Maryam Arabi

Dialectal variation presents a major challenge for deploying medical language models in real-world healthcare settings, where patient–clinician communication often occurs in regional vernaculars rather than standardized language forms. This challenge is particularly pronounced in the Arabic-speaking world, where clinical interactions frequently take place in diverse dialects that differ substantially from Modern Standard Arabic. Fine-tuning and maintaining separate models for each dialect is computationally inefficient and difficult to scale, motivating more integrated approaches. In this work, we present MENARA, an Arabic medical language model constructed by merging Egyptian Arabic, Moroccan Darija, and medical-domain specialists through model merging. We extend prior feasibility findings through comprehensive evaluation of cross-dialect performance, medical safety, and cross-lingual knowledge retention. Specifically, we introduce a fine-grained dialect composition analysis to quantify lexical purity and structured code-switching behavior, benchmark against state-of-the-art Arabic LLMs, conduct subject-matter-expert assessment of both dialectal fidelity and medical appropriateness. The results show that model merging preserves core medical competence while enabling robust dialectal adaptation, achieving strong cross-dialect fidelity while substantially reducing storage and deployment overhead compared to maintaining separate models. These findings establish model merging as a potentially practical and resource-efficient paradigm for dialect-aware medical NLP in linguistically fragmented healthcare environments.

Background: Digital health technology (DHT) plays an increasingly vital role in managing chronic diseases by enabling patients to actively manage their health. These tools have been shown to improve self-management and adherence to medical advice. However, for DHT to be fully effective, patients with chronic conditions must be digitally literate. The eHealth Literacy Scale (eHEALS), an 8‑item tool with scores ranging from 8 to 40, was developed to assess individuals’ perceived ability to find, evaluate, and apply digital health information. Assessing patients’ digital health literacy (DHL) and understanding the factors influencing it are essential for improving the accessibility and usability of health resources. Objective: This study aimed to assess DHL in patients with diabetes mellitus (DM), hypertension, and rheumatoid arthritis (RA) through a systematic review and meta‑analysis using eHEALS. We sought to determine average DHL scores, examine demographic and socioeconomic factors influencing DHL, and explore its impact on disease management to inform future strategies for enhancing DHL and improving chronic disease outcomes. Methods: Following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, we conducted a systematic review across 7 databases (PubMed, SCOPUS, Embase, ERIC, CINAHL, Library Literature and Information Science Index, and Google Scholar) from inception to August 14, 2022, with an updated search in October 2024. Eligible studies included adults (≥18 years) with DM, hypertension, or RA who reported DHL data using eHEALS (8-40) and were original research published in English. Exclusion criteria included studies involving participants younger than 18 years, reviews, meta‑analyses, studies not addressing the target diseases, or non‑English publications. Study quality was evaluated using the Newcastle‑Ottawa Scale (NOS). Results: Eight studies involving 2527 participants were included. The pooled mean eHEALS score was 27.03 (95% CI 25.08-28.98), indicating high overall DHL. Stratified by disease, scores were higher for DM (27.79) and hypertension (28.48) but lower for RA (24.74). Quality assessment indicated a high standard of included studies. Factors influencing DHL included age, education, employment, and perception of the internet as a health resource. Due to the limited number of studies, meta‑regression analysis could not be performed. Conclusions: DHL is critical for individuals with chronic conditions, empowering them to make informed decisions and manage their health effectively. However, the scarcity of studies limits comprehensive analysis of DHL determinants. While the internet offers abundant health information, unequal DHL and health skills remain barriers. More inclusive research is needed to fully understand DHL’s impact on health outcomes and mitigate disparities, ensuring equitable access to digital health resources and improving disease management.

The increasing complexity and scale of electronic health records (EHRs) demand advanced tools for efficient data retrieval, summarization, and comparative analysis in clinical practice. MERA (Medical Electronic Records Assistant) is a Retrieval-Augmented Generation (RAG)-based AI system that addresses these needs by integrating domain-specific retrieval with large language models (LLMs) to deliver robust question answering, similarity search, and report summarization functionalities. MERA is designed to overcome key limitations of conventional LLMs in healthcare, such as hallucinations, outdated knowledge, and limited explainability. To ensure both privacy compliance and model robustness, we constructed a large synthetic dataset using state-of-the-art LLMs, including Mistral v0.3, Qwen 2.5, and Llama 3, and further validated MERA on de-identified real-world EHRs from the MIMIC-IV-Note dataset. Comprehensive evaluation demonstrates MERA’s high accuracy in medical question answering (correctness: 0.91; relevance: 0.98; groundedness: 0.89; retrieval relevance: 0.92), strong summarization performance (ROUGE-1 F1-score: 0.70; Jaccard similarity: 0.73), and effective similarity search (METEOR: 0.7–1.0 across diagnoses), with consistent results on real EHRs. The similarity search module empowers clinicians to efficiently identify and compare analogous patient cases, supporting differential diagnosis and personalized treatment planning. By generating concise, contextually relevant, and explainable insights, MERA reduces clinician workload and enhances decision-making. To our knowledge, this is the first system to integrate clinical question answering, summarization, and similarity search within a unified RAG-based framework.

Large language models (LLMs) are increasingly being applied to clinical tasks, but it remains unclear whether medical-specific models consistently outperform smaller, generalpurpose ones. This study investigates that assumption in the context of cardiovascular disease (CVD) risk assessment. We fine-tuned eight LLMs—both general-purpose and medical-specific—using textualized data from the Behavioral Risk Factor Surveillance System (BRFSS) to classify individuals as “High Risk” or “Low Risk”. To provide actionable insights, we integrated a Retrieval-Augmented Generation (RAG) framework for personalized recommendation generation and deployed the system within an interactive chatbot interface. Notably, Gemma2, a compact 2B-parameter general-purpose model, achieved a high recall (0.907) and F1-score (0.770), performing on par with larger or medical-specialized models such as Med42 and BioBERT. These findings challenge the common assumption that larger or specialized models always yield superior results, and highlight the potential of lightweight, efficiently fine-tuned LLMs for clinical decision support—especially in resource-constrained settings. Overall, our results demonstrate that general-purpose models, when fine-tuned appropriately, can offer interpretable, high-performing, and accessible solutions for CVD risk assessment and personalized healthcare delivery.

Background: The SARS-CoV-2 virus elicited a major public concern worldwide since December 2019 due to the high number of infections and deaths caused by COVID-19. The Omicron variant was detected in October 2021 which evolved from the wild-type SARS-CoV-2 and was found to possess many mutations. Omicron exhibited high transmissibility and immune evasion as well as reduced severity when compared to the earlier variants. Although vaccinated individuals were largely protected against infections in previous waves, the high prevalence of both reinfections and breakthrough infections with Omicron was observed. The aim of this review is to understand the effectiveness of previous infection on subsequent reinfection, given its significance in driving public health policy, including vaccination prioritization and lockdown requirements. Methods: A comprehensive literature search was conducted using several databases to target studies reporting data related to the effectiveness of the previous infection with SARS-CoV-2 in protecting against the Omicron variant. Screening of the studies, quality assessment and data extraction were conducted by two reviewers for each study. Results: Only 27 studies met our inclusion criteria. It was observed that previous infection was less effective in preventing reinfections with the Omicron variant compared to the Delta variant irrespective of vaccination status. Furthermore, being fully vaccinated with a booster dose provided additional protection from the Omicron variant. Additionally, most infections caused by Omicron were asymptomatic or mild and rarely resulted in hospitalizations or death in comparison to the Delta wave. Conclusion: A majority of the studies reached a consensus that although previous infection provides some degree of immunity against Omicron reinfection, it is much lower in comparison to Delta. Full vaccination with two doses was more protective against Delta than Omicron. Receiving a booster dose provided additional protection against Omicron. It is therefore clear that neither vaccination nor previous infection alone provide optimal protection; hybrid immunity has shown the best results in terms of protecting against either Omicron or Delta variants. However, additional research is needed to quantify how long immunity from vaccination versus previous infection lasts and whether individuals will benefit from variant-specific vaccinations to enhance protection from infection.

The Omicron variant was first detected in October 2021, which evolved from the original SARS‐CoV‐2 strain and was found to possess many mutations. Immune evasion was one of the notable consequences of these mutations. Despite Omicron exhibiting increased transmissibility, the rates of hospitalizations and deaths among patients infected with this variant were substantially lower when compared to other strains. However, concluding that the Omicron variant is less severe than other variants of SARS‐CoV‐2 requires consideration of multiple factors, including the vaccination status of infected patients as well as any previous infections with other variants. This review compiled data about any reported indicators of severity in Omicron‐infected patients, including studies comparing Omicron with other variants while adjusting for confounders. A comprehensive search was conducted using different databases to target any studies about Omicron. In total, 62 studies met our inclusion criteria and were included in this study. Many studies reported a significantly reduced risk of hospitalization, ICU admission, need for oxygenation/ventilation, and death in Omicron‐infected patients compared to patients infected with other variants, such as Delta. Some studies, however, reported comparable severity in Omicron infected patients as to other variants emphasizing a substantial risk for severe illness. Furthermore, the COVID‐19 vaccines were less effective against Omicron relative to previous lineages, except after receiving the booster dose. One study recommended vaccination during pregnancy, which may help prevent future cases of severe SARS‐CoV‐2 pneumonia in neonates and young infants due to the transfer of humoral response from the mother.

Medical Humanities is an interdisciplinary field that has slowly been integrated into medical education. Despite developments in medical technologies, patients still experience feelings of grief, anger, and anxiety about their health and rely on their doctor for moral and emotional support. Thus, patients are inevitably finding doctors' empathy and understanding equally as important as their medical knowledge and clinical experience. The development and incorporation of medical humanities courses are vital for enhancing the delivery of treatment to the patient and optimizing their physical, mental, social, and emotional well-being. Nonetheless, some healthcare professionals and educators still argue that the medical field has advanced far beyond the theoretical approaches of medical humanities, especially since limited data is available to support their effectiveness. Another objection that is raised against the humanities is that the medical curriculum is already extremely rigorous, which makes it challenging to incorporate any new material into the standard coursework. Seven medical students from WCM-Q aim to uncover the potential values and limitations of medical humanities in current education and its efficacy in clinical settings. The discussion sheds light on the core skills that can be promoted by humanities including observation, listening, sensitivity, sensibility, and communication. Furthermore, the panelists engaged in a fruitful conversation exploring the narrative and scientific thinking of doctors and examining the cost-benefit analysis of modern medical practices with traditional therapeutic approaches.