Ourania Petropoulou

Background/Objectives: High-resolution Human Leukocyte Antigen (HLA) typing is required for unrelated donor selection, yet many registry records remain low- or intermediate-resolution. Haplotype frequency imputation is the standard remedy, but its accuracy decreases for rare haplotypes. Methods: A locus-specific Extreme Learning Machine (ELM) framework, trained on the Greek national donor registry (n = 117,345), upgrades low-/intermediate-resolution genotypes and imputes untyped loci; output weights are solved analytically. It was validated on two independent Greek cohorts (n = 20,100; 4353) against GRIMM and two expectation maximization baselines, on the same donors. Results: Accuracy depended on the high-resolution context available (94% for HLA-A with four anchor loci; 19–22% with none) and on the posterior probability threshold (78–91% ORAM, 70–84% GRPT), while call rate fell from 98.1% to ~41%; the two cannot be maximized together. GRIMM called fewer donors (1.8–47.9%) at comparable or higher accuracy. Allele-vocabulary abridgment excluded 14–47% of external-cohort allele calls; stratified by vocabulary membership, ORAM accuracy was 94.2% for in-vocabulary donors and 73.7% for the rest. The training procedure has no convergence guarantee. Conclusions: ELM upgrading converts legacy registry data into probabilistic high-resolution calls, with registry-scale training completing in minutes. These calls can narrow donor searches but cannot substitute for confirmatory typing.

Background/Objectives: High-resolution Human Leukocyte Antigen (HLA) typing is required for precise unrelated donor selection, yet many registry records remain low- or intermediate-resolution. Haplotype-frequency imputation is the standard remedy but loses accuracy for rare haplotypes and underrepresented populations. Methods: A locus-specific Extreme Learning Machine (ELM) framework, trained on the Greek national donor registry (n = 117,345), upgrades low/intermediate-resolution genotypes to high-resolution calls and imputes untyped loci; output weights were solved analytically via the Moore-Penrose generalized inverse. Validation used five-fold cross-validation plus two independent Greek cohorts (ORAM, Orama Elpidas Donor Center, n = 20,100; GRPT, Evangelismos Hospital Donor Center, n = 4,353), benchmarked against GRIMM, a HaploStats-style baseline, and an in-house expectation-maximization (EM) baseline. Results: Accuracy depended on the high-resolution context available (94% for HLA-A with four anchor loci; 19-22% per locus with none) and on the posterior-probability threshold (78-91% ORAM, 70-84% GRPT); call rate fell to ~41% at the highest threshold, so the two cannot be maximized together. GRIMM called far fewer donors (1.8-47.9%) at comparable or higher first-field accuracy. Allele-vocabulary abridgment excluded 14-47% of external-cohort allele calls. Conclusions: ELM upgrading converts legacy registry data into probabilistic high-resolution calls, with registry-scale training completing in minutes. These calls should prioritize and narrow donor searches, not substitute for confirmatory typing before selection.

Clinical question answering over electronic health records (EHRs) increasingly relies on large language model (LLM) agents that retrieve structured patient data through external tools. Published benchmarks, however, evaluate these systems at a single patient-population size, and rarely measure the effect of backend representation from that of the retrieval interface design. This paper compares six retrieval configurations that vary along two axes: backend (a property graph database, a relational database and a dense vector index) and interface design (curated domain-specific tool calls, model-generated queries, full-text search, and single-shot dense retrieval). The evaluation covers a 334-question bank spanning six categories (simple lookup, multi-hop, temporal, cohort, reasoning, and unanswerable), instantiated at three nested population scales: 200, 2000, and 20,000 alive patients from a single Synthea cohort. Four models are compared: Claude Haiku 4.5, Qwen 2.5 72B, Llama 3.1 8B, and Llama 3.3 70B, spanning closed-frontier and open-source alternatives. Curated tool-calling configurations improve accuracy over retrieval-augmented baselines for capable models, but reduce accuracy for a small open-source model due to function-calling protocol failures. We report how accuracy, latency, and cost evolve with each approach, model size, and cohort size, supported by paired statistical tests and confidence intervals. All benchmark components, databases, and evaluation code are publicly available.

Small-cell lung cancer (SCLC) is an aggressive malignancy that exhibits high prevalence for brain metastases. Furthermore, chemotherapy and metastasis-preventive approaches are also linked to neurotoxicity, further aggravating cognitive impairment. Despite evidence supporting structural and functional brain alterations in SCLC, the application of machine learning (ML) to new connectivity biomarkers has remained unexplored. This study is—to the best of our knowledge—the first to apply ML to structural brain connectomics in SCLC, using diffusion tensor imaging (DTI) to identify features discriminating between post-chemotherapy SCLC patients and healthy controls. Specifically, we constructed structural networks via deterministic tractography, applying an adapted feature reduction technique to identify the most informative connections without selection bias. This process isolated 16 connections involving 26 brain regions, predominantly in the frontal, temporal, and parietal lobes, showcasing primarily intra-hemispheric and left-lateralized alterations. Our optimal model leveraged a Gaussian Support Vector Machine (SVM), achieving a weighted accuracy of 0.92, a sensitivity of 0.93, a specificity of 0.91, and an area under the curve of 0.94. The selected feature subset retained high performance when tested with other classifiers, confirming its robustness. Our findings differ from prior studies based on statistically derived features, highlighting the ML-driven connectomics’ potential in uncovering DTI-derived SCLC patterns, offering interpretable insights for neuroimaging-based diagnostics.

Background: Healthcare robotics needs context-aware policy-compliant reasoning to achieve safe human–agent collaboration. The current ontologies fail to provide healthcare-relevant information and flexible semantic enforcement systems. Methods: HERON represents a modular upper ontology which enables healthcare robotic systems to communicate and collaborate while ensuring safety during operations. The system enables domain-specific instantiations through SPARQL queries and SHACL-based constraint validation to perform context-driven logic. The system models robotic task interactions through simulated eldercare and diagnostic and surgical support scenarios which follow ethical and regulatory standards. Results: The validation tests demonstrated HERON’s capacity to enable safe and explainable autonomous operations in changing environments. The semantic constraints enforced proper eligibility for roles and privacy conditions and policy override functionality during agent task execution. The HERON system demonstrated compatibility with healthcare IT systems and demonstrated adaptability to the GDPR and other policy frameworks. Conclusions: The semantically rich framework of HERON establishes an interoperable foundation for healthcare robotics. The system architecture maintains an open design which enables HL7/FHIR standard integration and robotic middleware compatibility. HERON demonstrates superior healthcare-specific capabilities through its evaluation against SUMO HL7 and MIMO. The future research will focus on optimizing HERON for low-resource clinical environments while extending its applications to remote care emergency triage and adaptive human–robot collaboration.

Background: Ulcerative colitis (UC) is a chronic inflammatory bowel disease characterized by continuous inflammation of the colon and rectum. Accurate disease assessment is essential for effective treatment, with endoscopic evaluation, particularly the Mayo Endoscopic Score (MES), serving as a key diagnostic tool. However, MES measurement can be subjective and inconsistent, leading to variability in treatment decisions. Deep learning approaches have shown promise in providing more objective and standardized assessments of UC severity. Methods: This study utilized publicly available endoscopic images of UC patients to analyze and compare the performance of state-of-the-art deep neural networks for automated MES classification. Several state-of-the-art architectures were tested to determine the most effective model for grading disease severity. The F1 score, accuracy, recall, and precision were calculated for all models, and statistical analysis was conducted to verify statistically significant differences between the networks. Results: VGG19 was found to be the best-performing network, achieving a QWK score of 0.876 and a macro-averaged F1 score of 0.7528 across all classes. However, the performance differences among the top-performing models were very small suggesting that selection should depend on specific deployment requirements. Conclusions: This study demonstrates that multiple state-of-the-art deep neural network architectures could automate UC severity classification. Simpler architectures were found to achieve competitive results with larger models, challenging the assumption that larger networks necessarily provide better clinical outcomes.

The rapid advancement of high-throughput technologies has led to the generation of vast amounts of omics data, including genomics, epigenomics, and metabolomics. Integrating these diverse datasets has become essential for gaining comprehensive insights into complex biological systems and enhancing personalized healthcare solutions. This critical review examines the current state of multi-omics data integration platforms, highlighting both the strengths and limitations of existing tools. By evaluating the latest digital platforms, such as GraphOmics, OmicsAnalyst, and others, the paper explores how they support seamless integration and analysis of omics data in healthcare applications. Special attention is given to their role in clinical decision-making, disease prediction, and personalized medicine, with a focus on their interoperability, scalability, and usability. The review also discusses the challenges these platforms face, such as data complexity, standardization issues, and the need for improved machine learning and AI-based analytics. Finally, the paper proposes directions for future research and development, emphasizing the importance of more advanced, user-friendly, and secure platforms that can better serve comprehensive healthcare needs.

Abdominal aortic aneurysm (AAA) is a complex vascular condition associated with high mortality rates. Accurate abdominal aorta segmentation is essential in medical imaging, facilitating diagnosis and treatment for a range of cardiovascular diseases. In this regard, deep learning-based automated segmentation has shown significant promise in the precise delineation of the aorta. However, comparisons across different models remain limited, with most studies performing algorithmic training and testing on the same dataset. Furthermore, due to the variability in AAA presentation, using healthy controls for deep learning AAA segmentation poses a significant challenge. This study provides a detailed comparative analysis of four deep learning architectures—UNet, SegResNet, UNet Transformers (UNETR), and Shifted-Windows UNet Transformers (SwinUNETR)—for full abdominal aorta segmentation. The models were evaluated both qualitatively and quantitatively using private and public 3D (Computed Tomography) CT datasets. Moreover, they were successful in attaining high performance in delineating AAA aorta, while being trained on healthy aortic imaging data. Our findings indicate that the UNet architecture achieved the highest segmentation accuracy among the models tested.

During the last few years, there has been a growing interest in students getting engaged in digital game-making activities so as to enhance their thinking skills. The findings of studies that have examined the impact of such initiatives are quite positive, especially concerning the promotion of 21st century skills; however, many students seem to face difficulties in getting a deeper understanding of the game development life cycle. Additionally, students often have difficulties in meaningfully reusing and applying the concepts from various subjects, mainly mathematics and physics, into their game-making tasks or in understanding advanced programming commands while creating their games. The present study presents an innovative game-making teaching approach that suggests a series of steps for the systematic design and rapid development of motion-based touchless games, i.e., games that are based on natural user interaction technologies, like the Microsoft Kinect camera. Findings from evaluation studies in two (2) secondary schools indicate that this approach can increase student motivation, strengthen their computational thinking, enhance their understanding of geometric principles and improve their social skills.