Search Results (88,625)

Background: The presence of left atrial fibrosis is a marker of advanced remodeling and is associated with a worse outcome after pulmonary vein isolation (PVI). Conventional fluoroscopy-only cryoballoon ablation (CBA) lacks this prognostic information. The addition of electroanatomical mapping (EAM) using the inner lumen spiral catheter allows accurate voltage assessment of the left atrial posterior wall. However, the value of the finding of posterior wall low-voltage zones (pwLVZs) is unknown. Purpose: To study the value of left atrial voltage maps during CBA by comparing clinical and procedural characteristics and clinical outcome between patients with and without pwLVZs. Methods: A cohort of 250 consecutive patients who underwent index CBA for atrial fibrillation was analyzed. All patients underwent pre- and post-procedural EAM using the Achieve^TM catheter and EnSite^TM mapping system. The presence of LVZs was evaluated at the postprocedural voltage map of the posterior wall. Clinical success was defined as freedom from documented AF or atrial tachycardia (AT) >30 s after 1 year. Results: PwLVZs were found in 41/250 (16.4%) of patients. Patients with pwLVZs were older (69.3 ± 8.5 vs. 64.2 ± 10.4; p = 0.003), more frequently female (63.4% vs. 32.5%; p < 0.001) and had higher CHA₂DS₂-VASc scores (3.0 ± 1.6 vs. 2.0 ± 1.5; p < 0.001). The incidence of obesity (31.7% vs. 25.8%; p = 0.048), structural heart disease (35.5% vs. 17.4%; p = 0.021) and persistent AF (68.3% vs. 43.8%; p = 0.004) was higher in the pwLVZs group. Kaplan–Meier analysis of clinical outcome showed a higher recurrence rate in the pwLVZs group. The finding of pwLVZs was a predictor of atrial arrhythmia recurrence during follow-up (HR 2.583; 95%CI: 1.334–5.002; p = 0.005). Conclusions: In CBA facilitated by integrated EAM, pwLVZ was associated with older age, female sex, higher CHADS-VASc scores, obesity, structural heart disease and persistent AF. The finding of pwLVZs is predictive of a worse clinical outcome. Full article

High-dimensional covariance estimation remains a fundamental challenge when the number of variables (p) substantially exceeds the sample size (n). In such settings, the sample covariance matrix is unstable, singular, and heavily contaminated by estimation noise. Although shrinkage estimators improve stability and thresholding methods promote sparsity, each approach alone may introduce bias or lose structural information. This study proposes a Noise-Adjusted Shrinkage Covariance (NASC) framework as a post-processing enhancement strategy for shrinkage-based covariance estimators. The framework first stabilizes the covariance structure through shrinkage toward a structured target, then suppresses noise-induced small covariance entries via thresholding, and finally applies a stabilization step to ensure positive definiteness of the resulting estimator. Sensitivity analyses were conducted to investigate the effects of the shrinkage and thresholding parameters, and the Monte Carlo simulations were subsequently performed using the best-performing parameter configuration. The simulation results showed that shrinkage alone may not sufficiently suppress entrywise noise, whereas NASC-adjusted estimators improved upon their corresponding shrinkage baselines in many scenarios, with the strongest gains observed for sparse covariance structures and for shrinkage estimators that do not explicitly suppress entrywise estimation noise. Improvements were more limited for highly optimized shrinkage estimators. Real-data analyses were conducted on the SRBCT and colon cancer benchmark datasets. On the SRBCT dataset, numerical stability and positive-definiteness properties were examined, while LOOCV-LDA classification performance without prior feature selection or dimensionality reduction was evaluated on the colon cancer dataset. The results suggest that NASC provides a computationally simple and numerically stable extension to classical shrinkage covariance estimation methods for high-dimensions. Full article

Phosphorites are a vital source of phosphorus for agricultural and industrial applications and are increasingly recognized for their potential as secondary repositories of critical raw materials (CRMs) such as rare earth elements plus yttrium (REYs). This study investigates deep-sea phosphorites from the Galicia Bank, Madeira, and Canary Seamounts, in the NE Atlantic Ocean, which are spatially associated with ferromanganese (Fe-Mn) mineralization. Through integrated petrographic, geochemical, and in situ isotopic analyses (O and Sr), we assess the timing, processes, and paleoenvironmental conditions of phosphogenesis and its implications for CRM enrichment. Rare earth element patterns in apatite reflect a predominant seawater-derived signature with variable Ce anomalies. Nevertheless, variable Y/Ho ratios point to evolving fluid sources including a hydrogenous component (directly derived from seawater), modified porewaters and, locally, volcanic or possibly hydrothermal inputs. Oxygen and strontium isotope compositions constrain phosphogenesis to several episodes ranging from the Upper Cretaceous to the Middle Miocene, with distinct isotopic shifts identifying both primary formation and later overprinting processes mostly linked to Fe-Mn oxyhydroxide growth or volcanic–hydrothermal activity. These findings highlight the dynamic and multiphase nature of phosphorite formation in deep-marine settings. The integration of high-resolution geochemical and isotopic tools proves essential for reconstructing genetic histories, defining metallogenic context and evaluating CRM prospectivity in complex submarine systems. Full article

Data on menstrual health, energy availability, and injury risk in women undertaking extreme ultra-endurance expeditions remain limited. We conducted a prospective cohort study of the first all-female UK military team competing in a 3000-mile transatlantic rowing race, aiming to characterize menstrual function, low energy availability (LEA) risk, eating disorder (ED) risk, and injury profiles. Four female British Army personnel completed the 46-day race. Menstrual symptoms, injuries, and illnesses were recorded daily, while reproductive, inflammatory, biochemical, and hematological markers were assessed before and after the race. LEA and ED risk were evaluated using the Low Energy Availability in Females Questionnaire and Brief Eating Disorder in Athletes Questionnaire, respectively. Analyses were primarily descriptive. Three athletes experienced amenorrhea during the expedition, including one with previously regular cycles. The fourth reported intermittent abnormal bleeding associated with injury and illness and screened positive for LEA risk before and after the race. Another athlete screened positive for ED risk at both time points. Most biomarkers remained stable post-race, whereas reproductive hormones showed consistent reductions in follicle-stimulating hormone and luteinizing hormone in all four participants, alongside increased oestradiol. These findings, based on a sample of four athletes, suggest that menstrual function may be sensitive to sustained physiological stress in extreme ultra-endurance settings, and support prospective monitoring in female ultra-endurance, military, and expeditionary populations. Full article

Arrhythmogenic cardiomyopathy (ACM/ARVC) is an inherited myocardial disease characterized by progressive fibro-fatty replacement, ventricular arrhythmias, and an increased risk of sudden cardiac death. In addition to mutations in desmosomal genes, growing evidence suggests that microRNAs (miRNAs) actively contribute to disease pathogenesis by regulating key processes such as fibrosis, cell adhesion, and cardiac remodeling. This systematic review analyzed the main miRNAs identified in studies of human cardiac tissue and animal models of ARVC. Materials and Methods: Studies based on human myocardial tissue analysis (including autopsy and biopsy samples) and animal models of arrhythmogenic cardiomyopathy were included, using RNA sequencing, small RNA sequencing, miRNA arrays, and RT-qPCR. Studies on circulating miRNAs and narrative reviews were excluded. miRNAs were analyzed in relation to their functional pathways and their role in disease pathogenesis. Results: The synthesis of studies on human and animal cardiac tissue reveals a consistent miRNA signature associated with arrhythmogenic cardiomyopathy. MiR-21-5p and miR-29b-3p are associated with fibrosis and extracellular matrix remodeling, whereas miR-133a-b and miR-130a are linked to cardiomyocyte integrity loss and desmosomal dysfunction. A second group of miRNAs, including miR-217-5p, miR-708-5p, and miR-135b, regulates key pathways such as Wnt/β-catenin and Hippo signaling, contributing to structural remodeling and loss of cellular identity. Furthermore, downregulation of miR-499-5p is associated with mitochondrial dysfunction and cellular vulnerability, while the miR-142-3p, miR-182-5p, and miR-183-5p clusters contribute to differential molecular signatures compared with other cardiomyopathies. Overall, miRNAs converge on three main pathogenic axes: myocardial fibrosis, desmosomal impairment, and remodeling of cellular signaling pathways. Conclusions: The available evidence indicates that arrhythmogenic cardiomyopathy is regulated by a coordinated network of miRNAs that actively drives myocardial damage progression. These miRNAs represent not only biomarkers but also functional mediators of disease, suggesting potential diagnostic and therapeutic applications based on tissue-specific molecular signatures, including in post-mortem settings. Full article

Background: Pain during orthodontic debonding is a common clinical concern. Although previous studies have mainly focused on mechanical approaches to reduce discomfort, the influence of emotional characteristics of patients on pain perception remains insufficiently explored. In this study, we aimed to evaluate the association between pain perception, emotional affect, and anxiety during orthodontic bracket removal using different clinical protocols. Methods: A prospective observational comparative study was conducted at Rey Juan Carlos University (Madrid, Spain). A total of 140 orthodontic patients underwent bracket removal according to four routine clinical protocols determined by clinical scheduling: ligated with interocclusal cotton rolls (used for tooth stabilization), non-ligated with cotton rolls, ligated without cotton rolls, and non-ligated without cotton rolls. Pain intensity was assessed using a Visual Analog Scale (VAS) immediately before (T0) and after (T1) bracket removal. Baseline pain (T0) was recorded to control for pre-existing discomfort. Anxiety and emotional affect were measured using the State–Trait Anxiety Inventory (STAI) and the Positive and Negative Affect Schedule (PANAS), respectively. Statistical analyses included ANOVA, factorial ANCOVA adjusted for baseline pain, and multivariable regression models. Results: No significant baseline differences were observed among groups. The highest post-debonding pain scores were found in the group without cotton rolls and without ligatures. ANCOVA revealed a significant main effect of cotton roll use, with lower adjusted pain scores in patients treated with cotton rolls, whereas ligation showed no statistically significant independent effect. In multivariable regression analyses, baseline pain, age, and negative affect were independently associated with higher post-debonding pain. Conclusions: Within the limitations of a non-randomized design, cotton roll use was associated with lower post-debonding pain, whereas ligation appeared to have a limited influence. Patient-related factors—particularly negative affect, age, and baseline pain—were also associated with pain perception, supporting a biopsychosocial perspective. These findings should be interpreted as exploratory evidence rather than causal effects. Full article

The prognostic value of post-radiotherapy (RT) lymphocyte recovery remains unclear in locally advanced esophageal squamous cell carcinoma (ESCC), and tools to predict recovery are lacking. This study evaluated lymphocyte recovery as a survival predictor and developed a prediction model. We analyzed 233 patients (2019–2024; training:validation = 7:3). Lymphocyte recovery was assessed at 1 and 3 months post-RT (ΔALC₁ > 0.41 and ΔALC₃ > 0.25 × 10⁹/L, calculated as ALC at each time point minus ALC at the end of RT). Patients were stratified into three groups by recovery status: no recovery (Group 0), recovery at both time points (Group 2), or at only one time point (Group 1). Multivariate logistic regression identified predictors of lymphocyte recovery, and a nomogram was developed and internally validated. Median overall survival (OS) was 26.4 months and median progression-free survival (PFS) was 13.9 months. OS differed significantly among groups: 16.0 months (Group 0), 26.0 months (Group 1), and 50.0 months (Group 2) (p < 0.001). Median PFS was 10.2, 12.0, and 36.6 months, respectively (p < 0.001). Independent predictors included ECOG 0 and thoracic spine V₅ < 57.3%; planning target volume < 210 cm³ showed a trend toward association (p = 0.051). The nomogram demonstrated AUCs of 0.77 and 0.75 in the training and validation cohorts. Superior lymphocyte recovery appears to be associated with improved survival. The model, if externally validated, may facilitate individualized risk stratification. Full article

Reliable insulation design for LNG feedthroughs requires fundamental dielectric breakdown data obtained under cryogenic LNG conditions. However, such data remain scarce owing to the explosion-proof requirements imposed by the flammable nature of LNG. Furthermore, the influence of phase differences between LNG and NG on creepage dielectric breakdown behavior along insulation surfaces has received little attention. In this study, an explosion-proof cryostat and test facility compliant with the IEC 60079 series of standards were developed, and dielectric breakdown tests were conducted over a range of electrode gap distances and pressures. Two electrode configurations were employed: rod–plate electrodes for dielectric breakdown characterization in LN${}_2$ and LNG, and creepage electrodes for surface dielectric breakdown evaluation in NG and LNG. Experimental results show that LNG requires approximately 1–2 bar of additional operating pressure above that of LN${}_2$ to achieve equivalent dielectric strength. Moreover, LNG exhibited higher creepage dielectric breakdown voltages than NG under all test conditions, with the difference becoming more pronounced as pressure and creepage distance increased. Post-breakdown surface analysis revealed distinct differences in carbonization patterns between the two media. The findings of this study are expected to serve as fundamental reference data for the insulation design of LNG-based cryogenic feedthroughs. Full article

Background and Objectives: Corneal collagen cross-linking (CXL) halts keratoconus progression, yet potential differences between conventional and accelerated protocols at one year remain uncertain. We analyzed the completed 12-month follow-up of a previously reported 6-month cohort to compare conventional (3 mW/cm² × 30 min; CXL 30) versus accelerated (9 mW/cm² × 10 min; CXL 10) CXL, interpreting outcomes within the ABCD framework alongside Kmax and curvature radii. Materials and Methods: In this single-center retrospective longitudinal analysis of prospectively collected routine clinical data, 22 eyes were included, with assessments performed at baseline and at 1, 3, 6, and 12 months of follow-up. Evaluated outcomes comprised ABCD stages (A–D), anterior and posterior radius of curvature (ARC and PRC), Kmax, pachymetric and elevation indices, as well as UDVA and BCVA. Within-group change used Friedman with Wilcoxon post hoc; between-group differences used Mann–Whitney (α = 0.05). Results: Both protocols resulted in significant visual improvement and Kmax reduction at 12 months (overall time effect: CXL 30 p < 0.001; CXL 10 p = 0.026). Median Kmax decreased 56.5 → 52.3 D (CXL 30) and 59.3 → 58.3 D (CXL 10). UDVA improved 0.2 → 0.6 (CXL 30) and 0.2 → 0.3 (CXL 10); BCVA 0.4 → 0.8 (CXL 30) and 0.2 → 0.5 (CXL 10). Tomographic analysis showed predominantly anterior changes, with a significant decrease in A stage in the CXL 30 group and an increase in ARC in both groups, more pronounced in CXL 30. In the late 6 → 12-month window, posterior metrics (PRC and posterior elevation) were largely stable; raw PRC change did not reach significance. Conclusions: Conventional and accelerated CXL both stabilized keratoconus at one year with meaningful functional gains. Beyond 6 months, remodeling was predominantly anterior; within-group findings suggested a more pronounced anterior tomographic response in the CXL 30 group. The 12-month visit may be useful for reassessing stability after CXL, although this study was not designed to determine optimal retreatment timing or optical rehabilitation strategy. Longer-term studies with standardized biomechanical and densitometric endpoints are warranted to assess durability and refine protocol selection. Full article

Objectives: This study investigated the relationship between interstitial and blood glucose concentrations following ingestion of carbohydrate-containing drinks differing in carbohydrate amount, osmolarity, and glycaemic index. Methods: Ten healthy adults (nine male; age: 22 ± 1 y; height: 177 ± 12 cm; weight: 75 ± 14 kg) completed a double-blind, randomised cross-over study with seven beverage trials varying in carbohydrate (CHO) characteristics. Blood samples were collected at rest and over two hours, while interstitial glucose ([iG]) was recorded using a continuous glucose monitor (Supersapiens, Atlanta, USA). Glycaemic metrics and mean absolute relative difference (MARD) were calculated for hypoglycaemic, euglycaemic, and hyperglycaemic ranges. Data were analysed using repeated-measures ANOVA with Bonferroni correction and paired t-tests (p ≤ 0.05). Results: Interstitial and blood glucose concentrations were similar at baseline but diverged post-ingestion. MARD varied by glucose rate and direction, exceeding 20% during rapid declines (>2 mg/dL/min), where [iG] underestimated [BG] by −7.3 ± 27.1 mg/dL. Accuracy was highest during stable glucose (MARD = 10.5 ± 8.6%). Carbohydrate amount and glycaemic index influenced peak glucose, whereas beverage concentration (5–20%) had minimal effect when CHO amount was fixed, though variation in CGM agreement appeared during post-peak declines. Conclusions: CGM tracked blood glucose overall but showed reduced accuracy during rapid falls or hypoglycaemia. Carbohydrate properties influenced glycaemic response but not sensor agreement when CHO load was constant. Glucose rate and direction of change are key considerations for interpreting CGM data in research and applied settings. Full article

Electric Vehicle (EV) charging infrastructure plays a critical role in modern energy systems, affecting energy load distribution, demand-response programs, and grid stability. As EV adoption accelerates globally, the varied charging habits and concurrent interactions among users, stations, and shared infrastructure create operational inefficiencies that existing machine learning and optimization approaches cannot fully diagnose, because these methods rely on aggregated or single-entity representations that discard cross-object process dependencies. To address this gap, we propose EnOptiMine (Energy Optimization Framework for Electric Vehicles through Object-Centric Process Mining), a novel four-phase analytical framework that applies Object-Centric Process Mining (OCPM) to EV charging infrastructure. EnOptiMine operates by transforming raw EV charging data into an Object-Centric Event Log (OCEL 2.0), discovering the complete charging lifecycle as a structured multi-object process through Object-Centric Directly-Follows Graphs (OC-DFGs), performing conformance analysis to detect and quantify process deviations across object-type lifecycles, and proposing process improvement interventions. Applied to the EV charging dataset, EnOptiMine identifies sessions that exhibit post-charge station idle-blocking, departure mismatch, and carry lifecycle ordering violations. In the present work, the real-world simulation confirms that a graduated idle fee policy recovers 22.9% of wasted station-hours, and a departure reconfirmation protocol reduces mismatch sessions by 54.0%. These results demonstrate that OCPM provides process-transparent diagnostic capabilities for EV charging infrastructure that are inaccessible to existing prediction- and optimization-based methods. Full article

Bio-based and biodegradable polymer composites based on polylactic acid (PLA) and polybutylene succinate-co-adipate (PBSA) were developed for rigid food packaging applications. Agro-industrial residues consisting of ground leaves and branches derived from tangerine tree cultivation (pruning) were used as fillers at high loading (30 wt%) before (PRE) or after (POST) extraction of bioactive compounds. The influence of blend composition (PLA/PBSA 60/40 and 30/70), filler extraction, and the addition of antioxidants (0.5 wt%) on material properties was systematically investigated. Composites were processed via extrusion and injection molding and characterized through FTIR, SEM, tensile testing and thermal analysis. The results show that polymer blend morphology affects mechanical behavior, with co-continuous structures (60/40) exhibiting improved ductility compared to dispersed systems (30/70). The incorporation of lignocellulosic residues increased stiffness but reduced elongation at break. Extraction treatment significantly modified filler morphology and interfacial interactions, slightly improving dispersion and processability. The effect of the extracted bioactive compounds on the thermal stabilization of biocomposites was also investigated. Overall, the findings demonstrate the potential of combining biodegradable polymer blends with treated agricultural residues to produce sustainable rigid packaging materials while supporting a bio-circular approach. In fact, preliminary extraction of valuable compounds from tangerine pruning waste appears to be a convenient strategy for its efficient cascade valorization. Full article

Background: Craniofacial growth is regulated by a complex interaction of genetic, functional, and systemic metabolic factors. Mechanical loading generated during mastication plays a fundamental role in mandibular development through bone modeling and remodeling mechanisms. In contrast, Type I diabetes mellitus is associated with impaired bone metabolism, which may compromise skeletal growth. Although the independent effects of functional loading and metabolic disorders on bone tissue have been widely investigated, their combined influence on mandibular development remains insufficiently understood. Objective: This study primarily aimed to evaluate the effect of Type I diabetes mellitus on mandibular growth in growing rats and, secondarily, to assess the impact of dietary consistency (hard versus soft food) on mandibular development under diabetic and non-diabetic conditions, as well as determine whether diabetes modifies the mandibular adaptive response to increased masticatory loading. Materials and Methods: An experimental animal study was conducted using twenty-four male Wistar rats aged one month. The animals were randomly allocated into four groups according to metabolic status (control or diabetic) and dietary consistency (hard or soft diet). Type I diabetes mellitus was experimentally induced in the relevant groups using streptozotocin. All animals underwent cone beam computed tomography (CBCT) scanning at baseline (Day 1) and at the end of the experimental period (Day 28). Linear measurements were obtained using specialized software following euthanasia. Two-way ANOVA was used to evaluate the effects of diabetes, diet, and their interaction, using appropriate post hoc tests for multiple comparisons. Categorical variables were analyzed using the chi-square test. A p-value < 0.05 was considered statistically significant. Results: Longitudinal morphometric analysis demonstrated that Type I diabetes mellitus significantly impaired mandibular growth. Diabetic animals exhibited reduced growth rates and smaller final mandibular dimensions compared with controls. Hard diet intake significantly enhanced mandibular development in control animals, reflecting a strong adaptive response to increased functional loading. However, this osteogenic response was markedly attenuated in diabetic rats. Under soft-diet conditions, differences between control and diabetic groups were diminished, indicating that reduced mechanical stimulation limits adaptive growth responses. Significant interaction effects between diabetes and dietary consistency were observed in several morphometric parameters, particularly those related to mandibular body length, ramus height, and condylar position. Conclusions: Type I diabetes mellitus exerts a detrimental effect on mandibular growth in growing rats and compromises the adaptive response of craniofacial structures to mechanical loading. Although a hard diet functions as a potent osteogenic stimulus, its growth-promoting effect is substantially reduced in the presence of metabolic dysfunction. Full article

Background/Objectives: Spontaneous intracranial hypotension (SIH) is caused by spinal cerebrospinal fluid (CSF) leakage and is typically diagnosed by clinical presentation and characteristic MRI signs; however, objective tools for monitoring physiological changes and treatment response remain limited. Cine phase-contrast MRI (PC-MRI) enables noninvasive quantification of aqueductal CSF dynamics, yet reliable analysis is challenging since the cerebral aqueduct is extremely small and susceptible to low contrast, partial volume effects, and ROI-dependent measurement variability—particularly in SIH where CSF pulsatility is often reduced. Methods: We propose an end-to-end automated framework that integrates (1) a cascade localization–segmentation strategy, consisting of Tiny YOLOv4 detection followed by MultiResUNet segmentation on a YOLOv4-derived cropped ROI; (2) physiology-informed pulsatility-based segmentation (PUBS) to refine anatomical masks into functional flow ROIs; and (3) one-dimensional convolutional neural networks (1D-CNNs) to extract exploratory waveform morphology features from 32-phase cardiac-cycle velocity waveforms. The study includes 39 participants, yielding 59 cine PC-MRI examinations: 11 controls, 28 Pre-treatment SIH scans and 20 Post-treatment Recovery scans. Results: The cascade model significantly improves segmentation robustness compared with a full-image baseline, achieving higher Dice scores and markedly lower boundary errors across cohorts (e.g., Pre-treatment SIH HD95: 1.66 ± 0.74 px vs. 15.37 ± 44.98 px). PUBS refinement reduces quantification deviation from expert manual references in SIH (mean relative error: 7.4% to 5.6%) and improves diagnostic performance for multiple hemodynamic parameters (e.g., downward mean flow AUC: 0.747 to 0.792). For waveform morphology analysis, the end-to-end 1D-CNN classifier was evaluated using repeated-seed participant-level grouped LOOCV. The repeated-seed ensemble prediction showed modest out-of-sample discrimination between Normal controls and Pre-treatment SIH scans, with an AUC of 0.646, a bootstrap 95% confidence interval of 0.455–0.826, and a permutation-test p-value of 0.072. Separately, exploratory analysis of the final baseline-trained 1D-CNN latent space showed marked, apparent Normal-versus-SIH separability and an intermediate recovery distribution in PCA space, suggesting that aqueductal waveform morphology may encode SIH-related physiological information. Conclusions: These findings suggest that SIH-related information may be reflected not only in flow magnitude but also in aqueductal CSF waveform morphology. However, the modest and statistically non-significant out-of-sample performance of the end-to-end 1D-CNN classifier indicates that morphology-based AI features should currently be regarded as exploratory biomarker candidates rather than validated stand-alone diagnostic tools. Larger independent cohorts are required to confirm their reproducibility, physiological meaning, and clinical utility. Full article

The growing complexity of web attacks highlights the need for adaptive, intelligent defense systems that overcome the limitations of traditional rule-based web security. Thus, the architecture proposed in this paper integrates data-driven deep learning with deterministic rule-based logic to enhance real-time detection accuracy and adaptability in dynamic web threat environments. The practical integration of a deep learning-based Gated Recurrent Unit (GRU) model with ModSecurity, an open-source Web Application Firewall (WAF), is employed to improve the detection and classification of malicious HTTP requests. The model, pre-trained on a large labeled up-to-date dataset of web traffic and attack types collected post-2020, is designed to classify requests in real-time, identifying both whether a request is malicious and the corresponding attack category (e.g., SQL Injection, Cross-Site Scripting, Command Injection). We demonstrate how the trained model is incorporated into ModSecurity’s inspection pipeline, allowing it to analyze real-time web traffic alongside traditional rule-based inspection. This hybrid approach aims to significantly reduce false positives and improve adaptability to new attack patterns. Evaluation metrics such as accuracy, receiver operating characteristic (ROC), area under the curve (AUC), Principal Component Analysis (PCA), confusion matrix, and t-Distributed Stochastic Neighbor Embedding (t-SNE) visualization are discussed, along with performance considerations and implementation architecture. The integration presents a robust framework for ML-improved intelligent web security defense. Full article