Communications medicine最新文献

{"title":"Benchmarking large language models for HIV medical decision support.","authors":"Gonzalo Cardenal-Antolin, Jacques Fellay, Bashkim Jaha, Huldrych F Günthard, Johannes Nemeth, Roger Kouyos, Niko Beerenwinkel, Diane Duroux","doi":"10.1038/s43856-026-01875-1","DOIUrl":"https://doi.org/10.1038/s43856-026-01875-1","url":null,"abstract":"<p><strong>Background: </strong>Large language models (LLMs) are emerging as tools to support clinical decision making. HIV management is a compelling use case due to its complexity and dynamic nature, involving diverse treatment options, comorbidities, and adherence challenges. However, integrating LLMs into clinical practice raises concerns about accuracy, safety, and clinician acceptance. Despite growing interest, their performance in HIV care remains poorly studied, and benchmarking is lacking.</p><p><strong>Methods: </strong>We developed HIVMedQA, a clinician-curated benchmark of HIV-related open-ended medical question-answer pairs spanning basic knowledge, clinical reasoning, complex patient vignettes, and bias-modified scenarios. We evaluated seven general-purpose and three medical LLMs. Performance was assessed using lexical similarity and an extended medical LLM-as-a-judge framework capturing key clinical dimensions, including question comprehension, reasoning, knowledge recall, bias, potential harm, and factual accuracy, to better capture nuances relevant to the medical domain, with additional evaluation by HIV-experienced physicians.</p><p><strong>Results: </strong>Performance varies substantially across models and task complexity. Gemini 2.5 Pro achieves the highest overall scores, followed by Claude 3.5 Sonnet and MedGemma-27B. Knowledge recall is generally stronger than question comprehension or clinical reasoning. Medical LLMs do not consistently outperform general-purpose models, and model size alone does not predict performance. Several models are sensitive to cognitive bias prompts. LLM-as-a-judge scoring aligns better with clinician assessment than lexical metrics.</p><p><strong>Conclusions: </strong>HIVMedQA provides a structured benchmark for evaluating LLMs in HIV clinical decision support. Current LLMs show promise, but limitations in reasoning, bias robustness, and safety indicate that careful validation, domain-specific evaluation, and clinician oversight remain essential before clinical deployment.</p>","PeriodicalId":72646,"journal":{"name":"Communications medicine","volume":"6 1","pages":""},"PeriodicalIF":7.4,"publicationDate":"2026-08-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149247655","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Pooled two-cohort MRI body composition phenotyping with open-source deep learning.","authors":"Christian J Mertens, Hartmut Häntze, Sebastian Ziegelmayer, Jakob Nikolas Kather, Daniel Truhn, Su Hwan Kim, Felix Busch, Dominik Weller, Benedikt Wiestler, Markus Graf, Fabian Bamberg, Christopher L Schlett, Jakob B Weiss, Steffen Ringhof, Elif Can, Jeanette Schulz-Menger, Thoralf Niendorf, Jacqueline Lammert, Isabel Molwitz, Avan Kader, Alessa Hering, Aymen Meddeb, Jawed Nawabi, Matthias B Schulze, Thomas Keil, Stefan N Willich, Lilian Krist, Martin Hadamitzky, Anke Hannemann, Florian Bassermann, Daniel Rueckert, Tobias Pischon, Alexander Hapfelmeier, Marcus R Makowski, Keno K Bressem, Lisa C Adams","doi":"10.1038/s43856-026-01888-w","DOIUrl":"10.1038/s43856-026-01888-w","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;Body mass index fails to capture variation in fat and muscle distribution that determines metabolic health and disease risk. MRI enables radiation-free quantification of regional body composition, yet scalable open-source tools applied in pooled cohorts with differing acquisition protocols have been lacking.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;MRSegmentator, an open-source nnU-Net-based pipeline, was applied to quantify visceral adipose tissue (VAT), abdominal subcutaneous adipose tissue (ASAT), gluteofemoral adipose tissue (GFAT), trunk musculature, and the liver mask used for liver fat-fraction estimation in 45,851 adults from the German National Cohort (n = 26,877, 3 T multi-centre Siemens) and UK Biobank (n = 18,974, 1.5 T Siemens). Population-scale compartment volumes were segmented from stitched in-phase gradient-echo (GRE) images in both cohorts; liver fat fraction was calculated from fat-only and water-only images. The annotated development data comprised NAKO T2-HASTE and UKB Dixon reconstructions. A single pooled model was applied without site-specific adaptation. A separate two-reader agreement study used 50 scans from these annotated development-sequence domains. Associations between BMI-adjusted body composition and cardiometabolic conditions were estimated using generalized linear mixed-effects models. Incremental discrimination beyond age, BMI, and waist-to-hip ratio was assessed.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;Five-fold participant-stratified internal cross-validation against curated human-in-the-loop development references comprising UKB Dixon and NAKO T2-HASTE yielded a mean Dice of 0.91. In a separate 50-scan reader study on these annotated development-sequence images, overall reader-reader Dice was 0.937 and overall algorithm-reader Dice was 0.908. The trained pipeline was then used to segment compartment volumes from stitched in-phase GRE inputs in both cohorts, while liver fat fraction was calculated from fat-only and water-only images; direct sequence-matched validation on NAKO GRE was not performed. VAT showed the strongest positive associations with cardiometabolic conditions, while GFAT showed inverse associations, most prominently for type 2 diabetes (OR 0.69, 95% CI 0.66 to 0.72). Disease-specific body-composition phenotypes were identified, with type 2 diabetes characterized by elevated VAT, reduced GFAT, and increased liver fat. MRI-derived compartments modestly improved discrimination for type 2 diabetes and hyperlipidemia beyond anthropometric measures.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusions: &lt;/strong&gt;A single open-source deep-learning pipeline enabled pooled body-composition phenotyping in two cohorts and captured distributional variation in fat and muscle beyond BMI. High agreement in internal cross-validation (mean Dice 0.91) and the separate two-reader study support the annotated development-sequence analysis, while the population-scale application identified distinct disease-associated phenotyp","PeriodicalId":72646,"journal":{"name":"Communications medicine","volume":"6 1","pages":""},"PeriodicalIF":7.4,"publicationDate":"2026-08-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13526011/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148857972","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Venular-centered thrombo-inflammation drives microvascular failure after arterial recanalization in acute mesenteric ischemia.","authors":"Déborah François, Lambert Kernanet, Melinda Bajul, Abigail Brami, Véronique Arocas, Marie-Christine Bouton, Dominique Cazals-Hatem, Kevin Guedj, Benoit Ho-Tin-Noé, Olivier Corcos, Yacine Boulaftali, Alexandre Nuzzo","doi":"10.1038/s43856-026-01878-y","DOIUrl":"10.1038/s43856-026-01878-y","url":null,"abstract":"<p><strong>Background: </strong>Acute mesenteric ischemia (AMI) remains associated with high mortality despite prompt revascularization, suggesting that downstream ischemia-reperfusion injury contributes to poor outcomes. However, the microvascular mechanisms underlying this process are not entirely defined.</p><p><strong>Methods: </strong>We analyzed admission blood samples from patients with arterial AMI and non-ischemic controls enrolled in the SURVIBIO diagnostic study (NCT03518099). We also investigated thrombo-inflammatory responses in a murine superior mesenteric artery occlusion (SMAO) model. In mice, intravital microscopy was used to directly visualize mesenteric microcirculatory flow and thrombo-inflammatory events during ischemia-reperfusion.</p><p><strong>Results: </strong>Patients with AMI display a marked systemic thrombo-inflammatory profile, characterized by elevated inflammatory markers, neutrophil activation, platelet activation, and alterations in coagulation-related proteins, which are closely mirrored in the SMAO model. Intravital microscopy reveals a dissociation between arterial and microvascular reperfusion: while arteriolar flow partially recovers after recanalization, venular perfusion remains severely impaired and is associated with early blood cell stasis and stable venular thrombus formation. Thrombi develop through a sequential process initiated during ischemia and amplified during reperfusion.</p><p><strong>Conclusion: </strong>Together, these findings suggest that venular-centered thrombo-inflammation is closely associated with microvascular dysfunction and intestinal injury in arterial AMI. This work provides a framework for targeting thrombo-inflammatory pathways beyond arterial reperfusion that may extend to other clinical forms of AMI, including non-occlusive mesenteric ischemia (NOMI).</p>","PeriodicalId":72646,"journal":{"name":"Communications medicine","volume":"6 1","pages":""},"PeriodicalIF":7.4,"publicationDate":"2026-08-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13530138/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148868167","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Evidence, use cases, and implementation safeguards of large language models in primary care.","authors":"Michael Christof, Krish Patel, Jiandong Zhou, Antonis A Armoundas","doi":"10.1038/s43856-026-01823-z","DOIUrl":"10.1038/s43856-026-01823-z","url":null,"abstract":"<p><p>Recent developments in large language models (LLMs) have created new opportunities to support primary care, where much of clinical work is text-mediated. This narrative review synthesizes evidence on LLM applications relevant to primary care workflows and summarizes implementation safeguards. Across studies, the most consistently supported near-term value is workflow augmentation, particularly documentation and inbox management (e.g., drafting portal replies and summarizing information for clinician review) and communication support, where benefits are reported primarily as process endpoints (time, acceptability, perceived communication quality) rather than hard patient outcomes. Evidence for improvements in clinician diagnostic reasoning, treatment planning, and downstream patient outcomes is more limited and context-dependent, and many evaluations remain simulated or conducted in adjacent settings, limiting generalizability to routine primary care. Accordingly, potential roles in population health and cost reduction should be treated as hypothesis-generating and evaluated prospectively. Challenges related to privacy, security, transparency, and model reliability shape organizational governance requirements and evolving regulatory expectations for the clinical use of generative AI in primary care. We emphasize a pragmatic adoption approach: prioritize high-volume, lower-risk clerical and communication workflows; maintain clinician verification and accountability; and apply governance and equity safeguards (e.g. privacy, security, transparency, auditability, monitoring for drift and error) before scale-up.</p>","PeriodicalId":72646,"journal":{"name":"Communications medicine","volume":"6 1","pages":""},"PeriodicalIF":7.4,"publicationDate":"2026-08-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13507062/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148820395","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"From frailty-driven to frailty-informed care in the age of wearable AI.","authors":"Elena Giovanna Bignami, Mattia Madeo, Carmine Siniscalchi, Sara Fedele, Nicoletta Cerundolo, Valentina Bellini, Tiziana Meschi","doi":"10.1038/s43856-026-01868-0","DOIUrl":"10.1038/s43856-026-01868-0","url":null,"abstract":"<p><p>As populations age, frailty is increasingly monitored through wearable technologies capable of continuously capturing behavioral and physiological change. While these systems enable earlier detection of vulnerability, they also risk transforming probabilistic signals into rigid clinical thresholds. We propose a frailty-informed framework in which artificial intelligence (AI) augments, rather than replaces, clinical judgment. By integrating longitudinal digital monitoring with multidimensional geriatric assessment and bedside evaluation, this approach supports anticipatory, patient-centered care. Proportional design, transparency and clinical oversight are essential to ensure that predictive systems guide care without narrowing therapeutic options.</p>","PeriodicalId":72646,"journal":{"name":"Communications medicine","volume":"6 1","pages":""},"PeriodicalIF":7.4,"publicationDate":"2026-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13498552/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148802333","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Using the Stressor-System-Resilience Resources-Outcomes (SSRO) scaffold to clarify the core components of resilience research in medicine.","authors":"Deepak K Ravi, Kaisa Koivunen, Kirsten Dj Bos, René Jf Melis","doi":"10.1038/s43856-026-01864-4","DOIUrl":"10.1038/s43856-026-01864-4","url":null,"abstract":"<p><p>Modern medicine increasingly extends beyond treating disease to supporting individuals through recovery, adaptation, and long-term functioning, all central to the concept of resilience. Despite growing interest, resilience remains conceptually difficult to structure and compare across disciplines, limiting its value for guiding research and clinical care. In this Perspective, we identify recurring features across many resilience traditions and propose a practical organizing scaffold for studying resilience in medicine. The scaffold specifies four elements that any resilience study should make explicit: the stressor, the system, resilience-relevant resources and processes, and outcomes. Using worked examples, we show how the scaffold clarifies resilience questions and improves comparability across studies. We also provide a step-by-step roadmap to help researchers new to the field design resilience studies, from framing a question through to interpreting results.</p>","PeriodicalId":72646,"journal":{"name":"Communications medicine","volume":"6 1","pages":""},"PeriodicalIF":7.4,"publicationDate":"2026-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13477472/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148765575","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Blood metal signatures and their association with Alzheimer's disease risk and cognitive decline in an Egyptian cohort.","authors":"Mai Othman, Eman Khedr, Gharib Fawi, Shimaa Heikal, Sara A Mostafa, Amna Ibrahim, Ahmed Qayaty, Eman M Ali, Mohamed Salama","doi":"10.1038/s43856-026-01844-8","DOIUrl":"https://doi.org/10.1038/s43856-026-01844-8","url":null,"abstract":"<p><strong>Background: </strong>Alzheimer's disease (AD) is a multifactorial disease in which environmental exposures may contribute to disease risk. This study aimed to investigate associations between blood concentrations of arsenic, lead, mercury, and lithium with AD risk and cognitive performance in an Egyptian cohort.</p><p><strong>Methods: </strong>We conducted a cross-sectional study in 50 AD patients and 50 cognitively healthy controls recruited across multiple centres in Egypt. Blood metal concentrations were measured by inductively coupled plasma optical emission spectrometry. A tiered analytical strategy addressed variable left-censoring: group-stratified Kaplan-Meier imputation was applied for arsenic, lead, and lithium, followed by logistic and linear regression adjusted for age and sex, with a binary detection sensitivity analysis for lithium, lead, and mercury. Multiple testing was corrected using the Benjamini-Hochberg false discovery rate method.</p><p><strong>Results: </strong>Here we show that higher arsenic concentrations are strongly associated with increased odds of AD (OR = 1321.61, p = 3.48×10⁻⁶) and lower cognitive scores (β = -16.44, p = 1.46×10⁻⁸), both surviving false discovery rate correction. Higher lithium is associated with reduced disease odds (OR = 0.634, FDR-adjusted p = 0.022) and better cognitive performance (β = 1.21, FDR-adjusted p = 0.020), corroborated in sensitivity analyses. Mercury detection is markedly more frequent in patients than controls (24% vs. 2%, p = 0.002) and associates with lower cognitive scores (β = -7.08, p = 0.0003). Lead shows a borderline association with disease (OR = 1.53, p = 0.046) that does not survive correction.</p><p><strong>Conclusions: </strong>These findings support arsenic neurotoxicity and lithium deficiency as potentially important contributors to AD risk in an Egyptian population, with implications for environmental public health in low- and middle-income settings.</p>","PeriodicalId":72646,"journal":{"name":"Communications medicine","volume":"6 1","pages":""},"PeriodicalIF":7.4,"publicationDate":"2026-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13627698/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149119345","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"High-frequency visual stimulation can increase medial temporal lobe ripple oscillation density.","authors":"Julian Keil, Victor Hernandez-Urbina, Liam Doherty, Hamed Bahmani, Martin Holtkamp, David Steinbart, Alexandra-Dikorey Diall, Jonas M Hebel, Christian Brandt, Thomas Cloppenborg, Anna Doll, Markus Müschenich","doi":"10.1038/s43856-026-01846-6","DOIUrl":"10.1038/s43856-026-01846-6","url":null,"abstract":"<p><strong>Background: </strong>Flickering visual stimulation can evoke neural oscillations, which can influence ongoing brain activity. Electrophysiological recordings of neural oscillations in the ripple band (80-180 Hz) show that these high-frequency oscillations occur in the neocortex and the hippocampus, that they phase-synchronize across long distances, and that ripple oscillations in the neocortex often precede those in the hippocampus during wakefulness. It is therefore possible that the neocortical ripple oscillations propagate beyond sensory areas to the hippocampus, inducing ripple oscillations.</p><p><strong>Methods: </strong>To test the hypothesis that neocortical ripple oscillations induced by visual stimulation induce hippocampal ripple oscillations, we conduct an exploratory experiment (N = 8) in humans, using ultra-high frequency visual stimulation to induce ripple oscillations recorded through electrodes implanted in or near the hippocampus. Although hippocampal ripple oscillations, so-called sharp-wave-ripples, mostly occur during quiet rest or slow-wave sleep, we aim to increase their abundance using visual stimulation during wakefulness in this exploratory study. We hypothesize that ultra-high frequency visual stimulation increases the number of sharp-wave-ripples relative to an eyes-open resting-state baseline.</p><p><strong>Results: </strong>In this exploratory and preliminary study, we observe significantly more sharp-wave-ripples per second during periods of stimulation compared to a resting-state baseline before and after the stimulation.</p><p><strong>Conclusions: </strong>The increased number of sharp-wave ripples during stimulation suggests that ultra-high-frequency visual stimulation can be used as a safe noninvasive tool to influence sharp-wave ripples, which offers the potential to improve memory.</p>","PeriodicalId":72646,"journal":{"name":"Communications medicine","volume":"6 1","pages":""},"PeriodicalIF":7.4,"publicationDate":"2026-08-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13458760/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148708628","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Author Correction: Comprehensive cross-cohort analysis reveals global gut microbiome signatures of celiac disease.","authors":"Peter J Prendergast, Haig V Bishop, Craig W Herbold, Elena F Verdu, Renwick C J Dobson, Andrew S Day, Olivia J Ogilvie","doi":"10.1038/s43856-026-01832-y","DOIUrl":"10.1038/s43856-026-01832-y","url":null,"abstract":"","PeriodicalId":72646,"journal":{"name":"Communications medicine","volume":"6 1","pages":""},"PeriodicalIF":7.4,"publicationDate":"2026-08-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13452810/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148698436","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Brain responses to food cue viewing in pregnant women with and without gestational diabetes mellitus.","authors":"Mariana Treviño Montemayor, Anna Lesniara-Stachon, Dan Yedu Quansah, Sybille Schenk, Martin Heni, Stephanie Kullmann, Ulrike Toepel, Micah M Murray, Chrysa Retsa, Jardena J Puder","doi":"10.1038/s43856-026-01792-3","DOIUrl":"10.1038/s43856-026-01792-3","url":null,"abstract":"<p><strong>Background: </strong>Pregnancy-induced food cravings are associated with excessive gestational weight gain and gestational diabetes mellitus (GDM). This study aimed to measure brain responses to food pictures with varying fat and carbohydrate content during pregnancy in women with and without GDM.</p><p><strong>Methods: </strong>To characterize brain-metabolism coupling, we recorded visual evoked potentials (VEPs) in pregnant individuals with and without GDM (24-36 weeks gestation) while they viewed food pictures varying in fat and carbohydrate content. Eighteen women with untreated GDM and 21 healthy pregnant controls participated. HbA1c levels were obtained from blood samples. Statistical analyses assessed group differences and interactions.</p><p><strong>Results: </strong>Within the initial ~150 ms post-stimulus onset, VEPs exhibited Group × Fat (p = 0.047, ηp² = 0.103), but no Group × Carb, interactions: Healthy pregnant controls showed stronger responses to images of high- versus low-fat foods, whereas individuals with GDM did not. Source estimations localized these effects to brain regions associated with control, attention, and reward. Moreover, brain responses correlated with HbA1c levels in GDM (p = 0.01, ηp² = 0.351), but not with pre-pregnancy BMI in either group.</p><p><strong>Conclusion: </strong>These results provide evidence of links between spatiotemporal brain responses to visual food cues and metabolic health in pregnancy.</p>","PeriodicalId":72646,"journal":{"name":"Communications medicine","volume":"6 1","pages":""},"PeriodicalIF":7.4,"publicationDate":"2026-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13448487/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148686866","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}