Search Results (256)

Background: Chest X-ray (CXR) is widely used for the assessment of thoracic diseases, yet automated multi-label interpretation remains challenging due to subtle visual patterns, overlapping anatomical structures, and frequent co-occurrence of abnormalities. While recent deep learning models have shown strong performance, limitations in interpretability, anatomical awareness, and robustness continue to hinder their clinical adoption. Methods: The proposed framework employs a hybrid ConvNeXtV2–Vision Transformer (ViT) architecture that combines convolutional feature extraction for capturing fine-grained local patterns with transformer-based global reasoning to model long-range contextual dependencies. The model is trained exclusively using image-level annotations. In addition to classification, three complementary post hoc components are integrated to enhance model trust and interpretability. A segmentation-aware Gradient-weighted class activation mapping (Grad-CAM) module leverages CheXmask lung and heart segmentations to highlight anatomically relevant regions and quantify predictive evidence inside and outside the lungs. An ontology-driven neuro-symbolic reasoning layer translates Grad-CAM activations into structured, rule-based explanations aligned with clinical concepts such as “basal effusion” and “enlarged cardiac silhouette”. Furthermore, a lightweight out-of-distribution (OOD) detection module based on confidence scores, energy scores, and Mahalanobis distance scores is employed to identify inputs that deviate from the training distribution. Results: On the VinBigData test set, the model achieved a macro-AUROC of 0.9525 and a Micro AUROC of 0.9777 when trained solely with image-level annotations. External evaluation further demonstrated strong generalisation, yielding macro-AUROC scores of 0.9106 on NIH ChestXray14 and 0.8487 on CheXpert (frontal views). Both Grad-CAM visualisations and ontology-based reasoning remained coherent on unseen data, while the OOD module successfully flagged non-thoracic images. Conclusions: Overall, the proposed approach demonstrates that hybrid convolutional neural network (CNN)–vision transformer (ViT) architectures, combined with anatomy-aware explainability and symbolic reasoning, can support automated chest X-ray diagnosis in a manner that is accurate, transparent, and safety-aware. Full article

Leadership lies at the core of public administration, yet research on boundary-spanning leadership has paid limited attention to the micro-level behaviors through which regional collaboration is enacted. Drawing on empirical evidence from China and a mixed-methods research design, this study examines relational leadership behaviors (RLBs) in regional collaborative governance (RCG). It identifies three types of collaborative leaders—leaders embedded in network administrative organizations, leaders within specialized collaborative departments, and leaders exchanged between regions—and four core RLBs: relational initiative, reconciliation, catalysis, and linkage. These behaviors enhance the perceived effectiveness of RCG by fostering trust, managing conflicts, and integrating diverse interests. The findings further show that RLBs are shaped by the collaborative context, including institutional arrangements, leader roles, task complexity, and the temporal dynamics of collaboration. By incorporating relational leadership into a process-oriented perspective, this study extends RCG theory and offers practical insights for improving governance effectiveness in RCG. Full article

Background: Breast cancer remains a leading cause of mortality among women worldwide, underscoring the need for timely and accurate detection. Conventional mammographic diagnosis, while widely used, is limited by subjectivity and variability in interpretation. Recent advances in deep learning (DL) have improved automated detection; however, the black-box nature of these models raises concerns regarding clinical trust and interpretability. Methods: To address this, we propose an explainable DL framework for breast cancer classification using mammographic images. The approach employs contrast limited adaptive histogram equalization (CLAHE)-based preprocessing to enhance lesion contrast, EfficientNetV2 for feature extraction, and the convolutional block attention module (CBAM) to refine salient features. For interpretability, gradient-weighted class activation mapping++ (Grad-CAM++) is used to highlight discriminative regions influencing predictions. Results: The framework is evaluated on three publicly available datasets—MIAS, DDSM, and InBreast—individually and under cross-dataset settings. Results demonstrate superior performance over existing methods, achieving classification accuracies of 99.85%, 99.40%, and 99.70% on MIAS, DDSM, and InBreast, respectively, with corresponding F1-scores of 99.75%, 99.10%, and 99.55%. Confusion matrix analysis confirms excellent sensitivity for malignant cases, and time complexity assessments show reduced training and inference overhead compared to conventional deep models. Conclusions: The framework thus provides a robust and interpretable solution for mammogram-based breast cancer screening. Full article

Agricultural producers experience elevated stress, limited mental health access, and cultural norms that can discourage help-seeking. This study examined farmers’ preferences for receiving mental health information and the barriers that impede care. Data came from a regional needs assessment of 1024 producers across 12 Midwestern states who completed online or paper surveys, including questions on willingness to seek or receive information and the 30-item Barriers to Access to Care Evaluation. Analyses included descriptive, bivariate, and multivariate methods to explore demographic and behavioral predictors. Results indicated that while 74.1% were open to receiving mental health information, notable proportions were unwilling to seek (27.8%) or receive (28.4%) it, and 18.7% were unwilling to do either. Preferred sources were medical providers, mental health professionals, and family members, with agricultural retailers least favored. Women, younger producers, veterans, those with mental health symptoms, and individuals with higher education, anxiety, or depression showed distinct patterns of openness and barrier endorsement. Attitudinal barriers were the most common across groups. Findings highlight the importance of culturally relevant approaches that leverage trusted messengers, reduce stigma, and tailor interventions to specific subgroups to strengthen mental health outreach in agricultural communities. Full article

Extractive regions are among the most visible frontlines of the Anthropocene as they are areas where the environmental and social consequences of intensive resource exploitation are concentrated. In Serbia, mining areas such as Bor and Majdanpek represent complex socio-spatial assemblages in which everyday life, work, and governance intersect under pressures of neoliberal development and ecological degradation. This study aims to identify the challenges and opportunities for citizen participation in mining regions, providing guidance on enhancing participatory processes in decision-making. To operationalise this aim, the study pursues three objectives: (1) to assess residents’ awareness, participation practices, access to information, and motivation to engage in planning; (2) to identify perceived barriers and opportunities for participation; and (3) to formulate recommendations for improving participatory and communication processes in extractive-region governance. Accordingly, the research is guided by the main question: How do residents of the Bor–Majdanpek mining region perceive opportunities and barriers to public participation in planning and decision-making processes? To address this question, a face-to-face field survey was conducted in the summer of 2024 with a random sample of residents (N = 300). In this mixed-methods exploratory study, primary survey data were analysed using descriptive and inferential statistical methods. In contrast, open-ended questions were analysed qualitatively to capture respondents’ detailed perceptions and suggestions. Findings indicate limited awareness of planning procedures, low participation experience, and structural barriers related to information access, trust, and institutional responsiveness. At the same time, respondents show a strong interest in more transparent, accessible, and dialogic forms of engagement. This study demonstrates that citizen participation in extractive landscapes is not only a procedural requirement but a mechanism to strengthen democratic governance and rebuild trust. Insights from Bor–Majdanpek provide an evidence base for improving participatory practices in mining regions undergoing socio-environmental transformation. Full article

Environmental factors motivate migration across the globe, calling for better planning. Although the US experienced such movements during the COVID-19 pandemic, literature on population mobility and outcomes for receiving communities in the US is scarce. We use a mixed-methods case study approach to explore the COVID-era population movement trends in the US Northeast (NE) Region and their outcomes for receiving communities to draw lessons for strategic regional planning aiming to achieve sustainable and equitable outcomes of disaster-induced movements. Utilizing the Statistics of Income data and focus group data collected from 27 local experts in 22 rural counties of NE, which experienced the highest relative numbers of in-movers between 2016 and 2020, the findings revealed the top receiving counties were predominantly rural areas where urbanites moved from within NE. This movement challenged the housing market and services, disproportionately burdening socioeconomically disadvantaged groups in receiving communities. The COVID-19 experience opened a window of opportunity for regional planning to prepare desirable outcomes of such mobilities by addressing existing issues in receiving communities while incorporating pulse and slow population movements into the agenda. The right policy timing and communication among communities are keys to building trust and ensuring integration of newcomers into receiving communities. Full article

Living Labs (LLs) are key for collaborative and value-based innovation, though their relational and governance mechanisms are still being explored. This study focuses on examining how relational dynamics and community leadership influence the design, governance, and replicability of a Digital Living Labs (DLLs) methodology. The research examines the DLLs of Catalonia using a combination of 15 qualitative interviews and 104 survey responses, with a mixed-methods design adopted. This regional initiative is based on Quadruple Helix (4-H) collaboration and value-driven innovation. The findings show that inclusive participation is enabled through core relational infrastructures. These relationships are built on trust-building, collaboration, facilitation, and knowledge exchange. Community leaders complemented facilitators through harmonizing institutional objectives with local priorities, reinforcing distributed governance, and generating public value. Inclusion, equity, transparency, and solidarity were essential to engagement and collective ownership. The study’s results indicate that effective DLLs transferability depends more on reinforcing relational foundations and shared values than on replicating fixed structures. Full article

In the context of dual-carbon goals, Park-Level Integrated Energy Systems (PIES) are pivotal for enhancing renewable energy integration and promoting clean, efficient energy use. However, the inherent non-linearity from multi-energy coupling and the high dimensionality of operational data present substantial challenges for conventional scheduling optimization methods. To overcome these obstacles, this paper introduces a novel multi-objective scheduling framework for PIES leveraging deep reinforcement learning. We innovatively formulate the scheduling task as a Markov Decision Process (MDP) and employ the Trust Region Policy Optimization (TRPO) algorithm, which is adept at handling continuous action spaces. The state and action spaces are meticulously designed according to system constraints and user demands. A comprehensive reward function is then established to concurrently pursue three objectives: minimum operating cost, minimum carbon emissions, and maximum exergy efficiency. Through comparative analyses against other AI-based algorithms, our results demonstrate that the proposed method significantly lowers operating costs and carbon footprint while enhancing overall exergy efficiency. This validates the model’s effectiveness and superiority in addressing the complex multi-objective scheduling challenges inherent in modern energy systems. Full article

Background: Laryngeal and hypopharyngeal cancers (LHCs) exhibit heterogeneous outcomes after definitive radiotherapy (RT). Large language models (LLMs) may enhance prognostic stratification by integrating complex clinical and imaging data. This study validated two pre-trained LLMs—GPT-4o-2024-08-06 and Gemma-2-27b-it—for outcome prediction in LHC. Methods: Ninety-two patients with non-metastatic LHC treated with definitive (chemo)radiotherapy at Linkou Chang Gung Memorial Hospital (2006–2013) were retrospectively analyzed. First-order and 3D radiomic features were extracted from intra- and peritumoral regions on pre- and mid-RT CT scans. LLMs were prompted with clinical variables, radiotherapy notes, and radiomic features to classify patients as high- or low-risk for death, recurrence, and distant metastasis. Model performance was assessed using sensitivity, specificity, AUC, Kaplan–Meier survival analysis, and McNemar tests. Results: Integration of radiomic features significantly improved prognostic discrimination over clinical/RT plan data alone for both LLMs. For death prediction, pre-RT radiomics were the most predictive: GPT-4o achieved a peak AUC of 0.730 using intratumoral features, while Gemma-2-27b reached 0.736 using peritumoral features. For recurrence prediction, mid-RT peritumoral features yielded optimal performance (AUC = 0.703 for GPT-4o; AUC = 0.709 for Gemma-2-27b). Kaplan–Meier analyses confirmed statistically significant separation of risk groups: pre-RT intra- and peritumoral features for overall survival (for both GPT-4o and Gemma-2-27b, p < 0.05), and mid-RT peritumoral features for recurrence-free survival (p = 0.028 for GPT-4o; p = 0.017 for Gemma-2-27b). McNemar tests revealed no significant performance difference between the two LLMs when augmented with radiomics (all p > 0.05), indicating that the open-source model achieved comparable accuracy to its proprietary counterpart. Both models generated clinically coherent, patient-specific rationales explaining risk assignments, enhancing interpretability and clinical trust. Conclusions: This external validation demonstrates that pre-trained LLMs can serve as accurate, interpretable, and multimodal prognostic engines for LHC. Pre-RT radiomic features are critical for predicting mortality and metastasis, while mid-RT peritumoral features uniquely inform recurrence risk. The comparable performance of the open-source Gemma-2-27b-it model suggests a scalable, cost-effective, and privacy-preserving pathway for the integration of LLM-based tools into precision radiation oncology workflows to enhance risk stratification and therapeutic personalization. Full article

The main objective of this article is to conduct a literature review on the use of open innovation (OI) for green transition to identify tools and methods that can make green transition more effective, efficient, and socially acceptable. This review is accompanied by an attempt to answer the following research questions: R1. How can open innovation be used in the economy and by individual entities to achieve the goals of the green transition? R2. How can individual stakeholders be activated and motivated to participate in the process of creating open innovation for the green transition? and R3. What are the real effects of using open innovation on a macroeconomic, social, and individual scale? The results allow concluding that OI is used by enterprises, cities, regions, and entire economies. Among the methods of activating and motivating individual stakeholders to engage in the process of creating OI for green transition, the following can be selected: (1) internal resources and competencies (knowledge management, internal programs, open leadership, trust, complementarity of resources); (2) partnership characteristics (modern business models, involvement of partnership intermediaries, strengthening relationships with suppliers and customers, involvement of prosumers, cooperation with universities and research institutions); (3) external legal and regulatory conditions (protection of intellectual property rights, pro-innovation and pro-environmental education systems, creation of a legal framework for cooperation between science and business); and (4) external technical and organizational solutions (online platforms, social media, Living Labs, external sources of knowledge). The most frequently mentioned individual effects of open innovation in the energy sector include: improved efficiency, effectiveness and competitiveness in environmental management and the implementation of sustainable development, as well as the use of modern technologies. At the economic level, OI supports investment and economic growth. It can also have a positive impact on reducing energy poverty and developing renewable energy sources, including in emerging economies. This form of innovation also promotes social integration and the creation of social values. The findings of this review can be utilized by scholars to identify current and future research directions. They may also prove valuable for practitioners as both an incentive to engage in open innovation and guidance for its design and implementation. Furthermore, the results can contribute to disseminating knowledge about open innovation and its role in the green transformation. Full article

Background/Objectives: In clinical practice, Explainable AI (XAI) enables non-specialists and general practitioners to make precise diagnoses. Current XAI approaches are limited, as many rely solely on either presenting explanations of clinical data or presenting explanations of MRI, or presenting explanations in unclear ways, reducing their clinical utility. Methods: In this paper, we propose a novel Hybrid Explainable AI (HXAI) framework. This framework uniquely integrates both model-agnostic (SHAP) and model-specific (Grad-CAM) explanation methods within a unified structure for the diagnosis of Alzheimer’s disease. The dual-layer explainability constitutes the main originality of this study, as it provides the possibility of interpreting quantitative (at the feature level) and spatial (at the region level) data within a single diagnostic framework. Clinical features (e.g., Mini-Mental State Examination (MMSE), normalized Whole Brain Volume (nWBV), Socioeconomic Status (SES), age) are combined with MRI-derived features extracted via ResNet50, and these features are integrated using ensemble learning with a logistic regression meta-model. Results: The corresponding validation reflects the explainability accuracy of these feature-based explanations, with removal-based tests achieving 83.61% explainability accuracy, confirming the importance of these features. Model-specific information was used to explain MRI predictions, achieving 58.16% explainability accuracy of visual explanations. Conclusions: Our HXAI framework integrates both model-agnostic and model-specific approaches in a structured manner, supported by quantitative metrics. This dual-layer interpretability enhances transparency, improves explainability accuracy, and provides an accurate and interpretable framework for AD diagnosis, bridging the gap between model accuracy and clinical trust. Full article

Background: Commissioning of ‘joint care’ across teenage and young adult (TYA) principal treatment centres (PTC) and regional designated hospitals was introduced to enable cancer care closer to home, while providing support through the TYA multidisciplinary team. We aimed to explore the processes being used to enable inter-organisational collaboration under joint care models through rapid ethnography. Methods: Healthcare professionals in TYA PTCs in England and Wales between June 2022 and December 2023 were identified by the TYA lead in each PTC as delivering TYA cancer care. Semi-structured interviews were conducted virtually or by telephone based on the structuration model of collaboration proposed by D’Amour. Data were analysed against the model through framework analysis. Results: Our study highlighted variation across the different dimensions of inter-organisational collaboration. We found that healthcare professionals delivering TYA cancer care were working toward a shared goal but this was not always achieved. Social interaction between professionals was required to develop relationships and trust, but opportunities for social interaction were not regularly available. Processes for sharing information were not streamlined, so there were instances when information could not be shared between organisations. Interventions to achieve coordinated care, such as an outreach team, supported the delivery of joint care but these were not available in every region. While there were some levels of leadership within aspects of services, there were limited examples nationally or across geographical regions, which hindered the development of coordinated care. Conclusions: Coordination of care is mostly developing; however, the shared vision and goals dimension did achieve full active collaboration. The implementation of a service specification will address regional leadership requirements, but resources are required to extend the delivery of interventions to support coordination and collaboration, allowing the commissioned model of care to be delivered safely. Full article

Background: The use of pesticides among smallholder farmers, agrochemical sellers, and agricultural officers involves a complex interplay of knowledge, economic factors, and regulatory frameworks. Therefore, this study explores the patterns, practices, and socio-environmental dynamics of pesticide use among smallholder farmers and agro-input sellers in Iringa and Njombe. Method: This study employed a qualitative, phenomenological design, guided by the socio-ecological model (SEM), to explore the lived experience of farmers, agro-dealers, and extension officers. It involved a total of 23 interviews performed in the Njombe and Iringa regions. Data were collected between October 2024 and March 2025, using a combination of in-depth phenomenological interviews, key informant interviews, and field observations, and were categorized into themes and subthemes analyzed using InVivo. Results: The study involved a total of 23 participants drawn from the Iringa and Njombe regions. The gender distribution was nearly balanced, with 52.1% male and 47.8% female respondents. The mean age of participants was 33 years (95% CI: 29.3–37.3). In terms of education, over half (52.17%) had completed primary school. The findings show that smallholders in Iringa and Njombe widely use mixed pesticides and fertilizers, rely on trusted brands, and adapt to climate impacts, but face challenges with regard to unsafe mixing, poor storage, fake products, and weak regulation, highlighting the need for better education, market oversight, and safer practices. Conclusion: Using the socio-ecological model, the findings indicate that pesticide use among smallholder horticultural farmers in Iringa and Njombe is influenced by a complex interaction of socio-economic constraints, market forces, climate variability, and institutional shortcomings. Although farmers have some awareness of safe practices, systemic barriers continue to limit the adoption of sustainable pesticide management. Full article

Alzheimer’s disease (AD) is a neurodegenerative condition that gradually deteriorates memory and cognitive abilities, posing a significant global health challenge. While convolutional neural networks (CNNs) applied to structural magnetic resonance imaging (MRI) have achieved high diagnostic accuracy, their decision-making processes often lack transparency, which can limit clinical trust. This study presents a structured evaluation framework by applying multiple gradient-based and model-agnostic interpretability methods, such as Grad-CAM, Grad-CAM++, HiRes-CAM, Backpropagation, Guided Backpropagation, Kernel SHAP, LIME, and RISE, to pre-trained and custom CNN architectures for AD classification. We utilized the ADNI MRI dataset and assessed models based on accuracy, sensitivity, specificity, and visual alignment of highlighted brain regions with established biomarkers. By analyzing both predictive performance and explanation validity, this study aims to assist clinicians in making informed diagnoses, ultimately strengthening trust in AI-assisted tools. Full article

Background: Early and accurate cancer detection remains a critical challenge in global healthcare. Deep learning has shown strong diagnostic potential, yet widespread adoption is limited by dependence on high-performance hardware, centralized servers, and data-privacy risks. Methods: This study introduces a browser-based multi-cancer classification framework that performs real-time, client-side inference using TensorFlow.js—eliminating the need for external servers or specialized GPUs. The proposed model fine-tunes the Xception architecture, leveraging depthwise separable convolutions for efficient feature extraction, on a large multi-cancer dataset of over 130,000 histopathological and cytological images spanning 26 cancer types. It was benchmarked against VGG16, ResNet50, EfficientNet-B0, and Vision Transformer. Results: The model achieved a Top-1 accuracy of 99.85% and Top-5 accuracy of 100%, surpassing all comparators while maintaining lightweight computational requirements. Grad-CAM visualizations confirmed that predictions were guided by histopathologically relevant regions, reinforcing interpretability and clinical trust. Conclusions: This work represents the first fully browser-deployable, privacy-preserving deep learning framework for multi-cancer diagnosis, demonstrating that high-accuracy AI can be achieved without infrastructure overhead. It establishes a practical pathway for equitable, cost-effective global deployment of medical AI tools. Full article