OBJECTIVES Left ventricular assist devices (LVAD) provide mechanical circulatory support for patients with advanced heart failure. Intracranial hemorrhage in this population represent a significant management challenge. The objective of this study is to report our initial experience on same-admission outcomes with LVAD patients that presented with various types of intracranial hemorrhage (ICH). PATIENTS AND METHODS A retrospective review of a large volume center over a two-year period was performed. LVAD patients with ICH requiring a neurosurgical consultation were identified. Hemorrhage type, interventions, discharge disposition and cause of death were recorded. RESULTS 27 LVAD patients with ICH received a neurosurgical consultation. The average INR at the time of ICH was 2.7 (1.0-8.8). EPZ015666 Hemorrhage types seen were lobar (10/27, 37 %), SAH (5/27, 19 %), SDH (4/27, 15 %), cerebellar ICH (3/27, 11 %), multiple ICH (3/27, 11 %), and hemorrhagic conversion (2/27, 7 %). The overall mortality rate was 48.2 % (13/27), with the highest mortality being in those patients who had multiple ICH at the time of presentation (3/3, 100 %). The majority of patients with ICH (85.2 %) were non-operative. Lobar IPH was less then 3 cm in 80 % (8/10) of these, and 6/8 (75 %) ultimately died. 11 %(3/27) received surgical intervention. Of these, 67 % ultimately withdrew care. 77 % (10/13) of patients died as a result of the ICH. 80 % of patients with SAH were ultimately discharged home. CONCLUSIONS Patients with a LVAD and ICH have a high rate of same-admission mortality (48 %). Hemorrhage location, intra-axial or extra-axial, resulted in patients being a risk for death secondary to either the hemorrhage itself or pump thrombosis, respectively. V.This study was aimed to investigate the inhibition effect of thiol-type antioxidants on protein oxidative aggregation caused by free radicals and the underlying mechanisms using six different thiol-type antioxidants (N-acetyl-L-cysteine, methionine, taurine, alpha-lipoic acid, glutathione and thioproline), Cu2+-H2O2 as a free radical generator (mainly a hydroxyl radical generator) and bovine serum albumin as the model protein. The inhibition effect of these antioxidants on protein oxidative aggregation and protective effect against oxidative damage in mouse brain tissues were investigated using SDS-PAGE, intrinsic fluorescence, simultaneous fluorescence, thioflavin T fluorescence, Congo red absorbance and inverted microscope. The results showed that all six antioxidants could inhibit protein oxidative aggregation by scavenging free radicals. In addition, alpha-lipoic acid could also bind to proteins via hydrophobic interactions and thioproline could bind to proteins via hydrogen bonds and van der Waals forces, thereby showing much stronger inhibition effect than others. Moreover, alpha-lipoic acid and thioproline could effectively prevent oxidative damage of mouse brain tissues. These results suggest that alpha-lipoic acid and thioproline can effectively inhibit free radical-induced protein aggregation and brain damage, which are worth testing for further anti-Alzheimer properties. The fabricating of metal oxide thin films onto conducting surfaces continues to grow and their potential applications as surfaces for biosensor applications is of paramount importance. The correct orientation of glucose oxidase redox enzymes yields very important biointerfaces capable of selectively detecting d-glucose as a measure of blood sugar for healthy and diabetic sick patients. The electrodeposition of redox enzymes, such as glucose oxidase enzymes, onto gold electrode surfaces pre-modified with nickel oxide was investigated in this work. The surface characterization confirmed the chemical nature, morphology and thin film composition of the modification of bare and modified gold electrodes. The electrodeposition of GOx enzyme onto nickel oxide/hydroxide thin film resulted in a surface with excellent bioelectrocatalytic properties towards the detection of d-glucose. The nickel within the nickel oxide/hydroxide thin film had a Ni(II) oxidation state. A well-defined redox peak of GOx enzyme co-factor (FAD/FADH2) was observed confirming the oriented immobilization onto NiO/Ni(OH)2 conducting surfaces. The amount of GOx enzyme deposited was determined by integrating the charge (Q = 0.368 μF) under the reduction peak and the surface coverage was found to be 1.43 × 10-10 mol. cm-2. A linear plot of electrocatalytic reduction currents against d-glucose concentrations was obtained up to 30.0 mM with a linear correlation coefficient (R2) of 0.99. The limit of detection (LoD) using S/N = 3 was calculated to be 1.54 ± 0.03 mM. The sensitivity of the biosensors was 1.95 ± 0.13 μA.mM.cm-2. The selectivity towards only d-glucose and not ascorbic acid and uric acid was evaluated and the Au-NiO/Ni(OH)2-GOx could not detect 1.0 mM of ascorbic acid and uric acid. Non-negative matrix factorization (NMF) is a knowledge discovery method that is used in many fields. Variational inference and Gibbs sampling methods for it are also well-known. However, the variational approximation error has not been clarified yet, because NMF is not statistically regular and the prior distribution used in variational Bayesian NMF (VBNMF) has zero or divergence points. In this paper, using algebraic geometrical methods, we theoretically analyze the difference in negative log evidence (a.k.a. free energy) between VBNMF and Bayesian NMF, i.e., the Kullback-Leibler divergence between the variational posterior and the true posterior. We derive an upper bound for the learning coefficient (a.k.a. the real log canonical threshold) in Bayesian NMF. By using the upper bound, we find a lower bound for the approximation error, asymptotically. The result quantitatively shows how well the VBNMF algorithm can approximate Bayesian NMF; the lower bound depends on the hyperparameters and the true non-negative rank. A numerical experiment demonstrates the theoretical result. Although it is one of the most widely used methods in recommender systems, Collaborative Filtering (CF) still has difficulties in modeling non-linear user-item interactions. Complementary to this, recently developed deep generative model variants (e.g., Variational Autoencoder (VAE)) allowing Bayesian inference and approximation of the variational posterior distributions in these models, have achieved promising performance improvement in many areas. However, the choices of variation distribution - e.g., the popular diagonal-covariance Gaussians - are insufficient to recover the true distributions, often resulting in biased maximum likelihood estimates of the model parameters. Aiming at more tractable and expressive variational families, in this work we extend the flow-based generative model to CF for modeling implicit feedbacks. We present the Collaborative Autoregressive Flows (CAF) for the recommender system, transforming a simple initial density into more complex ones via a sequence of invertible transformations, until a desired level of complexity is attained.