University of the Faroe Islands

BACKGROUND:The garden dormouse (Eliomys quercinus) is one of the fastest-declining mammals in Europe, and action is needed to prevent further population losses. The primary causes of declines are not well-understood, as the species experiences variable conditions and threats across its range, but likely include habitat fragmentation and loss. Previous genetic studies have provided evidence of highly structured garden dormouse populations in Western Europe, despite this region having been defined as a single clade with mitochondrial DNA analysis. Within Western Europe, the magnitude of declines has been recognized to be greater on the eastern edge of the species' range, which could explain differentiation within the clade as resulting from diversity loss and genetic drift for regions under greater risk of extirpation. Here, we focus on fine-scale genomic differentiation across the Western European clade to explore the consequences of genomic erosion on the eastern region and to help identify mechanisms driving genetic differentiation within this species. RESULTS:We found genetic differentiation both between and within major geographic regions. Populations located in the eastern edge of the species' range showed stronger signs of population isolation, including structure between spatially distant populations, lower genetic diversity, and greater rates of inbreeding. However, all populations exhibited signals of recent rapid population decline. Outlier analyses indicated that differentiation between regions was primarily due to genetic drift resulting from isolation-by-distance rather than adaptive differentiation. We also found genetic structuring between populations within the Rhine Valley, despite apparent lack of physical barriers preventing dispersal among groups within this region. CONCLUSIONS:Our findings indicate that population isolation following habitat loss and fragmentation has likely been a major contributor to garden dormouse declines. Dispersal among disparate garden dormouse sampling regions is restricted-even across local spatial scales-leading to loss of genetic diversity and potential erosion of evolutionary potential. With 21st century declines expected to continue across the species' range, even relatively common and well-connected populations are likely to follow the trajectory of the eastern populations, with increasing loss of diversity as populations contract and become more isolated.

Detecting Streptococcus pneumoniae traditionally relies on complex and time-consuming culture methods, limiting sensitivity, especially in samples with low bacterial loads. Molecular diagnostics, particularly real-time PCR (qPCR), have significantly improved detection thresholds compared to culture. This pilot study aimed to assess the sensitivity and specificity of a duplex droplet digital PCR (ddPCR) assay targeting pneumococcal-specific genes lytA and piaB in nasopharyngeal swabs from asymptomatic children. A total of 165 nasopharyngeal swab samples were collected, of which ten samples were excluded during quality control, leaving 155 for analysis. These included 73 culture-positive (47%) and 82 culture-negative (53%) samples. The ddPCR analysis yielded 105 samples (68%) positive for both lytA and piaB, representing a statistically significant increase in detection compared to culture (McNemar’s test, p < 1 × 10⁻⁶). An additional 12 samples exhibited single-gene positivity (lytA only: 7, piaB only: 5) but were not classified as definitive positives due to specificity concerns. Duplex ddPCR demonstrated enhanced sensitivity, particularly in samples with low pneumococcal density, and the requirement of dual-target positivity ensured high specificity. These results underline duplex ddPCR as a promising and robust diagnostic tool for pneumococcal carriage detection, with potential implications for epidemiological surveillance and clinical diagnostics. Future research directions include assay optimization, quantitative bacterial load analysis, and evaluation of the method’s applicability for rapid pneumococcal detection in invasive disease contexts.

Although deep learning has significantly improved the efficiency of medical image diagnosis, its inherent “black-box” nature hinders clinical trust. To address this, we propose an interpretable diagnostic framework based on prototypical concept representation. Targeting the semantic instability of weak-boundary and complex lesions, we first construct a multi-scale prototype library that fuses mid-level texture and high-level semantic features to simultaneously characterize local details and global structural patterns. Second, to enhance lesion localization accuracy, we design a frequency-constrained adaptive feature fusion strategy. This module dynamically selects the optimal scale and effectively suppresses background-induced pseudo-activations. Third, we introduce concept consistency and semantic alignment strategies to rectify anomalous prototype activations, thereby optimizing the class evidence distribution. Crucially, breaking the bottleneck of traditional correlation-based methods, we implement concept-level counterfactual perturbation within the prototype space to verify decision fidelity. By explicitly observing the impact of key semantic concept interventions, we ensure the credibility of the explanation path. Trained via a decoupled supervision protocol, experiments on VinDr-CXR and NIH ChestX-ray14 datasets demonstrate that our method significantly improves localization compactness and interpretation robustness while maintaining outstanding classification performance.

Transitions from early childhood education and care (ECEC) to school and after-school care (ASC) mark one of the most formative phases in children’s educational pathways. These transitions involve changes in relationships, routines, and pedagogical expectations that can both nurture and challenge continuity of learning. This hermeneutic literature review examines how transitions between ECEC, school, and ASC shape opportunities for learning, with a particular focus on how continuity is conceptualised and supported across settings. Guided by a hermeneutic model, the review followed iterative cycles of searching, reading, and interpretation to identify and connect central ideas within international research on early years transitions and their influence on continuity of learning. Through an ecological framework and later bioecological developments, the analysis revealed four interconnected themes: (1) increasing complexity in transition arrangements; (2) evolving roles in scaffolding and support across institutions; (3) the growing influence of schoolification on early learning; and (4) children’s agency as active brokers of continuity. Taken together, these themes portray continuity of learning as a relational and ecological achievement sustained through collaboration, play, and children’s participation. The review offers an integrative conceptual framework for understanding transitions as multidimensional processes and provides a theoretical foundation for future empirical inquiry.

This dataset provides Automatic Identification System (AIS) messages and per-message radio-frequency metadata captured in Tórshavn Harbour and the surrounding islands (Faroe Islands) over 76 days in 2025. The corpus contains 7.8 million messages from 547 vessels, with detections observed out to approximately 101 NM. Each record combines decoded AIS payloads with radio-frequency features—received signal power (dBFS) and frequency offset in parts-per-million (PPM)—enabling physical-layer characterisation, demodulation benchmarking, vessel analytics, and spoofing/interference studies. Signals were received using an RTL-SDR and a vertically-polarised VHF antenna with fixed 20 dB gain (AGC off); decoding used AIS-catcher (v0.61). Fixed gain ensures temporal consistency of signal power, while the PPM field provides a coarse indicator of oscillator drift and Doppler. The data are released as newline-delimited JSON in a single file (aisdata.csv) with a public DOI. This resource supports research across RF propagation and maritime data science, and facilitates reproducible benchmarking under real-world North Atlantic conditions.