Explore the vast range of titles available.

Generative models are becoming a tool of choice for exploring the molecular space. These models learn on a large training dataset and produce novel molecular structures with similar properties. Generated structures can be utilized for virtual screening or training semi-supervized predictive models in the downstream tasks. While there are plenty of generative models, it is unclear how to compare and rank them. In this work, we introduce a benchmarking platform called Molecular Sets (MOSES) to standardize training and comparison of molecular generative models. MOSES provides training and testing datasets, and a set of metrics to evaluate the quality and diversity of generated structures. We have implemented and compared several molecular generation models and suggest to use our results as reference points for further advancements in generative chemistry research. The platform and source code are available at https://github.com/molecularsets/moses .

To determine the role of pyroglutamic acid (PGA) in the pathogenesis of unexplained metabolic acidosis in critically ill patients. Case series in the medical ICU of an urban hospital. 23 patients admitted to the medical ICU with acidemia (pH <7.35 or HC0(3) < or = 16 mEq/l) not explained by the presence of ketoacidosis, lactic acidosis, renal failure or ingestion of drugs or toxins and who had an increase in the strong ion gap (SIG) greater than 5. Plasma levels of sodium, potassium, chloride, bicarbonate, calcium (ionized), magnesium, lactate, phosphate, albumin, blood urea nitrogen, and creatinine were measured. Arterial blood gases and urine dipstick for ketones were also analyzed. Plasma was assayed for PGA using gas chromatography. The patient's history and Kardex were reviewed for evidence of acetaminophen administration. The plasma PGA level was found to be very low in all patients studied. The correlation between SIG and PGA (r) was -0.01 (95% CI: -0.42 to 0.40). PGA therefore did not account for the observed increase in the SIG. There appeared to be no obvious influence of acetaminophen intake on levels of PGA in the plasma. We were unable to confirm the importance of PGA as a cause of unexplained metabolic acidosis and increased SIG in our critically ill patients.

The article proposes a new type of AI system that does not give solutions directly but rather points toward it, friendly prompting the user with questions and adjusting messages. Models of AI human collaboration can be deduced from the classic literary example of interaction between Mr. Holmes and Dr. Watson from the stories by Conan Doyle, where the highly qualified expert Mr. Holmes answers questions posed by Dr. Watson. Here Mr. Holmes, with his rule-based calculations, logic, and memory management, apparently plays the role of an AI system, and Dr. Watson is the user. Looking into the same Holmes-Watson interaction, we find and promote another model in which the AI behaves like Dr. Watson, who, by asking questions and acting in a particular way, helps Holmes (the AI user) make the right decisions. We call the systems based on this principle \"Dr. Watson-type systems.\" The article describes the properties of such systems and introduces two particular: Patient Management System for intensive care physicians and Data Error Prevention System.

Generative models are becoming a tool of choice for exploring the molecular space. These models learn on a large training dataset and produce novel molecular structures with similar properties. Generated structures can be utilized for virtual screening or training semi-supervised predictive models in the downstream tasks. While there are plenty of generative models, it is unclear how to compare and rank them. In this work, we introduce a benchmarking platform called Molecular Sets (MOSES) to standardize training and comparison of molecular generative models. MOSES provides a training and testing datasets, and a set of metrics to evaluate the quality and diversity of generated structures. We have implemented and compared several molecular generation models and suggest to use our results as reference points for further advancements in generative chemistry research. The platform and source code are available at https://github.com/molecularsets/moses.

A cycle of works on manufacturing and studying laser and magnetooptical ceramics with a focus on their thermo-optical characteristics performed by the research team is analyzed. Original results that have not been published before such as measurements of the Verdet constant in the Zr:TAG, Re:MgAl2O4, and ZnAl2O4 ceramics are also presented.

A glycine-nitrate self-propagating high-temperature synthesis (SHS) was developed to produce composite MgO-Gd 2 O 3 nanopowders. The X-ray powder diffraction (XRD) analysis confirmed the SHS-product consists of cubic MgO and Gd 2 O 3 phases with nanometer crystallite size and retains this structure after annealing at temperatures up to 1200 °C. Near full dense high IR-transparent composite ceramics were fabricated by spark plasma sintering (SPS) at 1140 °C and 60 MPa. The in-line transmittance of 1 mm thick MgO-Gd 2 O 3 ceramics exceeded 70% in the range of 4–5 mm and reached a maximum of 77% at a wavelength of 5.3 mm. The measured microhardness HV0.5 of the MgO-Gd 2 O 3 ceramics is 9.5±0.4 GPa, while the fracture toughness ( K IC ) amounted to 2.0±0.5 MPa·m 1/2 . These characteristics demonstrate that obtained composite MgO-Gd 2 O 3 ceramic is a promising material for protective infra-red (IR) windows.

Deep learning methods provide significant assistance in analyzing coronavirus disease (COVID-19) in chest computed tomography (CT) images, including identification, severity assessment, and segmentation. Although the earlier developed methods address the lack of data and specific annotations, the current goal is to build a robust algorithm for clinical use, having a larger pool of available data. With the larger datasets, the domain shift problem arises, affecting the performance of methods on the unseen data. One of the critical sources of domain shift in CT images is the difference in reconstruction kernels used to generate images from the raw data (sinograms). In this paper, we show a decrease in the COVID-19 segmentation quality of the model trained on the smooth and tested on the sharp reconstruction kernels. Furthermore, we compare several domain adaptation approaches to tackle the problem, such as task-specific augmentation and unsupervised adversarial learning. Finally, we propose the unsupervised adaptation method, called F-Consistency, that outperforms the previous approaches. Our method exploits a set of unlabeled CT image pairs which differ only in reconstruction kernels within every pair. It enforces the similarity of the network’s hidden representations (feature maps) by minimizing the mean squared error (MSE) between paired feature maps. We show our method achieving a 0.64 Dice Score on the test dataset with unseen sharp kernels, compared to the 0.56 Dice Score of the baseline model. Moreover, F-Consistency scores 0.80 Dice Score between predictions on the paired images, which almost doubles the baseline score of 0.46 and surpasses the other methods. We also show F-Consistency to better generalize on the unseen kernels and without the presence of the COVID-19 lesions than the other methods trained on unlabeled data.

This study looked at optimizing the composition of precursors for yttria nanopowder glycine–nitrate self-propagating high-temperature synthesis (SHS). Based on thermodynamic studies, six different precursor compositions were selected, including with excesses of either oxidant or fuel. The powders from the precursors of all selected compositions were highly dispersed and had specific surface areas ranging from 22 to 57 m2/g. They were consolidated by hot pressing (HP) with lithium–fluoride sintering additive and subsequent hot isostatic pressing (HIP). The 1 mm thick HPed ceramics had transmittance in the range of 74.5% to 80.1% @ 1μm, which was limited by optical inhomogeneity due to incomplete evaporation of the sintering additive. Two-stage HIP significantly improves optical homogeneity of the ceramics. It was shown that an excess of oxidizer in the precursor decreases the powders’ agglomeration degree, which forms large pore clusters in the ceramics.

Nanopowders of strontium fluoroapatite Sr5(PO4)3F (SFAP) were synthesized using a co-precipitation method with different starting strontium compounds. Based on the data of XRD, BET and SEM measurements, the nitrate-derived powders were chosen as the least agglomerated. The SFAP powders were hot pressed at 1000 °C to ceramic samples with a transmittance up to 82% in a mid-IR region. The designed approach was adopted to prepare 2 mol % of Er-doped SFAP powders and ceramics. It was established that Er:SFAP ceramics have luminescence in the range of 1.5–1.7 μm, the intensity of which increases with the calcination temperature of the initial powders.

Diffuse gliomas continue to be an important problem in neuro-oncology. To solve it, studies have considered the issues of molecular pathogenesis from the intratumoral heterogeneity point. Here, we carried out a comparative dynamic analysis of the different cell populations’ content in diffuse gliomas of different molecular profiles and grades, considering the cell populations’ functional properties and the relationship with patient survival, using flow cytometry, immunofluorescence, multiparametric fluorescent in situ hybridization, polymerase chain reaction, and cultural methods. It was shown that an increase in the IDH-mutant astrocytomas and oligodendrogliomas malignancy is accompanied by an increase in stem cells’ proportion and mesenchymal cell populations’ appearance arising from oligodendrocyte-progenitor-like cells with cell plasticity and cells’ hypoxia response programs’ activation. In glioblastomas, malignancy increase is accompanied by an increase in both stem and definitive cells with mesenchymal differentiation, while proneuronal glioma stem cells are the most likely the source of mesenchymal glioma stem cells, which, in hypoxic conditions, further give rise to mesenchymal-like cells. Clinical confirmation was a mesenchymal-like cell and mesenchymal glioma stem cell number, and the hypoxic and plastic molecular programs’ activation degree had a significant effect on relapse-free and overall survival. In general, we built a multi-vector model of diffuse gliomas’ pathogenetic tracing up to the practical plane.