Background/Objectives: Accurate post-mortem interval (PMI) estimation becomes increasingly difficult when bodies decompose under extreme heat. Hyperthermal Mediterranean environments accelerate soft-tissue degradation, induce early mummification, and distort classical thanatological indicators, often resulting in substantial PMI overestimation. This study analyzes three forensic cases affected by climate-driven decomposition anomalies and presents a climate-adaptive, AI-assisted diagnostic framework applied uniformly across all cases to improve PMI interpretation. Methods: A retrospective case series analysis was conducted on three individuals recovered during summer heatwaves. Crime scene investigation, post-mortem computed tomography (PMCT), autopsy, and genetic identification were integrated with 5–15-year meteorological datasets. Classical PMI estimations were compared with circumstantial data. A multimodal AI model, incorporating environmental features, decomposition morphology, and microenvironmental modifiers, was operationalized for each case using a hybrid Random Forest–LSTM architecture. Engineered indices included Accumulated Degree Days (ADD), a Decomposition Index, and climate-stress metrics (Thermal Load Index, Desiccation Pressure Factor, Microenvironmental Distortion Coefficient). Quantile regression provided calibrated prediction intervals. Results: Morphological assessments overestimated PMI in every case, suggesting intervals of 1–6 months despite true PMIs of approximately 20 days (Cases 1–2) or 36–48 h (Case 3). The AI model yielded conceptual outputs more consistent with verified PMIs, ~21 days (Case 1), ~23 days (Case 2), and ~42 h (Case 3), each accompanied by 50% and 90% prediction intervals. Explainability analyses identified thermal load, desiccation pressure, and microenvironmental distortion, particularly insulation in Case 3, as dominant drivers. Conclusions: Extreme heat fundamentally alters decomposition trajectories, rendering classical PMI methods unreliable. Applying a climate-aware, AI-assisted diagnostic framework across all three cases improved interpretability, providing uncertainty-aware estimates aligned with true PMIs. The AI framework is presented as a conceptual, non-trained, proof-of-concept system, and reported outputs represent operational demonstrations rather than validated predictions, offering a promising foundation for next-generation PMI diagnostics in hyperthermal forensic settings.

ABSTRACT Everyday communication is dynamic and multisensory, often involving shifting attention, overlapping speech, and visual cues. Yet, most neural attention tracking studies are still limited to highly controlled lab settings, using clean, often audio‐only stimuli and requiring sustained attention to a single talker. This work addresses that gap by introducing a novel dataset from 24 normal‐hearing participants. We used a wearable electroencephalography (EEG) system (44 scalp electrodes and 20 cEEGrid electrodes) in an audiovisual (AV) paradigm with three conditions: sustained attention to a single talker in a two‐talker environment, attention switching between two talkers, and unscripted two‐talker conversations with a competing single talker. Analysis included temporal response functions (TRFs) modeling, optimal lag analysis, selective attention classification with decision windows ranging from 1.1 to 35 s, and comparisons of TRFs for attention to AV conversations versus side audio‐only talkers. Key findings show significant differences in the attention‐related P2 peak between attended and ignored speech across conditions for scalp EEG. Interestingly, our results revealed strong cross‐condition generalization, with models trained in one condition maintaining good performance when evaluated on the other two. No significant change in performance between switching and sustained attention suggests robustness for attention switches. Optimal lag analysis revealed a narrower peak for conversation compared to single‐talker AV stimuli, reflecting the additional complexity of multi‐talker processing. Classification of selective attention was consistently above chance (55%–70% accuracy) for scalp EEG, whereas cEEGrid data yielded lower correlations, highlighting the need for further methodological improvements. These results demonstrate that wearable EEG can reliably track selective attention in dynamic, multisensory listening scenarios and provide guidance for designing future AV paradigms and real‐world attention tracking applications.

Small cell lung cancer (SCLC) is a highly aggressive neuroendocrine tumor that typically arises centrally in individuals with a significant smoking history. It is characterized by rapid progression and early metastatic spread, most commonly to the mediastinal lymph nodes, liver, bones, adrenal glands, and brain.1 SCLC is also known to be associated with several paraneoplastic syndromes, including Cushing’s syndrome, Lambert-Eaton myasthenic syndrome, and syndrome of inappropriate antidiuretic hormone secretion (SIADH). Metastasis to the breast from SCLC is exceedingly rare, especially as an isolated site of metastases. Only sporadic cases have been reported in the literature, particularly as an isolated site of spread2. A 45-year-old female with a 30-pack-year smoking history presented with dyspnea. CT chest showed a right lower lobe mass measuring 5.7 cm and an adjacent right hilar component measuring 4.8 x 4.2 x 5.6 cm, with extensive hilar lymphadenopathy and bronchovascular encasement, concerning for malignancy. Bronchoscopy with endobronchial ultrasound-guided biopsy and partial debulking of the right main bronchus mass confirmed SCLC. Staging imaging revealed a suspicious mass in the left breast. Ultrasound of the left breast showed an irregular, hypoechoic, solid lesion with lobulated margins measuring 2.5 x 2.4 cm, without associated axillary lymph node involvement. MRI of the brain showed no additional sites of metastases, indicating the breast as an isolated extrathoracic site of disease. Core needle biopsy of the breast lesion confirmed metastatic small cell carcinoma, with immunohistochemical staining positive for TTF-1, synaptophysin, and chromogranin, supporting lung origin. The patient was subsequently started on systemic chemotherapy. This case highlights a rare instance of isolated breast metastasis from SCLC. Given the unusual site, new breast lesions in patients with lung cancer should raise the suspicion for metastases, even in the absence of widespread disease. Early biopsy is essential to delineate a (second) synchronous primary vs unusual site of metastases to help guide therapy, preventing misdiagnosis as a primary breast malignancy. This abstract is funded by: None

OBJECTIVES Community-acquired bloodstream infections (CA-BSIs) are a major cause of mortality in older adults, yet prospective international data are limited. This study evaluated clinical features, microbiology, and mortality predictors in older patients with CA-BSI. METHODS This prospective, observational study was conducted in 72 referral centers across 16 countries between June 1 and September 1, 2024. Patients aged ≥65 years with confirmed CA-BSI were included. Demographic, clinical, laboratory, and treatment data were collected. The primary outcome was 30-day all-cause mortality. Univariate and multivariate logistic regression was used to identify predictors. RESULTS 574 patients were enrolled (median age 74 years; 48.1% female), and 30-day mortality was 26.3%. Urinary tract infection was the most common source (33.7%), and 30.5% had no identifiable origin. Gram-negative organisms predominated (58.6%). Independent mortality predictors included SOFA score ≥5, thrombocytopenia (<50,000/µL), Staphylococcus aureus bacteremia, pneumonia, bacteremia of unknown origin, and higher Charlson Comorbidity Index. Higher Katz Activities of Daily Living scores were protective. CONCLUSION CA-BSI in older adults is associated with high mortality. Outcomes are driven by comorbidity burden, functional status, infection severity, and causative pathogen, highlighting the need for early risk stratification and tailored management.

Hydatidiform mole, the most common form of gestational trophoblastic disease, presents as complete or partial form. Complete hydatidiform mole (CHM) usually presents with prominent histological criteria, whereas partial hydatidiform mole (PHM) shows a wide spectrum of presentations depending on gestational age. Molecular markers enhance the understanding of the variation and heterogeneous presentation of molar pregnancies, as well as their biological potential and behavior. This retrospective study included 50 CHM and 50 PHM specimens, terminated in first trimester via suction curettage. A second histopathological review of slides stained with hematoxylin and eosin was conducted, as well as a selection of representative tissue slides for p53 immunostaining. This study aimed to determine the precise correlation between specific morphological criteria and the patterns of p53 immunoexpression. Semi-quantitative analysis of samples for both pathological criteria and p53 immunostaining was performed. p53 positivity was defined as follow: the percentage of positive cells/nuclei: + (10-40%); ++ (40-70%); +++ (>70%); and staining intensity was scored as: 1 – weak, 2 – moderate, and 3 – strong intensity. P53 expression was estimated on at least 200 nuclei of cytotrophoblasts per slide. Significant difference in p53 expression exist between CHM and PHM in staining intensity. CHM shows significant correlation of p53 positivity with hydrops, central cisterns and atypia. In PHM trophoblast pseudoinclusions demonstrate strong significant correlation with p53 positivity. The irregular pseudoinclusions demonstrate lower expression compared with round or oval, being consistent with benign behavior of PHM. Prominent morphological criteria strongly correlate with p53 immunoexpression for both CHM and PHM.

This study examines the key competencies required of teachers in contemporary education and their role in preparing future generations. These competencies are widely recognized as foundational to societal development across multiple domains. In this context, the 4Cs model in initial teacher education was developed to establish conceptual and pedagogical preconditions for integrating critical thinking, creativity, collaboration, and communication competencies into teacher education programs. The study investigated the perceptions of students enrolled in teacher education programs at the University of Sarajevo regarding instructional activities through which university teachers foster and develop 4Cs competencies in higher education teaching. The findings indicate that, from the students’ perspective, instructional activities most strongly support the development of communication competency, followed by critical thinking and collaboration, whereas creativity is the least consistently fostered. This finding suggests that the integration of 4Cs competencies in higher education remains predominantly declarative rather than transformative, revealing a gap between curricular expectations and pedagogical practice. Furthermore, the results indicate a highly significant correlation between instructional activities fostering critical thinking and creativity. Significant correlations were also observed among the other competency scales, indicating that these competencies are interrelated and mutually reinforcing rather than developed in isolation. These findings highlight practical implications for teacher education programs.

Background: Osteoarthritis (OA) is a heterogeneous joint disorder traditionally considered mechanically driven; however, evidence indicates that inflammatory mechanisms contribute to symptom expression. Exploratory analyses of peripheral biomarkers may provide insights into systemic inflammation in symptomatic knee OA, but formal phenotypic validation requires dedicated clustering or longitudinal studies. Objective: To examine associations between clinical pain, functional impairment, and circulating inflammatory biomarkers in patients with knee OA compared with healthy controls. Methods: In this prospective, single-center study, patients aged 40–80 years with radiographically confirmed knee OA and chronic knee pain were compared with age- and sex-matched healthy controls. Pain intensity and functional status were assessed using the Visual Analogue Scale (VAS) and the Knee Injury and Osteoarthritis Outcome Score (KOOS). Circulating inflammatory biomarkers, including cytokines and matrix metalloproteinases, were quantified using multiplex immunoassays. Statistical analyses included adjusted linear regression models, with age and BMI as covariates, and multiple testing correction using the Benjamini–Hochberg procedure (FDR alpha error 5%). Results: OA patients exhibited higher circulating levels of TNF-α, IL-6, IL-8, MMP-1, MMP-3, TNFSF13, TNFSF13B, and pentraxin-3 compared with controls (p < 0.01). No significant sex differences were observed. KOOSs correlated with IL-6 and IL-10 levels, suggesting an association between systemic inflammatory activity and functional limitation. All findings are presented as exploratory and associative. Conclusions: Patients with knee OA display systemic inflammatory biomarker differences associated with pain and functional impairment. These results support the role of inflammation in OA symptoms within an exploratory framework. Larger, longitudinal studies are warranted to validate these observations.

This paper presents two offline, on-premise NLP proof-of-concept assistants built on a shared architecture for internal knowledge access in the Central Bank of Bosnia and Herzegovina: (i) a semantic document search tool for internal Word/PDF repositories and (ii) an HR chatbot that applies retrieval-augmented generation (RAG) over indexed HR policies and procedures. Rather than proposing a novel NLP method, the paper contributes by documenting a reusable offline architecture for institutional AI assistants in a security-constrained central banking environment and by providing pilot evidence on how established semantic retrieval and RAG techniques can be adapted to strict requirements of confidentiality, data sovereignty, and governance. The semantic search assistant combines exact phrase matching with embedding-based retrieval and hybrid re-ranking, while the HR chatbot generates source-grounded answers using locally hosted language models under explicit governance constraints, including B/H/S-only output, strict fallback behaviour, and transparent display of retrieved passages. Pilot results indicate that hybrid retrieval offers the most reliable performance across representative internal queries, while the HR chatbot demonstrates the feasibility of document-grounded employee support under offline institutional constraints. The findings provide preliminary evidence that offline NLP assistants can improve access to internal institutional knowledge while remaining compatible with the security and operational risk requirements typical of central banking environments.

The Sarajevo ion accelerator (SARAI) benefits from the successful transfer of the CERN’s 750 MHz radiofrequency quadrupole (RFQ) technology, adapted for societal and medical applications. SARAI incorporates an industrial electron cyclotron resonance supernanogan ion source operating at 14.5 GHz, designed to generate alpha particles and ions that can be used for ion beam analysis and experimental and applied physics research. The system is equipped with a 4-electrode extraction subsystem capable of extracting and focusing multiple ion species at voltages up to 30 kV. The source is being optimized at CERN for injection into a 750 MHz RFQ, initially with protons, followed by helium and carbon ions. This paper reports on the progress of a novel idea of integrating a supernanogan ECR ion source with the RFQ to get higher currents than currently achievable in the emerging domain of compact linacs for societal applications. The project is in its infancy, so this paper focuses on the source commissioning preparation, reports on the initial high voltage holding tests on the source extraction system in vacuum, and discusses further plans and applications.

This study investigates early‐age carbonate mineralization in cementitious systems using in situ Raman microspectroscopy. In the presence of dissolved CO 2 , clinker phases undergo accelerated dissolution, decomposing to form various calcium carbonate polymorphs and a transient amorphous silica gel network. Once the available CO 2 is consumed, clinker hydration resumes, leading to delayed calcium–silicate–hydrate (C–S–H) and portlandite (Ca(OH) 2 ) formation. The precipitation of portlandite through the pore network triggers a localized pozzolanic reaction at the silica gel–portlandite interface, yielding a distinct calcium–silicate–hydrate (C–S–H*). This templated mechanism produces a homogeneous and highly polymerized binder, leading to improved 24‐h compressive strength compared to reference samples. Correlation function analysis confirms that the evolution of silica gel, portlandite, and C–S–H follows a three‐stage sequence— mineralization , transition , and stabilization— quantitatively demonstrating a strong spatial anticorrelation between silica gel dissolution and portlandite precipitation. These findings establish a new chemomechanical framework for CO 2 mineralization in cement, highlighting transient silica gel as a critical intermediate for engineering sustainable, high‐performance concrete.