Search Results (45)

Context: Modern software systems have become increasingly complex, making it difficult to interpret raw requirements and effectively utilize traditional tools for software design and analysis. Unified Modeling Language (UML) class diagrams are widely used to visualize and understand system architecture, but analyzing them manually, especially for large-scale systems, poses significant challenges. Objectives: This study aims to automate the analysis of UML class diagrams by assessing their complexity using a machine learning approach. The goal is to support software developers in identifying potential design issues early in the development process and to improve overall software quality. Methodology: To achieve this, this research introduces a Visual Question Answering (VQA)-based framework that integrates both computer vision and natural language processing. Vision Transformers (ViTs) are employed to extract global visual features from UML class diagrams, while the BERT language model processes natural language queries. By combining these two models, the system can accurately respond to questions related to software complexity, such as class coupling and inheritance depth. Results: The proposed method demonstrated strong performance in experimental trials. The ViT model achieved an accuracy of 0.8800, with both the F1 score and recall reaching 0.8985. These metrics highlight the effectiveness of the approach in automatically evaluating UML class diagrams. Conclusions: The findings confirm that advanced machine learning techniques can be successfully applied to automate software design analysis. This approach can help developers detect design flaws early and enhance software maintainability. Future work will explore advanced fusion strategies, novel data augmentation techniques, and lightweight model adaptations suitable for environments with limited computational resources. Full article

Clinical notes often contain unstructured text filled with abbreviations, non-standard terminology, and inconsistent phrasing, which pose significant challenges for automated medical information extraction. Named Entity Recognition (NER) plays a crucial role in structuring this data by identifying and categorizing key clinical entities such as symptoms, medications, and diagnoses. However, traditional and even transformer-based NER models often struggle with ambiguity and fail to produce clinically interpretable outputs. In this study, we present a hybrid two-stage framework that enhances medical NER by integrating a fine-tuned BERT model for initial entity extraction with a Dictionary-Infused Retrieval-Augmented Generation (DiRAG) module for terminology normalization. Our approach addresses two critical limitations in current clinical NER systems: lack of contextual clarity and inconsistent standardization of medical terms. The DiRAG module combines semantic retrieval from a UMLS-based vector database with lexical matching and prompt-based generation using a large language model, ensuring precise and explainable normalization of ambiguous entities. The fine-tuned BERT model achieved an F1 score of 0.708 on the MACCROBAT dataset, outperforming several domain-specific baselines, including BioBERT and ClinicalBERT. The integration of the DiRAG module further improved the interpretability and clinical relevance of the extracted entities. Through qualitative case studies, we demonstrate that our framework not only enhances clarity but also mitigates common issues such as abbreviation ambiguity and terminology inconsistency. Full article

Common bean (Phaseolus vulgaris L.) is a critical protein-rich legume supporting food and nutritional security globally. However, Fusarium wilt, caused by Fusarium solani, remains a major constraint to production, with yield losses reaching up to 84%. While biocontrol strategies have been explored, most microbial agents are sourced from mesophilic environments and show limited effectiveness under abiotic stress. Here, we report the isolation and characterization of extremophilic Bacillus spp. from the hypersaline Lake Bogoria, Kenya, and their biocontrol potential against F. solani. From 30 isolates obtained via serial dilution, 9 exhibited antagonistic activity in vitro, with mycelial inhibition ranging from 1.07–1.93 cm 16S rRNA sequencing revealed taxonomic diversity within the Bacillus genus, including unique extremotolerant strains. Molecular screening identified genes associated with the biosynthesis of antifungal metabolites such as 2,4-diacetylphloroglucinol, pyrrolnitrin, and hydrogen cyanide. Enzyme assays confirmed substantial production of chitinase (1.33–3160 U/mL) and chitosanase (10.62–28.33 mm), supporting a cell wall-targeted antagonism mechanism. In planta assays with the lead isolate (B7) significantly reduced disease incidence (8–35%) and wilt severity (1–5 affected plants), while enhancing root colonization under pathogen pressure. These findings demonstrate that extremophile-derived Bacillus spp. possess robust antifungal traits and highlight their potential as climate-resilient biocontrol agents for sustainable bean production in arid and semi-arid agroecosystems. Full article

The underlying structural features of the mismatch between catalytic and cytostatic properties in L-asparaginase from Rhodospirillum rubrum (RrA) and three of its mutants were investigated. The rationale for selecting the specific mutations (RrA_{A64V, E67K}; RrA_{R118H, G120R}; RrA_{E149R, V150P, F151T}) is to elucidate the role of inter-subunit interaction in RrA and its impact on catalytic efficiency and stability. Bioinformatic modeling revealed a predominantly negative surface charge on RrA with limited positive charge clusters in the vicinity of the interface region. Thus, some negatively charged groups were replaced with positively charged ones to enhance the electrostatic interactions and stabilize the enzyme quaternary structure. RrA_{A64V, E67K} and RrA_{R118H, G120R} additionally contained an N-terminal 17-amino acid capsid peptide derived from the bacteriophage T7 (MASMTGGQQMGRGSSRQ), which could potentially affect the conformational stability of theenzymes. Circular dichroism (CD) spectroscopy was applied to the kinetic parameters analysis of Asn hydrolysis and showed that native RrA displayed a V_max of 30 U/mg and a K_M of 4.5 ± 0.5 mM. RrA_{E149R, V150P, and F151T} exhibited a substantially increased V_max of 57 U/mg. The catalytic efficiency of V_max/K_M also improved compared to the native enzyme: the V_max/K_M increased from approximately 7 U/mg × mM⁻¹ (for the native enzyme) to 9 U/mg × mM⁻¹ for Mut3. Other mutants exhibited less pronounced changes. Thermo-denaturation studies allowed us to determine the phase transition parameters of the RrA variants in comparison with commercial reference sample EcA. RrA_{A64V, E67K} and RrA_{R118H, G120R} exhibited the most favorable phase transition parameters, with melting temperatures (T_m) of 60.3 °C and 59.4 °C, respectively, exceeding that of the wild-type RrA (54.6 °C) and RrA_{E149R, V150P, F151T} (52 °C). The EcA demonstrated a slightly superior thermal stability, with a T_m of 62 °C. The mutations showed a significant effect on protein stability during trypsinolysis. Therefore, RrA_{E149R, V150P, F151T} showed higher resistance (45% activity remaining after 30 min of trypsin exposure) compared to the native RrA retained 20% activity. EcA preparations exhibited lower stability to trypsinolysis (losing over 90% activity in 15 min). The cytostatic effects were evaluated using MTT assays against K562 (leukemic) and A549 (lung carcinoma) cell lines. The MTT assays with K562 cells revealed that RrA_{E149R, V150P, F151T} (IC₅₀ of 10 U/mL) and RrA_{R118H, G120R} (IC₅₀ of 11.5 U/mL) exhibited superior antiproliferative activity compared to native enzymes RrA (IC₅₀ of 15 U/mL) and EcA (24 U/mL). RrA_{E149R, V150P, F151T} showed the most significant improvement in cytostatic activity. The results obtained indicate that the substitutions in RrA_{E149R, V150P, F151T} resulted in the improvement of the enzyme biocatalytic properties and an increase in the resistance to aggregation and trypsinolysis. This highlights the role of electrostatic interactions in stabilizing the oligomeric structure of the enzyme, which eventually translates into an improvement in cytostatic efficiency and antiproliferative forces. Full article

Understanding and interpreting medical images, particularly radiology images, is a time-consuming task that requires specialized expertise. In this study, we developed a deep learning-based system capable of automatically assigning multiple standardized medical concepts to radiology images, leveraging deep learning models. These concepts are based on Unified Medical Language System (UMLS) Concept Unique Identifiers (CUIs) and describe the radiology images in detail. Each image is associated with multiple concepts, making it a multilabel classification problem. We implemented several deep learning models, including DenseNet121, ResNet101, and VGG19, and evaluated them on the ImageCLEF 2020 Medical Concept Detection dataset. This dataset consists of radiology images with multiple CUIs associated with each image and is organized into seven categories based on their modality information. In this study, transfer learning techniques were applied, with the models initially pre-trained on the ImageNet dataset and subsequently fine-tuned on the ImageCLEF dataset. We present the evaluation results based on the F1-score metric, demonstrating the effectiveness of our approach. Our best-performing model, DenseNet121, achieved an F1-score of 0.89 on the classification of the twenty most frequent medical concepts, indicating a significant improvement over baseline methods. Full article

This study optimized operating parameters to enhance cellulase production and evaluated scale-up feasibility in submerged fermentation (SmF) using Trichoderma sp. KMF006. Flask-scale experiments assessed the effects of Avicel:cellulose ratios (4:0–0:4), agitation speeds (150–210 rpm), and turbulence (baffled vs. non-baffled flasks), with optimized conditions applied to a 10 L bioreactor. A 3:1 Avicel:cellulose ratio (A3C1) significantly accelerated cellulase production, reaching peak activity 6 days earlier than Avicel alone. An agitation speed of 180 rpm was optimal, balancing enzyme activity and energy efficiency. Turbulence enhanced cellulase yields, with baffled flasks increasing EG, BGL, and CBH activities 19.9-, 6.2-, and 8.9-fold, respectively, compared to the control. Biochar further improved cellulase production but only under turbulent conditions, demonstrating a synergistic effect. At the bioreactor scale, the A3-180_Imp (A3C1, 180 rpm, impeller-induced turbulence) achieved the highest enzymatic activity (33.60 U/mL EG, 3.46 U/mL BGL, and 0.63 U/mL CBH). The filter paper unit (FPU) was 84 FPU/mL, a two-fold increase compared to the control. However, excessive turbulence at 210 rpm reduced enzyme stability, emphasizing the importance of balancing shear stress. These findings provide a systematic framework for optimizing SmF conditions, highlighting the significance of balancing hydrodynamic conditions for efficient cellulase production at an industrial scale. Full article

Myocardial infarction (MI) is defined as a clinical event in which myocardial damage is evidenced in the setting of myocardial ischemia. However, patients without occlusive coronary artery stenosis can also have myocardial infarction, which is titled Myocardial Infarction with Non-Obstructive Coronary Arteries (MINOCA). In our study, we aimed to evaluate oxidative stress and inflammation responses between MINOCA and MI with coronary artery disease (CAD) patients. In this prospective, cross-sectional study, patients with elevated cardiac markers who were admitted to the cardiology clinic between March 2024 and May 2024 with the preliminary diagnosis of acute coronary syndrome were included. Patients were consecutively collected as those with an occlusive lesion on coronary angiography and those without. Routine blood samples and oxidative stress parameters were obtained and compared between groups. A total of 88 patients, including 44 MINOCA and 44 MI-CAD patients, were included in the study. The MINOCA group was significantly younger than the MI-CAD group (56.2 ± 12.5, vs. 64.7 ± 9.3, p: 0.001). While inflammatory parameters were similar between groups, dityrosine (5708 FU/mL (5311–6417) vs. 4488 FU/mL (3641–5238), p < 0.001), lipid hydroperoxide (3.6 nmol/mL (3.4–3.9) vs. 3.4 nmol/mL (3.1–3.9), p: 0.023), kynurenine (3814 ± 621 FU/mL vs. 3319 ± 680 FU/mL, p: 0.001), and malondialdehyde (17.4 nmol/mL (13.7–19.1) vs. 13.1 nmol/mL (12–14.9), p < 0.001) levels were higher in the MI-CAD group than in the MINOCA group. Although inflammation parameters did not differ between MI-CAD and MINOCA patients, oxidative stress parameters were higher in the MI-CAD group. Regardless of the presence and severity of inflammation, oxidative markers can help to assess the level of myocardial cell damage, risk stratification, and diagnosis of myocardial infarction. Full article

Cold-adapted pectin lyases are particularly useful in the extraction and clarification of freshly squeezed fruit juices at low temperatures, as they effectively reduce juice viscosity and improve light transmittance. With the increasing attention on low-temperature pectinase in industrial applications, the exploration of low-temperature pectinase with novel characteristics has become one of the key focuses of research and development. In this study, a 1026 bp gene, pel1Ba, encoding a 42.7 kDa pectin lyase, was cloned from sediment samples collected from the South China Sea and heterologously expressed in Escherichia coli. The purified Pel1Ba exhibited an optimal temperature of 40 °C and an optimal pH of 10, with a total enzyme activity of 5100 U/mL. Notably, Pel1Ba is a cold-adapted enzyme that retains 80% of its relative activity across the temperature range of 0–40 °C. When 20 U/mL purified Pel1Ba was added to orange juice, the juice volume increased by 43.00% and its clarity improved by 37.80%. Meanwhile, site-directed mutagenesis analysis revealed that the residual enzyme activities of the mutants A230I, F253I, and L292I were increased by 22.5%, 34.4%, and 25.1%, respectively, compared to the wild type. This study concludes that the cold-active pectate lyase Pel1Ba exhibits potential for applications in the food industry. Full article

Nattokinase (NK), a serine protease with high thrombolytic activity, has significant potential for application in foods intended for special health benefits. However, the NK production in wild-type Bacillus subtilis natto is relatively low. In this study, a high-yielding NK and genetically stable mutant strain (B. subtilis JNC002.001, 300.0 ± 4.7 FU/mL) was obtained through atmospheric and room temperature plasma (ARTP) mutagenesis. It increased NK activity by 1.84 times compared to the initial strain SD2, demonstrating significant prospects for NK production and food fermentation applications. Additionally, the B. subtilis JNC002.001 exhibited notable alterations in growth characteristics, glucose consumption, and sporulation. This study further elucidated the mechanism of enhanced NK production at the molecular level. Genome resequencing revealed that the mutant genes in JNC002.001 included 10 single nucleotide polymorphisms (SNPs) and one insertion, among which the kinA and gltA genes were associated with sporulation and NK synthesis, respectively. In terms of the transcriptional level, the NK-coding gene aprN was up-regulated 9.4 times relative to the wild-type strain. Most of the genes related to central carbon metabolism and the Sec secretion pathway were up-regulated. In addition, the expression of regulatory factors associated with the transcription of the aprN gene and the sporulation process provided evidence for high NK expression and sporulation deficiency in JNC002.001. These results could provide insights into the mechanism of NK production and facilitate the construction of engineered strains with high NK yield. Full article

Globally, an estimated 254 million people are living with chronic hepatitis B virus (HBV) infection, yet only 10.5% have been diagnosed, underscoring the urgent need to expand testing to meet the World Health Organization’s HBV elimination targets by 2030. Many HBV diagnostic tests remain expensive and inaccessible in resource-limited settings. In this study, we demonstrate how individually sourced, commercially available reagents can be used to develop cost-effective in-house assays for total DNA isolation, HBV viral load quantification by (q)PCR, and qHBsAg and qHBeAg measurement using sandwich ELISA. These assays were validated using known HBV-positive and HBV-negative plasma samples (genotypes A–F) and HepAD38 cells treated with tenofovir disoproxil fumarate (TDF). DNA isolation using a commercial column-based kit was compared to a high-throughput, column-free method, allowing for HBV quantification from 50 µL of plasma with lower limits of detection (LLOD) of 1.8 × 10³ and 1.8 × 10⁴ HBV DNA copies IU/mL, respectively. Both commercial and in-house DNA isolation methods yielded comparable half-maximal effective concentration (EC₅₀) values in TDF-treated HepAD38 cells. Additionally, in-house sandwich ELISA assays were developed for quantitative HBsAg and HBeAg detection, with LLOD values of 0.78 IU/mL and 0.38 PEI U/mL (Paul Ehrlich Institute), respectively. The in-house reagents for DNA isolation, molecular testing, and serological detection of HBV were estimated to be at least 10 times more cost-effective than commercially available kits, highlighting their potential for broader application in resource-limited regions. Full article

This work aimed to optimize the production of L-asparaginase (L-ASNase) from Aspergillus caespitosus CCDCA 11593 using Pereskia aculeata (Ora-pro-nóbis) leaf fiber as a substrate for solid-state fermentation (SSF), along with powdered whey protein as a substrate in submerged fermentation (SmF) processes. A centered face design was applied to evaluate the effect of the different parameters. Additionally, L-ASNase was partially purified on an ion-exchange cryogel column. For SSF, the experimental condition, inoculum concentration 10⁵ spores/mL, 120 h at 25 °C, 14% of substrate, and 1% of asparagine, corresponded to the highest enzymatic activity (2.75 U/mL) of L-ASNase. For SmF, the experimental condition of greater enzymatic activity (1.49 U/mL) was obtained in the medium containing 16% to 24% asparagine, 3.3% to 4.7% substrate, spore concentration of 7 × 10⁶ to 10⁷ spores/mL, temperature range of 29.8 to 34.8 °C, pH range of 5.7 to 6.3, and 87 to 105 h of fermentation. The L-ASNase obtained from SmF was subjected to adsorption tests, resulting in 4.4 U/mg of partially purified enzyme. This study suggested that whey protein and Ora-pro-nóbis leaf fiber could be a low-cost substrate for L-ASNase production. Additionally, using an ion-exchange cryogel column for enzyme purification holds promise for sustainable applications in the clinical and food industries. Full article

The Fusarium oxysporum species commonly found in soil include plant and human pathogens, and nonpathogenic species. F. oxysporum grown on lignocellulosic substrates under submerged conditions produces an extracellular enzyme profile with hemicellulolytic and cellulolytic activities. Our aim was to produce an extract of Fusarium oxysporum f. sp. melonis with β-1,4-xylanase activity after fermentation on a Brewers’ spent grain (BSG)-containing substrate. We prepared the BSG substrate, with or without sonication, for the submerged fermentation of Fusarium oxysporum previously isolated from local soil and preserved at 4 °C. First, an enriched inoculum was prepared, and later, the production of β-1,4-xylanase using the BSG substrates was monitored for up to 6 or 10 days in the enriched inoculum or in the enzyme extract, respectively. An activity of β-1,4-xylanase 12.0 U/mL (day 3) was obtained in the enriched inoculum with the untreated BSG, remaining constant for 3 days. A significant increase in the activity of this enzyme was observed (day 6), especially in the extract obtained using the sonicated BSG substrate (39 U/mL). Applying ultrasound to the BSG before its use in a submerged fermentation with Fusarium oxysporum f. sp. melonis could be an alternative for producing β-1,4-xylanase. Full article

Pancreatic cancer (PC) is the seventh most common cause of cancer-related death worldwide. The low survival rate may be due to late diagnosis and asymptomatic early-stage disease. Most patients are diagnosed at an advanced stage of the disease. The search for novel prognostic factors is still needed. Two miRNAs, miR-22-3p and miR-885-5p, which show increased expression in PC, were selected for this study. The aim of this study was to evaluate the utility of these miRNAs in the prognosis of PC. Other prognostic factors such as lipase-to-amylase ratio (LAR), neutrophil-to-lymphocyte ratio (NLR), and carbohydrate antigen 19-9 (CA19-9) were also evaluated in this study. This study was conducted in 50 patients previously diagnosed with pancreatic ductal adenocarcinoma in clinical stage (CS) III and IV. All patients underwent a complete medical history, physical examination, and routine laboratory tests including a complete blood count, C-reactive protein (CRP), CA19-9, lipase, and amylase. Two additional blood samples were taken from each patient to separate plasma and serum. Isolation of miRNA was performed using TRI reagent with cel-miR-39-3p as a spike-in control. Reverse transcription of miRNA was performed using a TaqMan Advanced miRNA cDNA Synthesis Kit. The relative expression levels of miR-22-3p and miR-885-5p were measured using RT-qPCR. Serum hsa-miR-22-3p was detected in 22 cases (44%), while hsa-miR-885-5p was detected in 33 cases (66%). There were no statistically significant differences in serum or plasma miRNA expression levels between patient groups based on clinical stage, gender, or BMI. There were no statistically significant differences in LAR between patients with different CS. For NLR, CRP and CA19-9 thresholds were determined using ROC analysis (6.63, 24.7 mg/L and 4691 U/mL, respectively). Cox’s F test for overall survival showed statistically significant differences between groups (p = 0.002 for NLR, p = 0.007 for CRP and p = 0.007 for CA19-9). Utility as prognostic biomarkers was confirmed in univariate and multivariate analysis for CA19-9, CRP, and NLR. The selected miRNAs and LAR were not confirmed as reliable prognostic markers in PC. Full article

This study investigated the psycho-physiological impact of a black belt examination. Older brown-belt judoka (Tori, F = 2, M = 4; age = 75.6 ± 4.5 yrs) and their 2nd–5th Dan black-belt coaches (Uke; M = 6; age = 36.5 ± 10.8 yr) were evaluated during a simulated and official examination and a resting day. Participants’ trait anxiety (STAI-Y2) was recorded prior to the study. State anxiety (STAI-Y1), ratings of perceived exertion (RPE), enjoyment (ENJ), and fear of falling (FoF) were collected 15 min before and after the experimental conditions. Saliva samplings at awakening (T0), PRE (T1), and POST (T2) exercise and during the recovery (15 min-T3, 30 min-T4, 60 min-T5) were collected for cortisol (sC), testosterone (sT), and alpha-amylase (sAA). Participants showed normal age-reference population trait anxiety. A difference (p ≤ 0.05) for role emerged for ENJ and sT only. For STAI-Y1, higher PRE values with respect to POST ones emerged (p = 0.005), and the highest values (p = 0.007) for PRE of the examination were with respect to the simulation. For sAA, differences for sampling were found in the examination conditions only, with peak values at T2 (370.3 ± 78.6 U/mL, p = 0.001). For sC, a significant peak value (0.51 ± 0.09 μg/dL, p = 0.012) emerged at T2 in the examination condition. With respect to Tori, Uke showed higher mean sT values in all conditions (p ≤ 0.05) and the highest T2 during examination (712.5 ± 57.2 pg/mL). Findings suggest the relevance of monitoring psycho-physiological stress-related responses in judo for optimizing both coaching effectiveness and sport performance, especially in older judo practitioners. Full article

This article investigates the application of a deep learning model for evaluating the conformity of model images to types of UML diagrams to be used in self-training and educational settings. Our approach leans on a feature-based dataset that captures a broad range of modeling elements from class, state machine, and sequence diagrams, enhancing the ability to recognize a larger variety of categories selected for this research. The neural network trained with these features representing parts of the UML concrete syntax demonstrates 90% in classification accuracy on average, in respect to our previous research on UML diagrams classification without using a feature-based dataset. This study concludes that a feature-based approach, combined with advanced neural network architectures, can improve the classification of such images, especially in edge cases where diagrams contain similar graphical details but the whole does not represent a UML diagram. For the given research, we obtained a 0.87 F1 score. Full article