nodes to a low-dimensional space of features that maximizes the likelihood of preserving network neighborhoods of nodes. We define a flexible notion of a node's network neighborhood and design a biased random walk procedure, which efficiently explores diverse neighborhoods. Our algorithm generalizes prior work which is based on rigid notions of network neighborhoods, and we argue that the added flexibility in exploring neighborhoods is the key to learning richer representations. We demonstrate the efficacy of node2vec over existing state-of-the-art techniques on multi-label classification and link prediction in several real-world networks from diverse domains. Taken together, our work represents a new way for efficiently learning state-of-the-art task-independent representations in complex networks.

Triamcinolone acetonide nanoparticles incorporated in thermoreversible gels for age-related macular degeneration.

Age-related macular degeneration (AMD) is one of the leading causes of blindness in the US affecting millions yearly. It is characterized by intraocular neovascularization, inflammation and retinal damage which can be ameliorated through intraocular injections of glucocorticoids. However, the complications that arise from repetitive injections as well as the difficulty posed by targeting the posterior segment of the eye make this interesting territory for the development of novel drug delivery systems (DDS). In the present study, we described the development of a DDS composed of triamcinolone acetonide-encapsulated PEGylated PLGA nanoparticles (NP) incorporated into PLGA-PEG-PLGA thermoreversible gel and its use against VEGF expression characteristic of AMD. We found that the NP with mean size of 208 ± 1.0 nm showed uniform size distribution and exhibited sustained release of the drug. We also demonstrated that the polymer can be injected as a solution and transition to a gel phase based on the biological temperature of the eye. Additionally, the proposed DDS was non-cytotoxic to ARPE-19 cells and significantly reduced VEGF expression by 43.5 ± 3.9% as compared to a 1.53 ± 11.1% reduction with triamcinolone. These results suggest the proposed DDS will contribute to the development of novel therapeutic strategies for AMD.

Brain-targeted delivery of docetaxel by glutathione-coated nanoparticles for brain cancer.

Gliomas are some of the most aggressive types of cancers but the blood-brain barrier acts as an obstacle to therapeutic intervention in brain-related diseases. The blood-brain barrier blocks the permeation of potentially toxic compounds into neural tissue through the interactions of brain endothelial cells with glial cells (astrocytes and pericytes) which induce the formation of tight junctions in endothelial cells lining the blood capillaries. In the present study, we characterize a glutathione-coated docetaxel-loaded PEG-PLGA nanoparticle, show its in vitro drug release data along with cytotoxicity data in C6 and RG2 cells, and investigate its trans-blood-brain barrier permeation through the establishment of a Transwell cellular co-culture. We show that the docetaxel-loaded nanoparticle's size enables its trans-blood-brain barrier permeation; the nanoparticle exhibits a steady, sustained release of docetaxel; the drug is able to induce cell death in glioma models; and the glutathione-coated nanoparticle is able to permeate through the Transwell in vitro blood-brain barrier model.

Preparation and characterization of methylene blue nanoparticles for Alzheimer's disease and other tauopathies.

Methylene blue (MB) has been shown to slow down the progression of the Alzheimer's disease (AD) and other tauopathies; however distribution of MB into the brain is limited due its high hydrophilicity. In this study, we aimed to prepare novel hydrophobic glutathione coated PLGA nanoparticles to improve bioavailability of MB in the brain. Glutathione coated poly-(lactide-co-glycolide) (PLGA-b-PEG) nanoparticles (NPs) were prepared and tested in two different cell culture models of AD expressing microtubule associated protein tau (tau). The NPs showed a particle size averaging 136.5±4.4nm, which is suitable for the blood brain barrier (BBB) permeation. The in vitro release profile of the NPs exhibited no initial burst release and showed sustained drug release for up to 144 hours. Interestingly, treatment of newly formulated MB-NPs showed a potent reduction in both endogenous and over expressed tau protein levels in human neuroblastoma SHSY-5Y cells expressing endogenous tau and transfected HeLa cells over-expressing tau protein, respectively. Furthermore, in vitro BBB Transwell™ study showed significantly higher permeation of MB-NP compared to the MB solution through the co culture of rat brain endothelial 4 (RBE4) and C6 astrocytoma cells (p<0.05). The proposed MB loaded nanoparticles could provide a more effective treatment option for AD and many other related disorders.

Loss of phosphatase and tensin homolog (PTEN) induces leptin-mediated leptin gene expression: feed-forward loop operating in the lung.

Elevated levels of systemic and pulmonary leptin are associated with diseases related to lung injury and lung cancer. However, the role of leptin in lung biology and pathology, including the mechanism of leptin gene expression in the pathogenesis of lung diseases, including lung cancer, remains elusive. Here, using conditional deletion of tumor suppressor gene Pten in the lung epithelium in vivo in transgenic mice and human PTEN-null lung epithelial cells, we identify the leptin-driven feed-forward signaling loop in the lung epithelial cells. Leptin-mediated leptin/leptin-receptor gene expression likely amplifies leptin signaling that may contribute to the pathogenesis and severity of lung diseases, resulting in poor clinical outcomes. Loss of Pten in the lung epithelial cells in vivo activated adipokine signaling and induced leptin synthesis as ascertained by genome-wide mRNA profiling and pathway analysis. Leptin gene transcription was mediated by binding of transcription factors NRF-1 and CCAAT/enhancer-binding protein δ (C/EBP) to the proximal promoter regions and STAT3 to the distal promoter regions as revealed by leptin promoter-mutation, chromatin immunoprecipitation, and gain- and loss-of-function studies in lung epithelial cells. Leptin treatment induced expression of the leptin/leptin receptor in the lung epithelial cells via activation of MEK/ERK, PI3K/AKT/mammalian target of rapamycin (mTOR), and JAK2/STAT3 signaling pathways. Expression of constitutively active MEK-1, AKT, and STAT3 proteins increased expression, and treatment with MEK, PI3K, AKT, and mTOR inhibitors decreased LEP expression, indicating that leptin via MAPK/ERK1/2, PI3K/AKT/mTOR, and JAK2/STAT3 pathways, in turn, further induces its own gene expression. Thus, targeted inhibition of the leptin-mediated feed-forward loop provides a novel rationale for pharmacotherapy of disease associated with lung injury and remodeling, including lung cancer.