Citations

Purpose of review Ophthalmology has emerged as the proving ground for medical artificial intelligence (AI) because of its imaging-centric workflows, standardized data acquisition, and well defined clinical endpoints. Recently, the field has started transitioning from narrowly trained, task-specific algorithms toward multimodal models that are trained on large datasets and can be adapted to many tasks. These models may integrate information from images, text, and clinical data and are increasingly trialed in real-world settings. This review examines recent advances and ongoing controversies in the use of AI as a basic science, translational, and clinical research tool in ophthalmology. Recent findings Recent literature highlights the growing role of self-supervised and foundation models for scalable phenotyping, biomarker discovery, and cross-task generalization, particularly in retinal imaging and optical coherence tomography (OCT). In parallel, large-scale clinical deployments, including cloud-based OCT platforms and autonomous diabetic retinopathy screening, demonstrate clinical impact while revealing persistent challenges related to workflow integration and patient generalizability. Regulatory and ethics-focused studies underscore increasing attention to validation rigor, bias, and postdeployment monitoring, exposing heterogeneity in research standards and physician practices. Summary Ophthalmic AI is maturing toward real-world clinical implementation. Future progress will depend less on performance gains alone and more on clinical alignment, rigorous validation, equity-aware evaluation, and postdeployment surveillance, alongside continued investment in basic science to advance disease understanding and inform robust, generalizable model development.

Background: Given aspirin’s biologic plausibility for antifibrotic and antineoplastic effects, we conducted a systematic review and meta-analysis to examine the association between aspirin use and major liver-related outcomes in metabolic dysfunction-associated steatotic liver disease (MASLD). To our knowledge, this is the first systematic review and meta-analysis restricted exclusively to patients with biopsy- or registry-confirmed MASLD. Methods: A comprehensive search of Ovid MEDLINE, Ovid EMBASE, Scopus, and Web of Science was performed in October 2025. Studies enrolling adults with a confirmed MASLD diagnosis were included; those with viral hepatitis or alcohol-related liver disease were excluded. Outcomes assessed included hepatocellular carcinoma (HCC), fibrosis progression, cirrhosis, all-cause and liver-related mortality, gastrointestinal (GI) bleeding, hemorrhagic stroke, and liver disease progression. Hazard ratios (HRs) with 95% CIs were pooled using random-effects models. Heterogeneity was assessed using I2 statistics. Results: Seven studies met the eligibility criteria, with approximately 720,000 individuals included. Pooled analysis showed that aspirin use was associated with a significantly lower HCC risk (HR 0.59; 95% CI 0.43–0.81; I2 = 83%). No statistically significant association was found between aspirin use and cirrhosis incidence (HR 0.55; 95% CI 0.13–2.37; I2 = 83.4%) or GI bleeding (HR 1.11; 95% CI 0.74–1.66; I2 = 98.9%). Among the two studies that explored all-cause mortality, aspirin was associated with a modest but statistically significant reduction in all-cause mortality (HR 0.86; 95% CI 0.78–0.95; I2 = 0%). Conclusions: Aspirin use is associated with a reduced risk of HCC and all-cause mortality in MASLD without significantly increasing GI bleeding or hemorrhagic strokes. These associations may reflect aspirin’s anti-inflammatory properties in liver disease. Further RCTs are needed to verify the causal role of aspirin in MASLD management.

ABSTRACT Incomplete aneurysm occlusion after flow disruption treatment strongly depends on the persistence of strong flow into the aneurysm, which in turn depends on the aneurysm and parent artery geometry as well as the device placement. However, at the time of treatment it remains challenging to predict future outcomes (complete or incomplete occlusion). This paper describes a strategy to evaluate the local hemodynamic environment created immediately after flow‐modifying device deployment to treat cerebral aneurysms and connect it to the subsequent long‐term outcomes. The approach is based on the concept of angiographic signatures that can be extracted from DSA sequences or computational models and represented as images that can be used to compare different aneurysmal environments. The methodology is demonstrated on two small series of four experimental rabbit aneurysms each, treated with either intrasaccular flow disruptors or endoluminal flow diverters. Noticeable differences in the angiographic signatures of completely and incompletely occluded aneurysms at follow‐up were observed. Specifically, larger values of mean transit times and longer time‐to‐peaks were observed deeper into the aneurysms that remained incompletely occluded in both series. The findings suggest that angiographic signatures can be used to assess the underlying hemodynamic environment immediately after device implantation and prognosticate the likelihood of future complete or incomplete occlusion.

Background: There is growing support for high-efficacy disease-modifying therapy (DMT) in multiple sclerosis (MS), but escalation (ESC) approaches remain common. Objective: To describe decision-making in a pragmatic trial of early high-efficacy treatment (EHT) versus ESC. Methods: DELIVER-MS is a multi-center, pragmatic, randomized controlled trial (RCT) with a parallel observational study (OBS), which enrolled treatment-naïve people with RRMS in 31 UK/US sites. Primary outcome was as follows: 36-month brain volume loss according to initial treatment approach (EHT vs. ESC). Stepwise multivariable logistic regression was used to predict participation in RCT versus OBS and choice of EHT versus ESC within the OBS cohort. Results: In total, 816 people with MS were enrolled. Participants declined randomization due to preference for a particular DMT (85%), efficacy concerns (20%), and safety concerns (9%). RCT versus OBS participation was associated with lower relapse rate ( p = 0.043) and greater brain parenchymal fraction ( p = 0.002). Among 374 in the OBS cohort, 125 (33%) chose ESC and 249 (67%) chose EHT. People commencing EHT had higher education attainment ( p < 0.001) and relapse rate ( p = 0.025). Conclusion: Baseline DELIVER-MS data demonstrate that participants with milder disease are more likely to participate in RCT. The choice of EHT versus ESC was associated with demographic factors and disease activity. Clinical trial registration: NCT03535298

Abstract Interstitial cells of Cajal (ICC) regulate gastrointestinal motility by generating electrical slow waves that orchestrate downstream calcium (Ca 2 + ) waves and coordinate smooth muscle contractions. A reduced ICC number has been reported in diabetic gastroparesis patients and in mouse models of delayed gastric emptying (DGE). However, functional changes in ICC propagating across the gastric regions in intact stomach are not well understood. The aim of this study was to develop an imaging approach to visualize ICC Ca 2+ waves in mouse stomach and determine spatiotemporal changes in mice with DGE. Tamoxifen‐treated B6.GCaMP5‐KitCreERT2 female mice were used as non‐diabetic controls (Non‐Db), while mice treated with streptozocin to induce diabetes were categorized as having normal (Db‐NGE) or delayed (Db‐DGE) gastric emptying based on emptying of solids. The ICC Ca 2+ network was imaged from excised intact mouse stomachs and subsequently opened along the lesser curvature and pinned flat. ICC Ca 2+ waves typically propagated along the greater curvature from the fundus–corpus line toward the antrum. The propagation length of ICC Ca 2+ activity was significantly decreased in Db‐DGE compared to Non‐Db and Db‐NGE and correlated to longer GE times. There was also a decrease in frequency, and the velocity was significantly faster in stomachs from Db‐DGE compared to Non‐Db and Db‐NGE. All experimental groups exhibited similar proportions of aberrant Ca 2+ propagation patterns (e.g., retrograde waves), but ectopic initiation only occurred in Db‐DGE stomachs. Thus, DGE is possibly driven by dysfunction in Ca 2+ activity in the ICC network that propagates rapidly over short distances, resulting in uncoordinated, non‐phasic muscle activity. image Key points Dysfunction of the interstitial cells of Cajal (ICC) has been implicated in diabetic gastroparesis but the changes in spatiotemporal properties of ICC network activity are not known. We developed an ex vivo Ca 2+ imaging approach to visualize ICC network activity across the entire stomach and determined changes in ICC Ca 2+ activity in a diabetic mouse model of delayed gastric emptying (GE). Decreased frequency and increased velocity of ICC Ca 2+ activity was observed in stomachs of diabetic mice with delayed GE. Ectopic initiation was a key feature associated with delayed GE. The propagation length of ICC Ca 2+ activity was inversely correlated to GE times and significantly reduced in stomach preparations from diabetic mice with delayed GE.

Abstract Background Ionizing radiation (IR) accelerates atherosclerosis through induction of cellular senescence, DNA damage, defective efferocytosis, and dysregulation of clonal hematopoiesis (CH) drivers. Although low-dose colchicine reduces ischemic cardiovascular events in coronary artery disease, the precise molecular mechanisms underlying its vasculoprotective effects remain incompletely defined, and whether it mitigates radiation-associated vascular injury is unknown. Methods Bone marrow–derived macrophages (BMDMs) were pretreated with low-dose colchicine and exposed to 2 Gy IR. Molecular effects were assessed by RNA-seq, immunoblotting, and molecular docking. In vivo effects were tested in a partial carotid ligation (PLCL) model using spatial proteomics. Human monocyte-derived macrophages (HMDMs) from thoracic malignancy patients were analyzed before and after radiation therapy (RT). Results Low-dose colchicine suppressed IR-induced macrophage senescence signaling while preserving NRF2 activity. In a cell-free assay, colchicine directly activated aldehyde dehydrogenase 2 (ALDH2) in a dose-dependent manner (EC 50 1–5 nM), identifying ALDH2 as a direct molecular target of colchicine. Following irradiation, colchicine restored ALDH2, reduced mitochondrial (mt)ROS - dependent p90 ribosomal S6 kinase ( p90RSK ) activation and lipid peroxidation, preserved TET2 and DNMT3A expression, and rescued impaired efferocytosis while preventing nicotinamide adenine dinucleotide (NAD⁺) and adenosine triphosphate (ATP) depletion. These protective effects were ALDH2-dependent, as they were lost with ALDH2 inhibition or depletion and were mimicked by pharmacologic ALDH2 activation. In vivo, colchicine attenuated radiation-induced atherosclerosis and macrophage senescence-associated stemness (SAS). Consistently, macrophages from patients after RT showed reduced ALDH2 with increased mtROS, lipid peroxidation, and senescence. Conclusion These findings identify ALDH2 as a previously unrecognized molecular target of colchicine that links mitochondrial redox control to suppression of radiation-induced macrophage senescence and atherosclerosis and may contribute to the efficacy of low-dose colchicine in cardiovascular disease.

Abstract Travelling waves are a prominent feature of hippocampal activity, but the mechanisms determining their propagation and influence on the cerebral cortex remain unclear. Using a biophysically-grounded model of neural activity evolving across hippocampal and cortical surfaces, we show that spatial gradients in external in-put or neural excitability are a sufficient mechanism for the emergence of travelling waves along the long axis of the hippocampus. These waves emerge only above a critical gradient threshold, propagate with biologically-plausible velocities and exhibit frequency-dependent reversals in direction across slow and fast theta regimes. When coupled to the cerebral cortex, anterior-to-posterior hippocampal waves selectively reorganise globally synchronous cortical activity into metastable travelling waves, whereas posterior-to-anterior hippocampal waves do not, revealing a directional asymmetry in hippocampal–cortical communication. Conversely, cortical waves aligned with large-scale functional gradients induce structured hippocampal waves, revealing complementary direction specific effects across the two systems. These analyses identify structured excitability gradients as a principle governing wave propagation in the hippocampus and suggest how wave-to-wave interactions may coordinate complex cortical and hippocampal interactions during mnemonic and perceptual processes.

Abstract Repetitive head impacts sustained during American football have been associated with neuropathological changes such as white matter shear injuries. However, the impact of specific factors, such as age of first exposure and cumulative head impact burden, on white matter integrity remains unclear. This study investigated in vivo white matter microstructural changes using diffusion tensor imaging and tract-based spatial statistics in 165 male former American football players (mean age 57.3 years, range 45–74) and 52 unexposed asymptomatic male controls (mean age 59.4 years, range 45–74) in the DIAGNOSE CTE Research Project. Compared to controls, former football players exhibited significantly higher fractional anisotropy (FA) in 1.97% of the white matter skeleton (1552 voxels; Cohen’s d = 0.587) and higher tissue-corrected FA (FAt) in 1.48% of the white matter skeleton (1004 voxels; Cohen’s d = 0.616). No significant differences were observed for mean diffusivity, axial diffusivity, radial diffusivity, or free water between football players and controls. Among football players, there were no significant differences in the white matter microstructure between players diagnosed with traumatic encephalopathy syndrome and those without the diagnosis. Lower FA was significantly associated with older age (P &lt; 0.00001) and an earlier age of first exposure to tackle football (P &lt; 0.01), while lower FAt was associated with greater cumulative head impact burden, specifically higher linear acceleration (P &lt; 0.04) and rotational force (P &lt; 0.02). This study highlights the influential role of exposure factors on white matter microstructure in former American football players, as well as the utility of diffusion tensor imaging to aid in characterizing the long-term effects of repetitive head impacts in contact sport athletes.