Is my model any good: differentially private regression diagnostics

Linear and logistic regression are popular statistical techniques for analyzing multi-variate data. Typically, analysts do not simply posit a particular form of the regression model, estimate its parameters, and use the results for inference or prediction. Instead, they first use a variety of diagnostic techniques to assess how well the model fits the relationships in the data and how well it can be expected to predict outcomes for out-of-sample records, revising the model as necessary to improve fit and predictive power. In this article, we develop \(\epsilon \)-differentially private diagnostics tools for regression, beginning to fill a gap in privacy-preserving data analysis. Specifically, we create differentially private versions of residual plots for linear regression and of receiver operating characteristic (ROC) curves as well as binned residual plot for logistic regression. The residual plot and binned residual plot help determine whether or not the data satisfy the assumptions underlying the regression model, and the ROC curve is used to assess the predictive power of the logistic regression model. These diagnostics improve the usefulness of algorithms for computing differentially private regression output, which alone does not allow analysts to assess the quality of the posited model. Our empirical studies show that these algorithms can be effective tools for allowing users to evaluate the quality of their models.     

A Gel Chromatographic Study of Molecular Weight Distribution in Native Wheat Starch and Its Oxidation Products

The distribution of molecular size of the native starch prepared according to a new method and of oxidised products of starch has been investigated with the help of agarose gel filtration. It has been noticed that the curve of the molecular distribution of the native wheat starch (product A) shows a big peak excluded by the gel with M̄_W > 2 · 10⁶ comprising about 58% of the sample and a part fractioned by the gel within the area 1 · 10⁴ < M̄_W < 2 · 10⁶. Comparative gel filterings of native maize, tapioca and potato starch have shown that the distribution of molecular size of wheat starch (A) most resembles that of maize starch. Tapioca starch and especially potato starch have a somewhat lower percentage of molecules within the part fractioned by the gel than wheat starch (A). Native wheat starch, product B has a somewhat higher percentage of molecules within the area 1 · 10⁵ > M̄_W > 2 · 10⁴ than the other native types of starch. Heat-treatening and oxidation of native starch causes a decrease or a disappearance of the macromolecular excluded part of the sample and instead increases the molecules within the fractioning area of the gel, M̄_W < 2 · 10⁶. The used gel chromatographic method is thus very suitable when investigating the distribution of molecules of oxidised starch products and in product control.     

The Endothelium and COVID-19: An Increasingly Clear Link Brief Title: Endotheliopathy in COVID-19

The endothelium has a fundamental role in the cardiovascular complications of coronavirus disease 2019 (COVID-19). Infection with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) particularly affects endothelial cells. The virus binds to the angiotensin-converting enzyme 2 (ACE-2) receptor (present on type 2 alveolar cells, bronchial epithelial cells, and endothelial cells), and induces a cytokine storm. The cytokines tumor necrosis factor alpha, interleukin-1 beta, and interleukin-6 have particular effects on endothelial cells—leading to endothelial dysfunction, endothelial cell death, changes in tight junctions, and vascular hyperpermeability. Under normal conditions, apoptotic endothelial cells are removed into the bloodstream. During COVID-19, however, endothelial cells are detached more rapidly, and do not regenerate as effectively as usual. The loss of the endothelium on the luminal surface abolishes all of the vascular responses mediated by the endothelium and nitric oxide production in particular, which results in greater contractility. Moreover, circulating endothelial cells infected with SARS-CoV-2 act as vectors for viral dissemination by forming clusters that migrate into the circulation and reach distant organs. The cell clusters and the endothelial dysfunction might contribute to the various thromboembolic pathologies observed in COVID-19 by inducing the formation of intravascular microthrombi, as well as by triggering disseminated intravascular coagulation. Here, we review the contributions of endotheliopathy and endothelial-cell-derived extracellular vesicles to the pathogenesis of COVID-19, and discuss therapeutic strategies that target the endothelium in patients with COVID-19.     

Development of High-Efficiency,Magnetically Separable Palladium-Decorated Manganese-Ferrite Catalyst for Nitrobenzene Hydrogenation

Aniline (AN) is one of the most important compounds in the chemical industry and is prepared by the catalytic hydrogenation of nitrobenzene (NB). The development of novel, multifunctional catalysts which are easily recoverable from the reaction mixture is, therefore, of paramount importance. Compared to conventional filtration, magnetic separation is favored because it is cheaper and more facile. For satisfying these requirements, we developed manganese ferrite (MnFe₂O₄)–supported, magnetically separable palladium catalysts with high catalytic activity in the hydrogenation of nitrobenzene to aniline. In addition to high NB conversion and AN yield, remarkable aniline selectivity (above 96 n/n%) was achieved. Surprisingly, the magnetic support alone also shows moderate catalytic activity even without noble metals, and thus, up to 94 n/n% nitrobenzene conversion, along with 47 n/n% aniline yield, are attainable. After adding palladium nanoparticles to the support, the combined catalytic activity of the two nanomaterials yielded a fast, efficient, and highly selective catalyst. During the test of the Pd/MnFe₂O₄ catalyst in NB hydrogenation, no by-products were detected, and consequently, above 96 n/n% aniline yield and 96 n/n% selectivity were achieved. The activity of the Pd/MnFe₂O₄ catalyst was not particularly sensitive to the hydrogenation temperature, and reuse tests indicate its applicability in at least four cycles without regeneration. The remarkable catalytic activity and other favorable properties can make our catalyst potentially applicable to both NB hydrogenation and other similar or slightly different reactions.     

Antitumor Effects of Ral-GTPases Downregulation in Glioblastoma

Glioblastoma (GBM) is the most common tumor in the central nervous system in adults. This neoplasia shows a high capacity of growth and spreading to the surrounding brain tissue, hindering its complete surgical resection. Therefore, the finding of new antitumor therapies for GBM treatment is a priority. We have previously described that cyclin D1-CDK4 promotes GBM dissemination through the activation of the small GTPases RalA and RalB. In this paper, we show that RalB GTPase is upregulated in primary GBM cells. We found that the downregulation of Ral GTPases, mainly RalB, prevents the proliferation of primary GBM cells and triggers a senescence-like response. Moreover, downregulation of RalA and RalB reduces the viability of GBM cells growing as tumorspheres, suggesting a possible role of these GTPases in the survival of GBM stem cells. By using mouse subcutaneous xenografts, we have corroborated the role of RalB in GBM growth in vivo. Finally, we have observed that the knockdown of RalB also inhibits cell growth in temozolomide-resistant GBM cells. Overall, our work shows that GBM cells are especially sensitive to Ral-GTPase availability. Therefore, we propose that the inactivation of Ral-GTPases may be a reliable therapeutic approach to prevent GBM progression and recurrence.