Exploring the Potential of Variational Autoencoders for Modeling Nonlinear Relationships in Psychological Data.

Latent variables analysis is an important part of psychometric research. In this context, factor analysis and other related techniques have been widely applied for the investigation of the internal structure of psychometric tests. However, these methods perform a linear dimensionality reduction under a series of assumptions that could not always be verified in psychological data. Predictive techniques, such as artificial neural networks, could complement and improve the exploration of latent space, overcoming the limits of traditional methods. In this study, we explore the latent space generated by a particular artificial neural network: the variational autoencoder. This autoencoder could perform a nonlinear dimensionality reduction and encourage the latent features to follow a predefined distribution (usually a normal distribution) by learning the most important relationships hidden in data. In this study, we investigate the capacity of autoencoders to model item-factor relationships in simulated data, which encompasses linear and nonlinear associations. We also extend our investigation to a real dataset. Results on simulated data show that the variational autoencoder performs similarly to factor analysis when the relationships among observed and latent variables are linear, and it is able to reproduce the factor scores. Moreover, results on nonlinear data show that, differently than factor analysis, it can also learn to reproduce nonlinear relationships among observed variables and factors. The factor score estimates are also more accurate with respect to factor analysis. The real case results confirm the potential of the autoencoder in reducing dimensionality with mild assumptions on input data and in recognizing the function that links observed and latent variables.

Catalytic Performance in Nitroarene Reduction of Nanocatalyst Based on Noble Metal Nanoparticles Supported on Polymer/s-Layer Protein Hybrids.

We present a novel bionanocatalyst fabricated by the adsorption-reduction of metal ions on a polyurethane/S-layer protein biotemplate. The bioinspired support was obtained by the adsorption of S-layer proteins (isolated from Lentilactobacillus kefiri) on polyurethane particles. Silver and platinum nanoparticles were well-loaded on the surface of the support after the combination with metallic salts and reduction with H2 at room temperature. Transmission electron microscopy analysis revealed the strawberry-like morphology of the bionanocatalysts with a particle size, dn, of 2.39 nm for platinum and 9.60 nm for silver. Both systems catalyzed the hydrogenation of p-nitrophenol to p-aminophenol with high efficiency in water at mild conditions in the presence of NaBH4. Three different amounts of bionanocatalyst were tested, and in all cases, conversions between 97 and 99% were observed. The catalysts displayed excellent recyclability over ten cycles, and no extensive damage in their nanostructure was noted after them. The bionanocatalysts were stable during their production, storage, and use, thanks to the fact that the biosupport provides an effective driving force in the formation and stabilization of the metallic nanoparticles. The successful bioinspired production strategy and the good catalytic ability of the systems are encouraging in the search for nontoxic, simple, clean, and eco-friendly procedures for the synthesis and exploitation of nanostructures.

Artificial Neural Networks for Short-Form Development of Psychometric Tests: A Study on Synthetic Populations Using Autoencoders.

Short-form development is an important topic in psychometric research, which requires researchers to face methodological choices at different steps. The statistical techniques traditionally used for shortening tests, which belong to the so-called exploratory model, make assumptions not always verified in psychological data. This article proposes a machine learning-based autonomous procedure for short-form development that combines explanatory and predictive techniques in an integrative approach. The study investigates the item-selection performance of two autoencoders: a particular type of artificial neural network that is comparable to principal component analysis. The procedure is tested on artificial data simulated from a factor-based population and is compared with existent computational approaches to develop short forms. Autoencoders require mild assumptions on data characteristics and provide a method to predict long-form items' responses from the short form. Indeed, results show that they can help the researcher to develop a short form by automatically selecting a subset of items that better reconstruct the original item's responses and that preserve the internal structure of the long-form.

Use of carbon from the agricultural industry for the synthesis of catalysts destined to oxyanions removal from water.

Remediation of water contaminated with oxyanions is of great importance due to the toxicity and environmental persistence of these chemical species. The present work describes the elimination of different oxyanions in water using catalysts supported on active carbon obtained from an agricultural residue (peanut shells, CPeanut) and active commercial carbon (CCom) in order to compare their structural and catalytic properties. The synthesized CPeanut and CCom were fully characterized by surface analysis, TGA, TPR, SEM-EDX, FT-IR, and TEM. It was observed that CPeanut presented similar superficial characteristics to CCom, being an adequate support to synthesize catalysts. With both carbons, catalysts based on Cu, Pd, and PdCu were prepared and evaluated in the elimination of NO3-, NO2-, and BrO3- from water using H2 as a reducing agent. The bimetallic catalysts prepared on both supports were active in the oxyanions reduction, obtaining good selectivities to the products of interest. In this sense, this work presents a potential re-use of agricultural wastes by preparing activated carbon from peanut shell residues in order to reduce the waste volume generated. In addition, the material synthetized is low cost due to its large-scale production and great availability in Argentina. The carbon obtained from the peanut shells provides a potential application in the environmental remediation of water contaminated with oxyanions.

Catalytic hydrogenation of nitrate in water: improvement of the activity and selectivity to N2 by using Rh(III)-hexamolybdate supported on ZrO2-Al2O3.

Catalysts prepared on ZrO2, Al2O3 and ZrO2-Al2O3 (ZrAl-10) supported with Anderson heteropolyanion (RhMo6) as active phase were investigated for the elimination of NO3- from water. Raman characterization of pure and supported RhMo6 phase showed the presence of polymolybdic species of different degrees of complexity when RhMo6 was supported. The temperature-programmed reduction study revealed the synergic effect between Rh and Mo species, through which the reducibility of Mo was promoted by Rh, and different phase/support interactions were verified. Among the supports, ZrAl-10 presented the highest acidity due to the presence of ZrO2 in the tetragonal modification and high specific surface area (due to Al2O3), favouring rhodium-molybdenum active phase/support interaction and high dispersion. All catalysts prepared were active in removing NO3-, the one prepared with the RhMo6 phase on the ZrAl-10 support being the most active. These results point to the formation of an active surface with a high dispersion of Rh and Mo. The highest selectivity to N2 (99.3) exhibited by the RhMo6/ZrAl-10 catalyst is proposed to be related to the high Rh dispersion (0.755) and to the presence of Lewis acid sites (oxygen vacancies) of the tetragonal ZrO2 modification that favour NO3- adsorption through electrostatic interactions.

Promoting Effect of Palladium on ZnAl2O4-Supported Catalysts Based on Cobalt or Copper Oxide on the Activity for the Total Propene Oxidation.

Palladium-modified Co-ZnAland Cu-ZnAl materials were used and found active for the catalytic oxidation of propene and propane. According to the results obtained by XRD, TPR and XPS, the zinc aluminate-supported phases are oxide phases, Co3O4, CuO and PdOx for Co-ZnAl, Cu-ZnAl and Pd-ZnAl catalysts, respectively. These reducible oxide species present good catalytic activity for the oxidation reactions. The addition of palladium to Co-ZnAl or Cu-ZnAl samples promoted the reducibility of the system and, consequently, produced a synergic effect which enhanced the activity for the propene oxidation. The PdCo-ZnAl sample was the most active and exhibited highly dispersed PdOx particles and surface structural defects. In addition, it exhibited good catalytic stability. The H2 pre-treated PdCu-ZnAl, PdCo-ZnAl and Pd-ZnAl samples showed higher activity than the original oxide catalysts, evidencing the important role of the oxidation state of the species, mainly of the palladium species, on the catalytic activity for the propene combustion. The synergic effect between metal transition oxides and PdOx could not be observed for the propane oxidation.

Synthesis and Catalytic Application of Silver Nanoparticles Supported on Lactobacillus kefiri S-Layer Proteins.

Research on nanoparticles obtained on biological supports is a topic of growing interest in nanoscience, especially regarding catalytic applications. Silver nanoparticles (AgNPs) have been studied due to their low toxicity, but they tend to aggregation, oxidation, and low stability. In this work, we synthesized and characterized AgNPs supported on S-layer proteins (SLPs) as bidimensional regularly arranged biotemplates. By different reduction strategies, six AgNPs of variable sizes were obtained on two different SLPs. Transmission electron microscopy (TEM) images showed that SLPs are mostly decorated by evenly distributed AgNPs; however, a drastic reduction by NaBH4 led to large AgNPs whereas a smooth reduction with H2 or H2/NaBH4 at low concentration leads to smaller AgNPs, regardless of the SLP used as support. All the nanosystems showed conversion values between 75-80% of p-nitrophenol to p-aminophenol, however, the increment in the AgNPs size led to a great decrease in Kapp showing the influence of reduction strategy in the performance of the catalysts. Density functional theory (DFT) calculations indicated that the adsorption of p-nitrophenolate species through the nitro group is the most favored mechanism, leading to p-aminophenol as the only feasible product of the reaction, which was corroborated experimentally.

Platinum Nanoparticles Obtained at Mild Conditions on S-Layer Protein/Polymer Particle Supports.

This work presents the synthesis of platinum nanoparticles supported on S-layer protein/polymeric particle systems, obtained by combining proteins isolated from Lactobacillus kefiri and an aqueous dispersion of acrylic particles. FTIR spectra of the protein/polymer supports did not show changes in the Amide I band of the proteins, suggesting that proteins maintained their conformation after adsorption. The SAXS spectra and DLS results are consistent with the formation of a protein corona around the polymer particles. After combining the supports with the platinum complex and subsequently reducing the combination with hydrogen at mild conditions, we obtained colloidal nanocomposite materials. In these, platinum nanoparticles with diameters around 3 nm located on the surface of the protein/polymer supports were observed by TEM. The obtained nanosystems showed catalytic activity in the reduction of p-nitrophenol with NaBH4 at room temperature with conversions of 100% for reaction times of 50 to 70 min.

Bimetallic PtSn/C catalysts obtained via SOMC/M for glycerol steam reforming.

A detailed study on the preparation of bimetallic PtSn/C catalysts using surface-controlled synthesis methods, and on their catalytic performance in the glycerol steam reforming reaction has been carried out. In order to obtain these well-defined bimetallic phases, techniques derived from Surface Organometallic Chemistry on Metals (SOMC/M) were used. The preparation process involved the reaction between an organometallic compound ((C4H9)4Sn) and a supported transition metal (Pt) in a H2 atmosphere. Catalysts with Sn/Pt atomic ratios of 0.2, 0.3, 0.5, and 0.7 were obtained, and characterized using several techniques: ICP, H2 chemisorption, TEM and XPS. These systems were tested in the glycerol steam reforming varying the reaction conditions (glycerol concentration and reaction temperature). The best performance was observed for the catalysts with the lowest tin contents (PtSn0.2/C and PtSn0.3/C). It was observed that the presence of tin increased the catalysts' stability when working under more severe reaction conditions.

Validity of the modified Charlson Comorbidity Index as predictor of short-term outcome in older stroke patients.

The modified Charlson Comorbidity Index (MCCI) has been proposed as a tool for adjusting the outcomes of stroke for comorbidity, but its validity in such a context has been evaluated in only a few studies and needs to be further explored, especially in elderly patients. We aimed to retrospectively assess the validity of the MCCI as a predictor of the short-term outcomes in a cohort of 297 patients with first-ever ischemic stroke, older than 60 years, and managed according to a clinical pathway. The poor outcome (PO) at 1 month, defined as a modified Rankin Scale score of 3-6, was the primary end point. Furthermore, a new comorbidity index has been developed, specific to our cohort, according to the same statistical approach used for the original CCI. The MCCI showed a positive association with PO (odds ratio [OR] 1.62; 95% confidence interval [CI] .98-2.68) and mortality (hazard ratio [HR] 1.85; 95% CI .94-3.61), not statistically significant and totally dependent on its association with the severity of neurologic impairment at onset. The new comorbidity index showed, as expected, a significant association with the PO and mortality with higher point estimates of OR (2.74; 95% CI 1.64-4.59) and HR (2.73; 95% CI 1.51-4.94), but this association was also dependent on stroke severity and premorbid disability. Our results do not support the validity of the MCCI as a predictor of the short-term outcomes in elderly stroke patients nor could we develop a more valid index from the available data. This suggests the need for development of disease- and age-specific indexes, possibly according to a prospective design. In any case, initial stroke severity, a strong predictor of outcome, is associated with the degree of comorbidity.

Impact of gender-age interaction on the outcome of ischemic stroke in an Italian cohort of patients treated according to a standardized clinical pathway.

BACKGROUND: Stroke outcome has been reported as worse in women, especially in terms of disability. As for mortality, the data are conflicting, with some reports suggesting a female advantage. Our objective was to explore such issues in an Italian cohort of patients managed by a standardized clinical pathway (CPW) and, as such, homogeneous in terms of clinical management. METHODS: Data from a cohort of 1993 patients (987 women and 1006 men) with first-ever ischemic stroke, consecutively referred to an in-hospital Clinical Pathway Program from January 1, 2001 to December 31, 2009, were retrospectively analyzed. The relationship between female gender and one-month outcome was assessed with adjustment for age, stroke severity and premorbid disability. RESULTS: The outcome was worse in women in terms of disability (age-adjusted odds ratio 2.03, 95% CI 1.69-2.46), while no difference was found for mortality. In multivariate models, female gender turned out to be associated with a lower case-fatality rate (adjusted hazard ratio 0.65, 95% CI 0.48-0.89, P=0.007), whereas the odds ratio for disability decreased but remained significant (OR 1.30; 95% CI 1.01-1.69). We found a significant interaction between gender and age in the case-fatality rate, and a female survival advantage was apparent only below 50 years. CONCLUSIONS: Our study confirms the excess risk of disability after stroke in women, although it is mostly explained by the occurrence of the most severe clinical syndromes. As for mortality, female gender seems to play a protective role, at least in the short-term and in younger patients.