University of Bari Aldo Moro
Archivio istituzionale della ricerca - Università di BariNot a member yet
151277 research outputs found
Sort by
UKANWeed: An Application of Kolmogorov-Arnold Networks to Weed Mapping
No full textEffective weed monitoring is a critical component of precision agriculture, directly impacting crop health and yield. This study presents UKANWeed, an application-specific adaptation of the recently proposed UKAN architecture, designed for image segmentation in agricultural settings. UKANWeed integrates Kolmogorov-Arnold Networks (KAN) into its layers, resulting in a lightweight yet highly efficient model that can distinguish weeds from crops. Evaluated on the WeedMap dataset, UKANWeed achieves an F1-score of 86.3, outperforming the widely used UNet architecture while requiring significantly fewer parameters. Additionally, we investigated the behavior of UKANWeed and UNet under extreme model compression, revealing a lower bound in the representational capacity below which task performance degrades sharply. The compactness and accuracy of UKANWeed make it suitable for deployment on edge devices such as drones, enabling real-time, in-field weed detection and crop monitoring—an essential step toward scalable, autonomous precision agriculture. The code is available at https://github.com/pasqualedem/UKANWeed
Sex disparities in tuberculosis outcomes: evidence from a multicenter Italian cohort (Italian South TB Network (ISTB‐Net)
No full textAbstract
Background Sex disparities in tuberculosis (TB) outcomes are not well characterized, especially in high-income countries where social vulnerability and migration influence access to care. Although men globally experience a higher TB burden, the interaction between sex, migration, and social determinants is complex and extends beyond biological factors. This study evaluated sex differences in clinical and programmatic TB outcomes in a high-income European country with a significant substantial migrant population.
Methods A retrospective multicentre cohort study was conducted across 16 Infectious Diseases Units in seven Italian regions from (January 2021 to September 2025). Outcomes included time to sputum conversion (in pulmonary TB), length of hospital stay (LOS), adverse events (AEs) and their severity, incomplete treatment (defined as failure, death, or loss to follow-up), and loss to follow-up (LTFU). Mixed-effects models were applied using two prespecified adjustment sets: sex, centre, and core confounders (Model A); and sex, centre, and clinically relevant baseline imbalances (Model B). Sub-analyses examined the impact of migration status.
Results Of 982 TB patients, 229 (23.3%) were women and 753 (76.7%) were men. Women exhibited lower rates of smoking (24.4% vs 36.7%), diabetes (7.9% vs 15.8%), and COPD/bronchiectasis (4.5% vs 10.3%). The median sputum conversion time was 21 days for both sexes. Adjusted analysesindicated shorter LOS among women (Model A: − 22% [95%CI − 32 to − 10]; Model B: − 19% [95%CI − 28 to − 9]). Time to sputum conversion was slightly shorter in women in Model A (− 13%; 95%CI −23% to −1%) but not in Model B (− 9%; 95%CI −17% to 1%). The risk and severity of AEs were similar between sexes. In Model B, women had lower odds of incomplete treatment (OR 0.64 [95%CI 0.41 to 0.99]) and LTFU (OR 0.62 [95%CI 0.38 to 0.99]). Migrants experienced worse overall outcomes, but the effect of sex did not differ by migration status. Conclusion Women had consistently shorter hospital stays and greater treatment continuity without increased toxicity, indicating that sex differences in TB outcomes are likely attributable to social and behavioural factors rather than biological differences. Supportive associative networks and non-governmental organisations may help reduce sex disparities, under- scoring the importance of sex- and migration-responsive TB care models in Europe
Liposome-metal nanoparticle based sensing systems for (bio)analytical applications
No full textLiposome-metal nanoparticle (MeNP) hybrids have emerged as promising platforms for biosensing due to their low toxicity, enhanced stability, and ability to improve selectivity and signal amplification. This review comprehensively explores the state-of-the-art applications of these hybrid systems in the (bio)analytical domain. Depending on the particular bioassay, MeNPs can be strategically positioned within liposomes in three distinct regions: encapsulated in the aqueous core, embedded in the lipid bilayer, or attached to the phospholipid membrane surface. These configurations enable MeNP-liposome hybrids to operate as (i) signal-generating labels, (ii) carriers for bioreagents, (iii) entities for target sequestration or interference mitigation, and (iv) signal readout amplification. We delve into various analytical applications based on the signal transduction system, including electrochemistry, fluorescence, colorimetry, electrochemiluminescence, photoelectrochemistry, surface plasmon resonance, and surface-enhanced Raman spectroscopy. Representative examples from the last decade illustrate the diverse and innovative uses of these composite materials in the field of biosensing
Triazole-benzodiazepine derivatives: One-pot synthesis, characterization, hirshfeld surface analysis, and computational insights into anticancer potential as KIF11 inhibitors
No full textIn this study a [3 + 2] cycloaddition reaction of the benzodiazepine BZD1 with N-aryl-C-ethoxycarbonylnitrilimines 3(a-b) was explored. The reaction was carried out in a basic medium to obtain new triazole-benzodiazepine derivatives 4(a-b) with improved selectivity. Structural characterization was performed using 1H, 13C NMR spectroscopy, as well as X-ray diffraction analysis. A comprehensive theoretical study was conducted to elucidate the electronic structure and reactivity of two newly synthesized triazole-benzodiazepine derivatives (4a and 4b). Frontier Molecular Orbital (FMO), Electrostatic Potential (ESP), Fukui function, and global descriptor analyses consistently revealed that the triazole-benzodiazepine core drives charge transfer, while the ester and triazole groups are the main sites of nucleophilic and electrophilic attacks. According to Hirshfeld surface analysis, there was no significant π–π stacking interactions were observed, and weak hydrogen bonding between oxygen and nitrogen atoms play a critical directing function, even if dispersion forces (H···H contacts) predominate. Drug similarity and in silico ADMET studies indicated that both compounds exhibit high predicted oral bioavailability, blood-brain barrier crossing, and good gastrointestinal absorption all of which suggest that the central nervous system may be activated. Docking simulation results demonstrated that both compounds represent promising prospects for the development of oral bioavailable KIF11 inhibitors with potential anticancer effects
On the use of the principle of maximum entropy to improve the robustness of bivariate spline least-squares approximation
No full textWe consider fitting a bivariate spline regression model to data using a weighted least-squares cost function, with weights that sum to one to form a discrete probability distribution. By applying the principle of maximum entropy, the weight distribution is determined by maximizing the associated entropy function. This approach, previously applied successfully to polynomials and spline curves, enhances the robustness of the regression model by automatically detecting and down-weighting anomalous data during the fitting process. To demonstrate the effectiveness of the method, we present applications to two image processing problems and further illustrate its potential through two synthetic examples. Unlike the standard ordinary least-squares method, the maximum entropy formulation leads to a nonlinear algebraic system whose solvability requires careful theoretical analysis. We provide preliminary results in this direction and discuss the computational implications of solving the associated constrained optimization problem, which calls for dedicated iterative algorithms. These aspects suggest natural directions for further research on both the theoretical and algorithmic fronts
Assessment of Bottom Trawl Impacts on the Status of Seabed Communities in European Seas
No full textBottom trawling affects seabed habitats, but its large- scale impacts remain poorly quantified. Assessment of trawling impacts is essential to support monitoring and achieving sustainability objectives under international conventions, sustainable development goals, and seafood certification programs. We present a Europe- wide quantitative assessment of bottom trawling impacts, accounting for regional seabed- community sensitivity drivers, across the Baltic, Atlantic, Mediterranean and Black Sea continental shelves. Using two risk-based indicators of seabed status—Relative Benthic Status determined as benthic community biomass relative to seabed fauna carrying capacity (RBStot) and RBSsen (biomass of the 10% most sensitive fauna relative to carrying capacity)—we found substantial regional and habitat differences. The Black, Baltic and Aegean-Levantine Seas showed low trawling intensity and high seabed status across habitats. In contrast, the Western Mediterranean, Ionian and Central Mediterranean and Adriatic Seas were the most severely impacted. Trawling affected the sensitive species biomass fraction more strongly than the total community biomass. RBStot was in good condition (here chosen as RBS > 75% for epifauna) for over 79% of habitat-ecoregion combinations. In contrast, RBSsen met this threshold in only 46% of these. A strong correlation emerged between the mean trawling intensity and RBStot and RBSsen, allowing the use of SAR to estimate ecosystem status. This relationship can support decisions on where, and by how much, SAR reductions are needed to achieve good environmental status in regions where no detailed assessment is available. Our approach provides a quantitative framework to balance fishery production with ecosystem sustainability, offering tools for environmental and fisheries management in Europe
Brain vascular stability relies on PAK2-cilia-PDGF-BB-HSPGs on basolateral side of endothelium
No full text: Endothelial cells (ECs) in the brain communicate with mural cells to facilitate vascular stability. Platelet-derived growth factor-BB (PDGF-BB)/platelet-derived growth factor receptor-β (PDGFR-β) signaling mechanism at EC-mural cell interface helps stabilize the vasculature. How this paracrine signaling is mediated is not known. Our laboratory studies endothelial cilia, a microtubule-based organelle, and its role in promoting vascular stability. We discovered that brain endothelial cilia are located primarily on the basolateral side, and PDGF-BB is expressed in EC cilium. Thus, we hypothesized that endothelium cilium in conjunction with PDGF-BB on the basolateral side is responsible for mural cell recruitment. In this study, using a combination of zebrafish, mice, and human brain model systems, we have established a signaling paradigm wherein p21-activated kinase (PAK2) and ADP-ribosylation factor-13b (ARL13b) in ECs induce secretion of PDGF-BB. PDGF-BB associates with heparan sulfate proteoglycans (HSPGs) to form a gradient around ECs. Disrupting PAK2 affects ciliogenesis, HSPGs, and PDGF-BB gradient. We unravel a new mechanism involving endothelial cilia/PAK2-mediated PDGF-BB secretion, and retention by periendothelial HSPGs to promote vascular stability via recruiting mural cells
Influence of energy supplementation on mitigating energy deficit and enhancing dairy performance in Holstein Friesian cows
No full textThis work sought to investigate the effects of different dosages of energy supplements (XE) on energy balance, dairy performance, and blood metabolites in Holstein Friesian cows. Ninety nursing Holstein Friesian cows were randomly distributed to three groups: Control (C), low dosage (0.1 kg/cow/day, XE1), and high dose (0.2 kg/cow/day, XE2). Over the course of one week, known as the acclimatization period, we gradually exposed the cows to the extra food. Using XE components glycerol and propylene glycol made via fermentation with single-cell protein, the treatment ran for eight weeks. Milk composition and its production, along with blood metabolites, were evaluate. The XE2 group significantly (P < 0.05) improved total digestible nutrients and net energy versus control group. Compared to other groups, cows in the XE2 group produced the most milk during the weeks; their relative increase from 29.5 L/day in week 1 to 31.5 L/day in week 8 is statistically significant (P < 0.05). Milk fat and protein notably (P < 0.05) increased in the high-dose group (XE2) by roughly 14% and 9%, respectively, versus control cows. Blood metabolites like beta-hydroxybutyrate, non-esterified fatty acids went down (P < 0.05), and urea nitrogen was nonsignificant, even though insulin and blood glucose levels went up, which means metabolic health and energy balance got better. Particularly in cows with outstanding production potential, the study found that introducing energy supplementation into dairy diets improved energy balance, milk yield, milk composition, and blood metabolites
A Weakly-Supervised Learning Approach for RGB Crop Detection Using UAV Imagery
No full textPrecision agriculture increasingly relies on accurate crop monitoring to optimize yields and resource use, yet faces significant challenges due to the scarcity of labeled data and the high cost of advanced imaging technologies. To address these limitations, we propose a novel weakly-supervised learning approach for crop detection in aerial RGB imagery, in which pseudo-labels are automatically generated through zero-shot segmentation—thus minimizing the need for manual annotation. Our pipeline combines the Segment Anything Model for zero-shot segmentation, DBSCAN for clustering and label inference, and Faster R-CNN with a ResNet-101 backbone for object detection. This strategy enables effective crop detection even in scenarios with little or no human supervision, offering a scalable and cost-efficient solution for diverse agricultural contexts. We evaluate our approach using a newly collected dataset of drone-based RGB images, which comprises vineyards, orchards, olive groves, and wheat fields. Experimental results demonstrate high precision, recall, and F1 scores across crop types, validating the robustness and applicability of the proposed method in real-world agricultural environments. Our findings highlight the potential of integrating modern self-supervision techniques with object detection frameworks to enhance sustainable and data-efficient precision farming
Balancing carbon footprint and algorithm performance in recommender systems: A comprehensive benchmark
No full textIn this paper, we present a reproducible pipeline to benchmark the trade-off between carbon emissions and recommendation performance across 14 algorithms and three publicly available datasets. In particular, we contribute: (a) a standardized protocol to account for carbon emissions of recommendation algorithms; (b) an empirical quantification of the carbon cost of hyperparameter tuning, and (c) an evaluation of data-reduction strategies as a low-cost approach to reduce emissions while improving certain non-accuracy metrics. Unlike previous literature, which mainly focused on the trade-off between performance and emissions, our benchmark reveals the cost of hyperparameter tuning. It examines the impact of data reduction techniques on the path toward sustainability-aware recommender systems. Our results show that simpler algorithms often deliver competitive accuracy at significantly lower emissions, and that exhaustive tuning can dramatically increase carbon costs with limited accuracy gains. Generally speaking, this study aims to discuss the challenges of energy consumption in recommender systems and to develop a new generation of algorithms that prioritize sustainability. All code and experiment traces are publicly released for reproducibility on Github.