Background: Frailty is a condition characterized by an age-related decrease of physiological reserve, which increases vulnerability to stressors, leading to higher risk of negative health events. Evidence has demonstrated that frailty can be reversed through a multidisciplinary approach that encompasses multicomponent exercise, nutritional advice and medication management. However, the role of Information and Communication Technologies (ICTs) in the implementation of such programs has been scarcely explored in the scientific literature. We describe the rationale and methods of a multimodal intervention in frailty using innovative ICTs, the MIRATAR study.

Methods/design: The MIRATAR study is a quasi-experimental, longitudinal, multicentre, clustered, superiority, and prospective study. A sample of 50 community-dwelling older adults (aged > 65 years), identified as pre-frail or frail according to the Fried Frailty Phenotype (FFP), will be recruited in two Spanish sites (Getafe and Albacete). Our primary objective is to evaluate if a personalized multi-modal intervention supported by the MIRATAR technological ecosystem can improve frailty status after 6 months, either based on a reduction in more than 2.5 points in the 5-item Frailty Trait Scale (FTS-5) and/or a reduction in more than 1 points in the FFP score. Secondary outcomes include participant’s autonomy, nutritional status, quality of life and cost-effectiveness. The core of the MIRATAR ecosystem is an AI-powered virtual assistant/carer that will interact with participants via an intelligent mirror, enabling natural interaction. It will provide guidance on physical exercise and nutrition, tailored to each participant's functional status and characteristics. Additionally, MIRATAR will use multiple sensing technologies to gather supplementary information like physiological data (e.g., weight, blood pressure), environmental data (e.g., temperature, presence), and activity tracking through wearable sensors.

Conclusions: The MIRATAR study will provide evidence on the usability, acceptance, and clinical and functional effectiveness of a personalized multi-modal intervention for frail and pre-frail older people provided through innovative ICTs.

Introduction: Obesity involves complex inter-organ communication, potentially mediated by gut microbiota-derived exosomes carrying molecular cargo. This study investigated whether these exosomes mediate signaling between the gut and metabolic tissues, contributing to obesity in a diet-induced model.

Methods: We isolated exosomes from the feces of diet-induced obese mice and lean controls. Their molecular cargo (miRNAs, proteins, lipids) was characterized using sequencing and proteomics. In vivo, exosomes from obese mice were administered to germ-free mice to assess their impact on weight gain, adiposity, and metabolic markers. In vitro studies exposed adipocytes and hepatocytes to these exosomes to investigate changes in gene expression and signaling pathways related to lipid metabolism and inflammation.

Results: Analysis identified distinct molecular cargo profiles in exosomes isolated from obese versus lean mice. The administration of obese-derived exosomes to germ-free recipients impacted metabolic parameters. In vitro experiments revealed that these exosomes influenced responses in target metabolic cells.

Conclusion: This research provides evidence suggesting that gut microbiota-derived exosomes function as molecular messengers affecting host metabolism in obesity. The findings support their potential role in inter-organ communication and highlight areas for further investigation into specific molecular targets and therapeutic strategies.

Background:

Pulmonary embolism (PE) is a life-threatening diagnosis that often prompts urgent imaging with CT Pulmonary Angiography (CTPA). While CTPA is highly sensitive, its liberal use in low-risk patients raises concerns around unnecessary radiation, contrast exposure, and resource strain. This study investigates whether validated clinical algorithms can outperform clinicians' judgment in guiding CTPA use.

Methods:

We retrospectively reviewed all patients who underwent CTPA for suspected PE at a major tertiary center in Bahrain between December 2019 and August 2024. Pre-test probability was retrospectively calculated using the Wells score, and D-dimer testing was analyzed for compliance with NICE-recommended diagnostic pathways.

Results:

Among the 743 patients evaluated, only 18.7% had CTPA-confirmed PE; this was significantly lower than the 15.4–37% diagnostic yield benchmark recommended by the Royal College of Radiologists. Notably, 47.3% of patients with a low Wells score and negative D-dimer underwent CTPA despite being eligible to avoid imaging per the existing guidelines. Additionally, the Wells score was undocumented in nearly two-thirds of cases, indicating that clinical intuition frequently overrode structured decision-making tools.

Conclusion:

Our data reveal a substantial gap between guideline-based diagnostic pathways and real-world clinical practice. The overuse of CTPA not only inflates healthcare costs but also exposes patients to avoidable harm. This study supports the integration of algorithm-based tools into electronic ordering systems to standardize decision-making, reduce low-yield scans, and enhance diagnostic accuracy. As imaging technology advances, so too must our frameworks for using it wisely—not as a reflex, but as a deliberate, evidence-informed step in patient care.

Background: Multiple sclerosis (MS) is a chronic autoimmune disease of the central nervous system that presents significant diagnostic and therapeutic challenges. Recent advancements in artificial intelligence (AI) have shown promise in improving early detection and personalizing the treatment strategies for MS.
Methodology: A systematic review was conducted, analyzing studies published between 2022 and 2025 that employed AI techniques in MS diagnosis and treatment. The review focused on studies using magnetic resonance imaging (MRI), optical coherence tomography (OCT), and machine learning algorithms to assess diagnostic accuracy and treatment efficacy.
Results: The AI models demonstrated high diagnostic accuracy, with the combined sensitivity and specificity ranging from 92% to 93% across various studies. For instance, a study by Hernandez et al. (2023) used explainable AI to analyze OCT data, achieving significant accuracy in early MS detection. Furthermore, AI algorithms have facilitated the identification of new MS subtypes, such as ‘cortex-led’, ‘normal-appearing white-matter-led’, and ‘lesion-led’, which correlate with different disease progression patterns and treatment responses .
Conclusions: AI technologies are revolutionizing MS management by allowing for early and accurate diagnoses, uncovering disease subtypes, and personalizing treatment plans. The integration of AI into clinical practice holds the potential to improve patient outcomes through personalized therapeutic approaches. However, challenges remain, including the need for large, diverse datasets and a reduction in biases in AI models. Future research should focus on verifying AI tools in diverse populations and creating standardized protocols for their clinical application.