Fresenius (Germany)

Handheld ultrasound (HHUS) devices are increasingly used in clinical practice due to their portability and cost-effectiveness. However, HHUS image quality remains variable, and a standardized expert-based evaluation protocol is lacking. This study compares the image quality of B-mode recordings of five HHUS models with a high-end reference system (HEUS) using a blinded assessment to identify differences, covariates, offering a standardized evaluation framework.In this blinded, multi-center study, video sequences of the liver, pancreas, and sigmoid colon were recorded in ten volunteers using 5 HHUS devices - Vscan Air, Butterfly iQ+, Philips Lumify, Clarius C3, and Clarius C3 HD3 - and one HEUS (GE Logiq E10). Twenty-two physicians with many years of abdominal ultrasound experience rated blinded clips for 5 image quality parameters.The HEUS consistently received the highest ratings. Image quality differed significantly between HHUS devices. Clarius systems showed the smallest deviation from the reference, followed by Vscan Air and Philips Lumify. All image parameters significantly correlated with the overall impression, without a single dominant factor. The patients' BMI had no relevant influence. Examiner experience was less impactful than device familiarity and personal preference.The image quality of HEUS remained superior to all tested HHUS devices. Further subdivision of image quality aspects did not provide additional information. Furthermore, we could not identify significant covariates in our setup. These findings could justify simplified setups with reduced questionnaires and requirements for examiners.

The integration of generative AI (GenAI) and large language models (LLMs) in healthcare presents both unprecedented opportunities and challenges, necessitating innovative regulatory approaches. In this perspective, we discuss the risks of GenAI and LLM-based medical devices, the limitations of current medical device regulation frameworks when applied to GenAI or LLMs, and advocate for global collaboration in regulatory science research through engaging multidisciplinary expertise and focusing on the needs of diverse populations.

The convergence of (multi-)omics and artificial intelligence is transforming precision medicine by enabling data-driven diagnostics, treatment prediction, and patient stratification. Yet, regulatory frameworks lag behind this innovation. This commentary discusses emerging issues around data integrity, algorithm transparency, validation and real-world evidence integration, highlighting gaps between device and pharmaceutical regulations. Drawing on examples of regulatory-approved AI-enabled (multi-)omics tools in the EU and US, we explore evolving regulatory pathways shaping the clinical adoption of next-generation tools for precision medicine.

Artificial intelligence (AI) offers the promise of automating key tasks in haematology and unveiling novel biomarkers. However, model outputs may be biased, lack interpretability or fail to generalise. AI literacy is essential for haematologists to critically evaluate model outputs, effectively using them as decision support tools.

AimsAge‐associated clonal haematopoiesis of indeterminate potential (CHIP) has been linked to increased incidence and worse prognosis of chronic heart failure (CHF). CHIP arises from somatic mutations in haematopoietic stem and progenitor cells. Mosaic loss of Y chromosome (LOY), the most common somatic mutation in male blood cells, increases with age, drives clonal expansion of myeloid cells, and has been experimentally associated with cardiac fibrosis and heart failure in mice. However, its prognostic value and interplay with CHIP in CHF patients remain unclear.Methods and resultsWe analysed 781 male CHF patients across the full spectrum of left ventricular ejection fraction to assess the prevalence and prognostic relevance of LOY and the two most common CHIP‐driver mutations, DNMT3A and TET2. Both LOY and CHIP mutations increased with age and co‐occurred in 27.1% of men >70 years. LOY independently predicted all‐cause mortality in patients with heart failure with reduced ejection fraction (HFrEF). The co‐occurrence of LOY and DNMT3A/TET2 mutations further increased mortality among CHIP carriers. This detrimental prognostic effect of LOY was confirmed in a validation cohort of HFrEF patients. Single‐cell RNA sequencing of peripheral blood mononuclear cells from HFrEF patients with ischaemic heart failure revealed elevated pro‐fibrotic signalling in LOY monocytes, characterized by increased inflammatory and remodelling markers (S100A8, TLR2, CLEC4D) and decreased expression of transforming growth factor‐β inhibitors (SMAD7, TGIF2). In patients with both LOY and DNMT3A mutations, monocytes showed enhanced pro‐inflammatory gene expression, including alarmins (S100A8, HMGB2) and interferon‐related genes (IFNGR1, TRIM56, CD84).ConclusionsSomatic mutations in blood cells—particularly LOY—are associated with increased mortality in male CHF patients, with LOY emerging as an independent prognostic marker.

Mitochondrial membrane potential is highly dependent on coupled as well as uncoupled respiration. While brown adipose tissue (BAT) mediates non-shivering thermogenesis (NST), a highly adaptive bioenergetic process critical for energy metabolism, the relationship of coupled and uncoupled respiration in thermogenic adipocytes remains complicated. Uncoupling protein 1 (UCP1)-mediated proton leak is the primary driver of NST, but recent studies have shown that oxidative phosphorylation may be an underappreciated contributor to UCP1-dependent NST. Here, we highlight the role of ATP synthase for BAT thermogenesis and discuss the implications of fine-tuning adrenergic signaling in brown adipocytes by the protein inhibitory factor 1 (IF1). We conclude by hypothesizing future directions for mitochondrial research, such as investigating the potential role of IF1 for mitochondrial substrate preference, structural dynamics, as well as its role in cell fate decision and differentiation.

Background and study aimsEndoscopy video recordings are valuable data for training and deploying artificial intelligence (AI) models. However, collecting these data is challenging and time-consuming, demanding new workflows and robust data management strategies.MethodsHere, we outline the challenges associated with routinely recording endoscopy data in clinical practice and share experiences and solutions from three endoscopy centers in Germany and the United States.ResultsEach center uses a recording setup tailored to specific needs of that endoscopy unit. Common challenges include integrating with the hospital’s electronic health records, automating video recording, and addressing data privacy concerns. In all cases, having dedicated research staff to manage daily operations has proven essential for successful implementation.ConclusionsBy describing successful strategies, we aim to inspire gastroenterology divisions worldwide to adapt routine video recording for endoscopy procedures, thereby increasing the volume and diversity of datasets necessary for developing clinically impactful AI applications.

ABSTRACT Public value theory argues that public value emerges from collectively shared individual need fulfilment. The perception of public value is expected to yield effects comparable to those resulting from the satisfaction of individual needs – for instance, with respect to the experience of flow. In a design with two measurement waves (N1&2 = 797) in one of the largest German public administrations, we found evidence for our model that the perception of public value is positively related to future purposeful goal setting, and purpose is positively related to professional self-efficacy beliefs. Thus, public value-related need fulfilment fosters the long-term development of flow, sequentially mediated by purpose- and self-efficacy-related implicit beliefs.

Recent advancements in voice-enabled artificial intelligence (AI), particularly end-to-end speech-to-speech models, are reshaping communication within health care. These models, such as OpenAI's Advanced Voice Mode (AVM), offer real-time, nuanced human-like interactions by capturing intonation and pitch, thereby enabling more natural machine-human dialogue. In this paper we explore the integration of voice-enabled AI into medical practice, highlighting the potential to improve clinical efficiency, medical education, and patient engagement by providing self-recorded use cases. While the benefits are promising-ranging from increased accessibility to reduced clinician workload-challenges remain in data security, reliability, integration with existing systems, and ethical use. Addressing these concerns through robust regulation, transparent development, and targeted training will be essential. Ultimately, voice-enabled AI holds transformative potential to bridge communication gaps in medicine and support more equitable, efficient, and patient-centered care.