Universidad Yachay Tech

Kalanchoe pinnata, Lantana camara L., and Melia azedarach L. are common plants in Cumana, Venezuela, and are traditionally used by locals to treat infections. This research aimed to evaluate the phytochemical profile and antibacterial and cytotoxic activities of the ethanolic leaf extracts from these plants. Specimens were collected and identified. Using qualitative colorimetric and precipitation tests, the profile of secondary metabolites in the ethanolic extract of the leaves of each plant was determined. These metabolites were confirmed by FT-IR and UV-visible spectroscopies. Antimicrobial and cytotoxic activities were evaluated using the disk diffusion method and the Artemia salina model, respectively. All three extracts contain coumarins and polyphenols. K. pinnata and Lantana camara L. contain unsaturated sterols, but alkaloids and saponins were only detected in L. camara. None of the extracts exhibited antibacterial activity at the concentrations tested. However, according to the Clarkson Toxicity Index, all extracts showed medium toxic activity (LC50: 100–500 µg mL-1) against A. salina nauplii, with LC50 values of 245.20, 261.27, and 151.40 µg∙mL⁻¹ for K. pinnata, L. camara, and M. azedarach, respectively. These results demonstrate that the plants studied are a potential source of polar chemical compounds with useful biological activity.

Academic writing presents significant challenges for beginner learners of English as a Foreign Language (EFL), where learners tend to make mistakes in grammar, coherence, and structure. Feedback is essential for the development of writing, yet in traditional instructional models it is not always provided with the quality or frequency required. This study explores the effectiveness of ChatGPT-generated feedback in enhancing academic writing skills among A2-level English as a Foreign Language learners. Involving a pre-post experimental design, 45 students participated over a fourteen-week intervention period within two groups: one receiving traditional instruction and the other using ChatGPT guided by a tailored prompt created by the principal researcher: "Evaluate this paragraph based on Cambridge A2-level standards (CEFR) for Content, Organization, and Language (vocabulary and grammar). Provide specific feedback". The results showed statistically significant gains in writing performance for the experimental group, especially in language accuracy and structural organization, with no regression in any skill area, confirming that AI feedback has the potential to facilitate language acquisition among participants at a beginner level, without compromising learner autonomy, when conducted through a structured approach. The research indicates that ChatGPT can be utilized as a pedagogical tool in addition to regular education, serving as a scalable resource with the aid of clear evaluative criteria. Future research is encouraged to expand this approach to other CEFR proficiency levels and explore learners' qualitative experiences with AI feedback. Highlights The first quasi-experimental study used structured ChatGPT prompts to improve A2-level academic writing in EFL settings. Inter-rater reliability validates the statistically significant increases in grammar, vocabulary, and organization demonstrated. Provide a feasible 12-week intervention framework that teachers can incorporate into their weekly lesson plans. Emphasize the balance between AI feedback and human mediation, supporting the pedagogical foundation. Provides implications for extending AI-assisted feedback across multiple CEFR competency levels and investigating cognitive processes involved in L2 writing

Unlike traditional centralized learning-which requires aggregating all raw data in a single repository-federated learning (FL) enables collaborative model training across distributed devices while preserving data privacy by exchanging only model parameters. This paper presents a privacy-preserving FL framework deployed on a wireless ad hoc edge network of lowcost Raspberry Pi devices. A minimal Bash script automates the configuration of the MANET topology, ensuring scalability and resilience across embedded nodes. We implemented a lightweight TCP message protocol to transmit model parameters and control signals, keeping local datasets private throughout. To validate the framework, we employed the IBM Telco Customer Churn dataset-chosen for its complexity and imbalance-though the approach generalizes to other tabular problems. We systematically evaluated four aggregation methods (FedAvg, FedProx, FedNova, and FedAvgM) under both IID and non-IID partitions, and compared them to a centralized baseline trained on the entire combined dataset collected from all clients in one place. Results show that while FedNova excelled under balanced conditions, FedAvgM was more robust to heterogeneous client distributions. These findings highlight the practicality of deploying federated neural networks on infrastructure-free edge environments, achieving competitive performance without sharing sensitive raw data.

Context. Dwarf-dwarf galaxy encounters are a key aspect of galaxy evolution as they can ignite or temporarily suppress star formation in dwarfs and can lead to dwarf mergers. However, the frequency and impact of dwarf encounters remain poorly constrained due to limitations in spectroscopic studies, such as surface-brightness incompleteness in optical studies and poor spatial resolution in single-dish neutral hydrogen (H I ) surveys. Aims. We aim to quantify the frequency of isolated gas-rich dwarf galaxy multiples using the untargeted, interferometric Apertif H I survey and study the impact of the interaction on star formation rates of galaxies as a function of the on-sky separation. Methods. Our parent dwarf sample consists of 2481 gas-rich galaxies with stellar masses in the range ∼10 6 < M ⋆ / M ⊙ ≲ 5 × 10 9 , for which we identified close companions based on projected separation ( r p ) and systemic velocity difference (|Δ V sys |). We explored both constant thresholds for r p and |Δ V sys | corresponding to 150 kpc and 150 km s −1 on all galaxies in our sample as well as mass-dependent thresholds based on a stellar-to-halo mass relation. Results. We find the average number of companions per dwarf in our sample to be 13% (20%) when considering mass-dependent (constant) thresholds. We find that the frequency (∼11.6%) of dwarf companions in the stellar mass regime of 2 × 10 8 < M ⋆ / M ⊙ < 5 × 10 9 is three times higher than previously determined from optical spectroscopic studies, highlighting the power of H I for finding dwarf multiples. Furthermore, we find evidence for an increase in star formation rates (SFRs) of close dwarf galaxy pairs of galaxies with similar stellar masses.

Abstract Accurate heavy petroleum fractions characterization is challenging when modeling hydroprocessing reactors1,2, given that predicting the hydrogen solubility in the hydrocarbon mixture is essential3,4. The continuous thermodynamics approach has been successfully used to characterize complex mixtures such as crude oils and various petroleum fractions in the liquid-vapor equilibrium prediction5,6. For this reason, this technique used in the hydrogen solubility prediction specifically for heavy petroleum fractions can contribute to improving the accuracy of the results. In this work, we present a characterizing methodology for vacuum gas oils using the continuous thermodynamics approach, implemented for the hydrogen solubility estimation by the Augmented Grayson Streed method7, achieving a reduction in the global average absolute deviation from 22.5% to 11% compared with the experimental data, under temperature and pressure ranges of 459-653 K and 1.0-12.5 MPa, respectively. Therefore, it is concluded that the characterization methodology implemented contributes to reliably estimate the hydrogen solubility in vacuum gas oils, which in turn represents a potential benefit in the design and operation of hydroprocessing reactors models.

ABSTRACT We present a comparative analysis of interstellar hydrogen (H i) and potassium (K i) absorption from the radio and optical surveys, GASKAP (Galactic Australian Square Kilometre Array Pathfinder) and GALAH (Galactic Archaeology with HERMES), to study the physical and kinematic properties of the cold interstellar medium (ISM) in the Milky Way foreground towards the Magellanic Clouds. By comparing GASKAP H i absorption with interstellar K i absorption detected in GALAH spectra of nearby stars (within 12 arcmin angular distance or a spatial separation of $\sim$0.75 pc), we reveal a strong kinematic correlation between these two tracers of the cold neutral ISM. The velocity offsets between matched H i and K i absorption components are small, with a mean (median) offset of –1.3 (–1.2) $\mathrm{km\, s^{-1}}$ and a standard deviation of 2.3 $\mathrm{km\, s^{-1}}$. The high degree of kinematic consistency suggests a close spatial association between K i and cold H i gas. Correlation analyses reveal a moderate positive relationship between H i and K i line-of-sight properties, such as K i column density with H i column density or H i brightness temperature. We observe a $\sim$63 per cent overlap in the detection of both species towards 290 (out of 462) GASKAP H i absorption lines of sight, and estimate a median K i/H i abundance ratio of $\sim 2.3\times 10^{-10}$, in excellent agreement with previous findings. Our work opens up an exciting avenue of Galactic research that uses large-scale surveys in the radio and optical wavelengths to probe the neutral ISM through its diverse tracers.

Diabetes is a chronic metabolic disorder characterized by persistent hyperglycemia. The globally rising prevalence of diabetes has made early and accurate diagnosis imperative to avoid long-term sequelae and decrease the cost burden on health facilities. Machine Learning (ML) has emerged as a highly effective tool of clinical science because of its capability of discovering complex patient data patterns and diagnostic performance optimization. This paper is a comparative performance evaluation of five ML models—Decision Tree (DT), Random Forest (RF), Logistic Regression (LR), K-Nearest Neighbors (KNN), and Multilayer Perceptron (MLP)—utilized for diabetes prediction. Data were processed using the “Healthcare Diabetes Dataset,” made up of eight commonly used clinical parameters. For making the models trustworthy, a strong data preprocessing pipeline was utilized, made up of outlier detection using the Interquartile Range (IQR), normalization of data, and class balancing using the Synthetic Minority Over-sampling Technique (SMOTE). Results reveal that the RF and DT models achieved the highest performance based on accuracy rates of 98.15% and 97.51%, respectively, though more moderate outcomes were recorded by LR and MLP. They reveal the remarkable potential of ML models, particularly ensemble-based ML models such as RF, at supporting early diagnosis of diabetes. When implemented complementarily to clinical decision-making processes, these models can serve a cost-effective and effective replacement of conventional diagnostic methods.

Water is essential for active life, yet some organisms, such as tardigrades, can survive prolonged periods of drying‐induced dormancy. Cytoplasmic abundant heat‐soluble (CAHS) proteins are disordered proteins that undergo a phase transition from the solution to gel state. CAHS proteins help tardigrades survive extreme drying, increase hyperosmotic stress tolerance in heterologous systems, and preserve the function of labile enzymes during drying in vitro. It has been speculated that the ability of CAHS proteins to form gels might be mechanistically linked to their protective capacity. However, recent evidence suggests that while gelation enhances hyperosmotic stress tolerance, it is not required for this phenomenon. Still, the extent to which gelation is necessary for other CAHS‐based protective functions, such as enzyme protection during drying, is unknown. Here, we show that rather than the solution or gel state of CAHS proteins being the sole protective phase, each phase is optimized to protect different enzymes during drying. Using in vitro assays that provide clear functional readouts and allow for precise control over CAHS and client enzyme ratios, we show that the gelled state of CAHS D, a model CAHS protein, promotes the protection of the enzyme lactate dehydrogenase during drying. We find that the opposite is true for the enzyme citrate synthase, with variants of CAHS D that do not gel providing optimal protection to this enzyme. Correlative analysis between protective capacity and sequence/ensemble features of CAHS D variants supports the notion that phase is a major driver of differential enzyme protection. Finally, we show that enhanced water binding is an emergent property of gelation that positively correlates with the protein's ability to protect LDH. These results demonstrate a link between the phase of CAHS proteins and their protective function, providing insights into how CAHS proteins help tardigrades counteract the spectrum of stresses encountered during different stages of drying. Broadly, this study advances our understanding of desiccation tolerance, while providing insights into engineering strategies to tune protein‐based excipients to protect specific clients. This study contributes to a broader discussion in the protein field about the functionality of phase behavior and states.