Search Results (6,713)

Fish disease and species identification is critical for intelligent aquaculture, directly influencing productivity, sustainability, and economic viability. However, existing approaches largely treat species identification and pathological classification as independent tasks, limiting their ability to capture interdependent features under complex real-world conditions such as occlusion, low contrast, dynamic backgrounds, and high inter-class similarity. Moreover, challenges including class imbalance, cross-species variability, and fine-grained feature discrimination remain insufficiently addressed. To overcome these limitations, this paper proposes a hybrid ConvNeXt–BiLSTM–multi-head self-attention (MHSA) framework for joint fish species and disease classification, where a ConvNeXt-Small backbone extracts hierarchical spatial features that are transformed into a structured sequence and processed by a bidirectional LSTM to capture contextual dependencies, followed by an MHSA module for adaptive feature refinement. An auxiliary species classification branch is incorporated to provide multi-task regularization without additional inference costs. The training pipeline integrates CLAHE-based image enhancement, square-root inverse-frequency focal loss, targeted minority oversampling, and a two-stage progressive learning strategy with differential-rate cosine annealing, complemented by five-view test-time augmentation. For practical deployment, a YOLOv8s detector is employed for fish localization prior to classification. The experimental results demonstrate that the proposed model achieves superior performance, attaining overall top-1 classification accuracy of 94.33%, precision of 97.1%, recall of 90.9%, 96.1% mAP50, and an F1-score of 0.9264, while achieving a macro AUC of 0.994 and maintaining high computational efficiency (213.3 FPS), demonstrating a robust and efficient solution for real-time fish disease screening. Full article

Background and Objectives: Although the anterior mandible is generally considered a safe region for implant placement, injury to the medial lingual foramen (MLF) may result in significant vascular complications. Accurate identification of this structure is challenging due to its small size, low volumetric representation, and anatomical variability. This study aimed to evaluate the anatomical characteristics of the MLF using cone-beam computed tomography (CBCT) and to develop and validate a deep learning-based approach for its automated detection and segmentation. Materials and Methods: A total of 106 CBCT scans were retrospectively analyzed to assess the morphology and position of the MLF. Manual pixel-wise annotations of the complete canal trajectory were performed on sagittal slices and used to train convolutional neural network models based on a U-Net-derived framework. Multiple configurations, including multi-class, binary, two-dimensional, and three-dimensional approaches, were evaluated. Given the extremely limited volumetric representation of the MLF, severe class imbalance represented a major challenge during model training and evaluation. Model performance was assessed using the Dice similarity coefficient, precision, recall, and Hausdorff distance. External validation was performed on an independent dataset of 10 CBCT scans. Results: The MLF was identified in all patients, with a single canal observed in 63% of cases. The sagittal-plane binary segmentation model achieved the best performance, with a test Dice score of 0.79, precision of 0.88, and recall of 0.73. External validation demonstrated a Dice score of 0.81, precision of 0.89, and recall of 0.71. The 95th percentile Hausdorff distance was 2.6 mm, and the mean center-point localization error was 1.2 mm. The model correctly detected the MLF in 90% of external cases. Conclusions: Deep learning-based segmentation of the MLF is feasible and may support automated localization assistance during preoperative CBCT assessment. Performance was influenced by the alignment between the annotation strategy and model input, highlighting an important consideration for small-structure segmentation. Further validation on larger multicenter datasets is required before clinical implementation can be considered. Full article

Background: Pre-Chronic Obstructive Pulmonary Disease (pre-COPD) and Preserved Ratio Impaired Spirometry (PRISm) phenotypes represent important components of the early obstructive lung disease spectrum, characterized by respiratory symptoms and structural lung abnormalities prior to the development of overt airflow limitation. Emphysema is considered one of the major structural phenotypes underlying airway disease and the COPD spectrum. Although cigarette smoking is the best recognized risk factor for these conditions, non-tobacco exposures may also contribute to early structural lung changes. In this study, we evaluated the radiological features, pulmonary function parameters, and dyspnea severity of CT-detected emphysema in symptomatic patients classified as having pre-COPD or PRISm, with particular attention paid to the potential influence of smoking status on disease characteristics. Methods: In this retrospective, single-center study, symptomatic patients aged 20–50 years classified as having pre-COPD or PRISm and in whom emphysema was detected on high-resolution computed tomography (HRCT) were evaluated. Only symptomatic patients who underwent HRCT for clinical indications and in whom emphysema was identified were included. Demographic characteristics, emphysema type and quantitative emphysema severity, pulmonary function parameters, and Modified Medical Research Council (mMRC) dyspnea scores were analyzed. The PRISm and pre-COPD groups were compared in terms of clinical and symptomatic characteristics. In addition, smoking-related clinical and radiological characteristics were also evaluated. Results: A total of 232 patients were included in the study. The median age was 43 years (38–48), and 84.1% of the participants were male. Among the study population, 68.5% were classified in the pre-COPD group and 31.5% in the PRISm group. The most frequently identified emphysema patterns were paraseptal (44.4%) and centrilobular (40.5%). The median total lung emphysema area was 18% (13–22). A weak negative correlation was observed between the degree of emphysema and FEV₁ (r = −0.185; p = 0.005), whereas a weak positive correlation was found between emphysema extent and the mMRC dyspnea score (r = 0.214; p = 0.001). Dyspnea severity was significantly higher in the PRISm group compared with the pre-COPD group (p < 0.001). In the smoking-status subgroup analysis, ever-smokers demonstrated significantly greater dyspnea severity and lower FEV₁ values, whereas never-smokers had a significantly higher proportion of emphysema extent > 18% (all p < 0.05). Conclusions: Radiologically detected emphysema in symptomatic patients without airflow limitation was associated with statistically significant but weak alterations in pulmonary function and dyspnea burden. Dyspnea severity was significantly higher in the PRISm phenotype. In a smoking-status subgroup analysis, ever-smokers had significantly greater dyspnea severity, whereas never-smokers showed a significantly higher proportion of extensive emphysema (>18%), despite similar functional impairment across groups. These findings underscore the importance of non-tobacco exposures in the development of emphysema within pre-obstructive spirometric phenotypes. Multicenter prospective studies incorporating healthy controls and systematic exposure documentation are needed to confirm these observations. Full article

Background and Objectives: Head trauma (HT) has been suggested as a risk factor for Parkinson’s disease (PD), but its impact on the motor and non-motor manifestations remains unclear. We investigated whether patients with early PD and a history of HT differ from those without HT in terms of their motor and non-motor symptom profiles. Materials and Methods: We enrolled patients with early PD (disease duration of ≤5 years, modified Hoehn and Yahr stages [mHY] 1–3). HT history was ascertained by structured questionnaire. Motor and non-motor symptoms were evaluated using standardized clinical rating scales. Motor subtypes—tremor-dominant (TD), akinetic-rigid (AR), and mixed—were determined according to established criteria based on the tremor-to-AR score ratio. Subtype distribution and motor scores were compared between HT and non-HT groups using univariate tests, mHY-adjusted ANCOVA, and multivariable models adjusting for age, sex, disease duration, education, mHY stage, MMSE, and BDI score. Results: Of 237 patients, 35 (14.8%) reported HT. The HT group had a higher mHY stage than the non-HT group and showed lower total and rest tremor scores on the Unified Parkinson’s Disease Rating Scale, whereas rigidity scores were similar. Bradykinesia and gait/posture scores tended to be higher in the HT group, but these differences did not persist after adjustment for disease severity. The TD subtype was less frequent in the HT group than in the non-HT group (5.7% vs. 30.2%), whereas the AR subtype was more common (82.9% vs. 61.9%). The categorical subtype redistribution remained significant after multivariable adjustment (adjusted OR for AR vs. TD = 4.61, 95% CI 1.28–16.67). No detectable between-group difference in non-motor symptom burden was observed. Conclusions: In this single-center cross-sectional cohort, a self-reported history of HT in early PD was associated with a redistribution of motor phenotype categories toward AR-predominant presentations, with no detectable difference in non-motor symptom burden. Given the retrospective binary exposure assessment, these findings should be interpreted as hypothesis-generating, and prospective studies with structured exposure ascertainment are needed to clarify how HT may shape PD motor phenotype expression. Full article

Purpose: Survivors of childhood lower extremity and pelvic osteosarcoma have a high burden of functional, strength, and range of motion (ROM) impairment. However, longitudinal trajectories of these outcomes have not been well studied. Methods: This longitudinal study included patients enrolled on an institutional therapeutic protocol with lower extremity or pelvic osteosarcoma. Functional assessments (functional mobility assessment, FMA) were performed at diagnosis, prior to surgery, 10–12 weeks post-surgery, end of therapy, and 6, 18, and 48 months post-therapy. Generalized mixed models were used to evaluate differences in FMA scores, strength, and ROM across time points. Results: The study included 35 patients (mean age 13.1 ± 3.4 years) with osteosarcoma of the lower extremity (n = 34) and pelvis (n = 1). The majority underwent a limb salvage procedure (74.3%). FMA scores were lowest at the pre-therapy time point (least squared (LS) mean 27.2, standard error (SE) 2.2) and highest at the 48-month follow-up time point (LS mean 48.4, SE 2.4, p-trend < 0.01). A large proportion of participants had scores indicating functional impairment throughout the study period including 80% of participants at the 48-month follow-up. Strength and ROM scores demonstrated similar trends in improvement over time; however, they remained below normative scores at long-term follow-up. Conclusions: A high burden of impairment in function, strength, and ROM persists among survivors of childhood lower extremity and pelvic osteosarcoma from diagnosis through long-term follow-up. Implications for Cancer Survivors: A lifelong focus on mitigating impairments in function, strength, and ROM is needed among survivors of childhood pelvic and lower extremity osteosarcoma. Full article

In rock tunnel engineering, the discontinuity roughness plays a crucial role in rock mass stability. Nevertheless, no suitable parameter is currently available to describe the multi-scale morphological complexity associated with varying shear directions. Improved 1D and 3D fractal dimensions are proposed and systematically evaluated for their ability to characterize the anisotropy and morphology complexity. The results show that the trends of improved 1D and 3D fractal dimensions are consistent with the Grasselli parameter GP(θ) along the orthogonal direction of the major and minor axes, which effectively characterizes the anisotropy. Meanwhile, they effectively characterize the profile and discontinuity morphology complexity, respectively. On the basis of the evaluation results, a comprehensive parameter is proposed to quantify directional multi-scale morphological complexity, which characterizes the discontinuity anisotropy, local roughness, and global morphological complexity simultaneously. Furthermore, the proposed parameter is compared with GP(θ) based on the Huashansong tunnel, using the concordance correlation coefficient and macro-trend similarity as evaluation criteria. The results show a moderate correlation between the two parameters, with a concordance correlation coefficient of 0.716. The overall similarity score of 85.8 and the macroscopic trend similarity of 71.0 indicate strong consistency in morphological features. In addition, the RMSE and MAE of γ(θ) are 0.25 and 0.19, which is lower than that of the existing GP(θ) and PSD-based methods in the study of characterizing the dominant morphology complexity direction. Full article

We propose QwenMoE-SC, a semantic communication framework that integrates a Mixture-of-Experts (MoE) Large Language Model with three complementary modules: (1) a Graph Neural Network (GNN)-based Unequal Error Protection (UEP) plug-in that assigns semantic importance scores via syntactic dependency graph message passing for adaptive bit allocation, without modifying the pre-trained LLM; (2) NEFTune noise injection during fine-tuning for channel robustness; and (3) a Communication-aware Direct Preference Optimization (C-DPO) strategy that favors semantically faithful yet token-efficient transmissions. Comprehensive ablation studies on AWGN and Rayleigh fading channels show that each component contributes distinct gains, and their combination consistently outperforms traditional separation-based methods and neural baselines in sentence similarity, BLEU score, and semantic-level BER, with the largest improvements at low-to-mid SNR regimes. QwenMoE-SC can also serve as a semantic interface layer within expert and decision-support systems, enabling robust, bandwidth-efficient communication between data sources, inference engines, and human users. Full article

Introduction: Advances in pediatric oncology have transformed cancer into a condition with chronic and long-term developmental consequences. While survival rates have improved significantly, the literature on psychosocial outcomes remains fragmented and inconsistent, with a notable lack of person-centered analyses that account for the heterogeneity of adaptive trajectories. Current evidence fails to explain why survivors with similar clinical profiles exhibit divergent psychological phenotypes, particularly regarding the late effects of multimodal treatments. The aim of this study was to identify heterogeneous psychosocial profiles among adolescent cancer survivors and to examine their associations with treatment complexity and quality of life. Materials and Methods: This cross-sectional study included 165 adolescents aged 12–18 years (mean age: 14.64 years) who were in clinical remission following oncological treatment. Standardized assessment tools were used: the Children’s Depression Inventory 2 (CDI-2™) to measure depressive symptoms, the KIDSCREEN-10 index to assess health-related quality of life (HRQoL), and a scale evaluating satisfaction across 14 life domains. Adaptive profiles were identified using a Two-Stage Cluster Procedure, and risk factors were examined using multinomial logistic regression. Results: Four clusters were identified in the study population: a depressive–dysphoric profile, an anhedonic-withdrawn profile, a highly adaptive profile, and a mixed (struggling) profile. Treatment complexity was identified as a significant independent predictor of membership in the high-distress (depressive) cluster. While each additional therapeutic modality beyond standard chemotherapy was associated with a markedly increased risk (OR = 8.91; p < 0.001), the relatively wide confidence interval (95% CI: 3.27–24.31) suggests that the exact magnitude of this effect should be interpreted with caution. The high lower bound of the interval (3.27), however, strongly supports the directional association of cumulative iatrogenic burden with psychological adaptation. Subjective quality of life functioned as a protective factor against depressive symptoms (OR = 0.57); however, paradoxically, higher self-reported quality of life increased the likelihood of classification into the anhedonic group (OR = 1.81). This divergence between high self-reported HRQoL and social withdrawal potentially suggests a ‘well-being paradox’. It is hypothesized that standard HRQoL instruments may primarily capture physical remission and relief from acute somatic symptoms, potentially masking underlying social–emotional deficits. This suggests that HRQoL scores in survivors should be interpreted with caution and complemented by specific affective screenings. Conclusions: The absence of a uniform pattern of psychological response to cancer among adolescent survivors supports the validity of a patient-centered approach. The burden associated with intensive multimodal treatment significantly increases the likelihood of full-syndrome depression during adolescence. Moreover, the identification of a cluster suggestive of anhedonic and socially withdrawn features highlights the limitations of standard screening tools focused solely on the detection of overt sadness. This heterogeneity underscores the need for personalized psycho-oncological care and the implementation of intensified monitoring for patients at high medical risk. Full article

Objectives: Contrast-enhanced magnetic resonance imaging (CE-MRI) plays an important role in the diagnosis, treatment monitoring, and follow-up of brain tumors. However, the use of gadolinium-based contrast agents (GBCAs) is limited in patients with contraindications, such as severe renal impairment or situations requiring repeated examinations. This study aimed to develop a diffusion model-based Difference-Aware Guided Control Network (DAGCN) for synthesizing high-quality contrast-enhanced T1-weighted MRI (T1-CE) from non-contrast T1-weighted images in combination with an auxiliary sequence. Methods: Using the BraTS 2021 dataset, we proposed a two-stage generative framework that first localizes lesion-related enhancement cues and then guides image synthesis. In the first stage, a Difference-Aware Fusion and Prediction (DAFP) module was designed to extract complementary information from non-contrast T1-weighted images and an auxiliary sequence (T2-weighted or FLAIR) through dual-branch feature extraction and cross-modal channel attention fusion, followed by prediction of a lesion-related discrepancy map. In the second stage, the predicted discrepancy map was concatenated with the original T1-weighted images and introduced into a ControlNet-guided diffusion model to constrain the reverse denoising process and generate the target T1-CE image. Model performance was evaluated by visual comparison, quantitative metrics including peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), visual information fidelity (VIF), and normalized cross-correlation (NCC), as well as blinded radiologist scoring of image quality (IQ), clinical replaceability (IC), contrast enhancement (CE), and lesion conformity (CF). Results: DAGCN generated synthetic T1-CE images with preserved global anatomical structure and faithful local lesion enhancement without the need for contrast agent administration. Compared with baseline methods, DAGCN achieved the highest PSNR and NCC under both T1 + T2 and T1 + FLAIR settings, while showing competitive SSIM and VIF performance. Visual comparison and radiologist-based subjective evaluation further indicated improved lesion-focused enhancement fidelity and reduced false-positive enhancement. Among the two auxiliary sequence settings, the T1 + FLAIR configuration provided more specific lesion localization and cleaner background suppression than the T1 + T2 configuration, particularly by reducing interference from cerebrospinal fluid signals. Conclusions: The proposed DAGCN framework enables the synthesis of clinically informative contrast-enhanced-like MRI from non-contrast multi-sequence inputs and may provide a promising alternative for patients in whom gadolinium administration is contraindicated or should be avoided. In particular, the FLAIR-guided setting showed advantages in lesion specificity, background cleanliness, and overall diagnostic quality. Full article

Background: Respiratory syncytial virus (RSV) is a serious disease in older adults and is associated with various comorbidities; however, comparative data across epidemic waves, both clinically and in terms of inflammatory profiles and their diagnostic and prognostic utility, remain limited. Methods: We conducted a retrospective study of adults hospitalized with RSV infection across two epidemic waves (2022–2023 and 2024–2025). Data on clinical characteristics, comorbidities, severity scores, and outcomes were collected, and serum interleukin-6 (IL-6), C-reactive protein (CRP), and hematological parameters were analyzed and compared with those in healthy controls. Results: A total of 152 patients were included in this study (81 in wave 1 and 71 in wave 2). Patients in wave 2 were older and had a higher burden of comorbidities, although ICU admission and in-hospital mortality were similar across waves. RSV induced a consistent systemic inflammatory response in both waves, characterized by elevated IL-6 and CRP levels, neutrophilia, lymphopenia, and increased neutrophil-to-lymphocyte ratios, with no relevant inter-wave differences. All biomarkers demonstrated good diagnostic performance. The neutrophil-to-lymphocyte ratio showed the highest accuracy, while IL-6 exhibited high rule-in capacity. However, none of the evaluated biomarkers were associated with disease severity or mortality. Conclusion: RSV infection in older adults is associated with a similar inflammatory profile across waves. Although biomarkers showed strong diagnostic utility, they did not show any significant prognostic discrimination in this cohort. We suggest that disease severity is primarily associated with host-related factors, particularly comorbidities, rather than with differences in the inflammatory response, highlighting the need for improved preventive and risk-stratification strategies in this population. Full article

Background and Objectives: Adenoidectomy is a common pediatric procedure performed and endoscopic, instrumented techniques such as a powered microdebrider and coblation aim to improve visualization and completeness of resection while potentially reducing pain, bleeding and recurrence compared with blind curettage. To compare clinical outcomes and safety of endoscopic-assisted adenoidectomy using a microdebrider versus coblation in pediatric patients. Materials and Methods: This retrospective comparative cohort study included 50 children who underwent endoscopic-assisted adenoidectomy using either a microdebrider (n = 25) or coblation (n = 25) during a 6-month study period. Operative time (skin-to-skin), pain scores at 24 h, early complications at prespecified intervals, and endoscopic assessment at 6 months were extracted from medical records and compared. Results: Groups were comparable at baseline (mean age ≈ 9.6 years, all baseline p > 0.05). Mean skin-to-skin operative time was significantly shorter with the microdebrider (24.9 ± 6.9 min) than with coblation (31.1 ± 8.4 min), p = 0.004. Mean 24 h VAS did not differ significantly between the microdebrider and coblation (3.20 ± 1.21 vs. 2.73 ± 1.30, p = 0.191). Early complication rates (postoperative bleeding within 48 h, infection within 7 d, halitosis, neck pain) were low and showed no statistically significant differences (all p > 0.05). Conclusions: Both microdebrider and coblation endoscopic adenoidectomy were safe and effective in this cohort study. The microdebrider was associated with a shorter operative time in our setting, while early pain, short-term complications and six-month endoscopic residual/recurrence were similar between techniques. Full article

Background: Intrathecal morphine (ITM) is a widely used reference approach for post-caesarean analgesia but is associated with neuraxial opioid-related side effects, particularly pruritus and nausea. The transversalis fascia plane (TFP) block is a relatively recent regional technique targeting the transversalis fascia; however, comparative effectiveness data in the obstetric setting remain limited. This study aimed to evaluate whether TFP block provides analgesic outcomes broadly similar to those observed with ITM with respect to 24 h rescue analgesic requirement following elective caesarean delivery under spinal anesthesia, and to compare patient-centered recovery quality. Methods: This prospective, single-center, observational comparative cohort study was conducted in a tertiary obstetric unit. Sixty American Society of Anesthesiologists (ASA) physical status II parturients undergoing elective caesarean delivery under spinal anesthesia were included. Postoperative analgesic strategy was determined according to routine clinical practice, with patients receiving either ITM (100 µg) or bilateral ultrasound-guided TFP block with 0.25% bupivacaine. The primary outcome was cumulative tramadol consumption within the first 24 h postoperatively. Secondary outcomes included Numerical Rating Scale (NRS) pain scores at rest and on movement at 0, 3, 6, 12, and 24 h; postoperative nausea and vomiting (PONV) and pruritus scores; and Obstetric Quality of Recovery-11 (ObsQoR-11) scores at 24 and 48 h. A non-inferiority analytical framework was applied to the primary outcome. Results: Both groups had a median tramadol consumption of 0 mg (interquartile range (IQR) 0–0). Rescue analgesic rates were 23.3% (ITM) versus 16.7% (TFP; OR 0.66, 95% CI 0.18–2.36; p = 0.748). Within an exploratory observational non-inferiority analytical framework, the findings were consistent with non-inferiority of TFP block; however, these analyses should be interpreted as hypothesis-generating rather than confirmatory (risk difference −6.7 percentage points; upper 95% CI +13.5% relative to a prespecified margin of +15%). TFP block was associated with higher ObsQoR-11 scores at 24 h (median 96 vs. 88; p = 0.010; Cliff’s δ = −0.39) and 48 h (median 96 vs. 91; p = 0.017; Cliff’s δ = −0.36). Pruritus at 6 and 12 h was nominally lower in the TFP group. Conclusions: In this prospective observational cohort, TFP block was associated with analgesic outcomes broadly similar to those observed with ITM, with exploratory differences in patient-centered recovery measures. Within the applied exploratory observational analytical framework, these findings were broadly consistent with non-inferiority; however, given the observational design, results should be interpreted cautiously, and the present study does not establish formal non-inferiority or equivalence. TFP block may represent a potential opioid-sparing option warranting confirmation in adequately powered randomized studies. Full article

Background/Objectives: Immediate patient access to radiology reports has increased the need for communication that patients can understand, yet it remains unclear whether simplifying report language improves comprehension without worsening psychological distress. This study aimed to determine whether AI-based simplification of a brain MRI report improves patient understanding, to assess whether anxiety differs between standard and AI-simplified reports, and to examine the relationships among anxiety, report understanding, and health literacy. Methods: We conducted a minimal-risk, survey-based randomized experimental study using Qualtrics and Amazon Mechanical Turk. A total of 803 participants were randomized 1:1 to view either an original radiology report (control, n = 402) or an AI-simplified version of the same report (intervention, n = 401). The stimulus was a single de-identified brain MRI/MRV report. Primary outcomes were report understanding and post-exposure anxiety, and secondary measures included radiology literacy and general health literacy assessed with the Short Test of Functional Health Literacy in Adults (S-TOFHLA). Between-group differences were analyzed using Mann–Whitney U tests, and associations between variables were examined using correlation analyses. Results: Participants who received the AI-simplified report achieved significantly higher understanding scores than those who viewed the original report (mean 5.78 ± 1.31 vs. 5.61 ± 1.49; p = 0.007). Anxiety scores were similar between groups (mean 3.24 ± 0.84 vs. 3.23 ± 0.85; p = 0.103). A positive correlation was observed between anxiety and general health literacy (r = 0.283, p < 0.001), and report understanding was also positively correlated with anxiety (r = 0.182, p < 0.001). Age was negatively associated with anxiety, whereas income showed a weak positive association. Conclusions: AI-based simplification improved patient understanding of radiology reports but did not reduce anxiety. Greater understanding was associated with higher anxiety, suggesting that clearer language alone may be insufficient to address the emotional burden of reading radiology results without clinical context or reassurance. Full article

Objectives: This study aimed to investigate the impact of the food environment surrounding schools on nutritional status and body mass index (BMI) of children and adolescents, offering insights for developing evidence-based policies to promote healthier school surroundings. Methods: Based on 357,782 physical examination records from 140,578 children and adolescents aged 6 to 19 in the Shenzhen Student Health Surveillance System for the 2018–2025 academic years, this study employed latent class mixed models to analyze BMI Z-score trajectory changes among children and adolescents. Furthermore, multinomial logistic regression and logistic regression models were utilized to examine the association between the number of catering points of interest (POIs) near schools, including total number, fast-food restaurants, pastry shops, and beverage stores, and the nutritional status and BMI trajectories of children and adolescents, respectively. Data from Huairou District, Beijing, was used to verify the applicability of the findings in Northern China. Results: 20.71% of children and adolescents in Shenzhen were overweight or obese, and 44.70% were consistently overweight from 2018 to 2025. The increase in catering POIs around schools was significantly associated with nutritional status and overweight trajectory, with pastry shops having a particularly pronounced effect. Each interquartile range (IQR) change in pastry shop was associated with 4.25% (95% CI: 2.96%, 5.56%) increase in the odds of overweight compared with the normal nutritional group, and with 5.03% (95% CI: 3.62%, 6.45%) increase in the odds of the overweight trajectory compared with the normal weight trajectory. Moreover, schools in above-median GDP regions required more attention. A similar association between the number of catering POIs near schools and long-term overweight among children and adolescents was observed in Huairou District, Beijing. Conclusions: The food environment surrounding schools might play a contributory role in shaping the BMI trajectories of children and adolescents. The study emphasized the importance of focusing on the food environment near schools, providing insights for weight management interventions among children and adolescents as well as healthy urban planning. Full article

Fruit detection and maturity recognition are crucial for intelligent tomato harvesting and management. However, in complex field environments, challenges such as the similarity in color between fruits and leaves, cluttered backgrounds, and severe occlusions significantly hinder accurate tomato detection. To address these issues, this paper proposes a lightweight tomato maturity detection model, termed FDA-YOLO. Building upon the YOLOv11 framework, the proposed model enhances global perception in complex scenarios by introducing a multiscale feature enhancement module. In addition, a foreground–background dual-path attention mechanism is designed to better distinguish fruits from the background, thereby improving detection robustness. Furthermore, a lightweight asymmetric detection head is constructed to reduce computational cost while maintaining high accuracy. These improvements enable the model to achieve more efficient and accurate tomato maturity detection under complex conditions. Extensive experiments are conducted on the LaboroTomato dataset. The results demonstrate that FDA-YOLO achieves the best performance with relatively low computational overhead, reaching 83.4% and 67.5% in mAP50 and mAP50–95, respectively, while also attaining a near-optimal F1 score. Overall, the proposed model achieves an excellent balance between accuracy and efficiency, providing an effective solution for intelligent agricultural monitoring and automated harvesting systems. Full article