Citations

At a time when there was no unifying hypothesis and no broadly effective treatments for heart failure with a preserved ejection fraction (HFpEF), it was proposed that HFpEF represented a multitude of different disorders. Accordingly, characterization of phenotypic heterogeneity was envisioned as a means of identifying novel causal pathways that could lead to new treatments while simultaneously discerning patients who would selectively benefit from a specific therapy. However, efforts over the past decade to characterize phenotypic diversity in diverse and complex ways have not achieved these goals. Subgroup analyses of neutral HFpEF trials have failed to reliably identify responders to treatments. Neither proteomics nor unsupervised cluster analysis across multiple phenotypic domains has yielded reproducible results (even in the same dataset), elucidated novel mechanistic pathways for new drug development, or identified patients most likely to benefit from treatment. In marked contrast to these efforts to characterize phenotypic heterogeneity, large-scale trials in HFpEF enrolled patients using broad eligibility criteria, and they established the benefits of sodium-glucose cotransporter 2 inhibitors and mineralocorticoid receptor antagonists without evidence for subgroup effects. However, it is noteworthy that patients enrolled in recent large-scale HFpEF trials were characterized by general uniformity with respect to the presence of excess adiposity, with central obesity being a feature of the vast majority of enrolled participants. Of note, patients with obesity showed particularly large benefits when treated with effective drugs for HFpEF. These observations raise the possibility that, if these trials had permitted the participation of patients with class III obesity or with lower natriuretic peptide levels, the magnitude of the observed treatment effects might have been greater than originally reported. These findings are consistent with a central role for visceral adiposity (and the secretion of an altered suite of adipokines) in the pathogenesis of HFpEF. Therefore, because the field of HFpEF has a unifying hypothesis (applicable to the large majority of patients), enrolled a broad population of patients in large-scale trials without subgroup interactions, and has several broadly applicable effective treatments (as is true for heart failure with a reduced ejection fraction), the motivation to characterize phenotypic heterogeneity in HFpEF in complex ways may no longer be supported or needed.

Abstract Objective This study evaluated sequential multi-site fine-tuning of large language models (LLMs), simulating incremental deployment across institutions for extracting mobility functional status from unstructured clinical notes. Materials and Methods We assembled a corpus of 600 clinical notes from 3 institutions, yielding 3810 annotated note sections (1200 training; 2610 testing). Building upon our previously established LLM mobility extraction pipeline, which utilized optimized prompt engineering, we applied parameter-efficient low-rank adaptation (LoRA) fine-tuning. We evaluated sequential multi-site fine-tuning, where the model is trained locally and only the updated LoRA weights are transferred to the next site. We compared this against joint fine-tuning, which requires pooling all multi-site patient data centrally before training. Performance was measured by micro F1-scores for 2 distinct subtasks: mobility extraction and impairment classification. Results Sequential multi-site fine-tuning yielded comparable results to centralized joint fine-tuning for mobility extraction (F1, 0.861 vs 0.894) and impairment classification (F1, 0.912 vs 0.900). Both sequential and joint fine-tuning outperformed an untuned 70B model. Additionally, single-site LoRA models showed robust cross-site generalization. Discussion Sequential multi-site fine-tuning offers a resource-friendly approach to adapt LLMs to clinical environments. By retaining previously acquired knowledge without severe interference, this strategy allows for local fine-tuning at each institution without the need to pool or share sensitive patient data required for joint fine-tuning. Conclusion Sequential multi-site fine-tuning provides a viable solution for the scalable, incremental deployment of LLMs across institutions, showing promising generalizability for clinical information extraction from electronic health record narratives.

This Viewpoint proposes restoring science-based policymaking around vaccines and adding a new component—a transparent process for explicitly identifying and weighing the inherent values and trade-offs.

Background Lumbar facet arthropathy (LFA) is a degenerative, arthritic joint disorder and a common specific cause of axial low back pain and functional impairment. Currently, no disease-modifying therapies are available for symptomatic LFA. Mesenchymal stem/stromal cell (MSC)-based therapy represents an emerging regenerative treatment modality with the promise to target the underlying pathophysiology of LFA. Despite encouraging clinical evidence supporting the safety and efficacy of MSC therapy in other osteoarthritis, its therapeutic utility in LFA remains largely unexplored. We recently completed a prospective, single-arm, open-label phase I clinical trial that demonstrated a favourable safety profile and preliminary clinical benefit following intra-articular administration of bone marrow-derived MSCs (BMSCs) in patients with symptomatic LFA. Building on these findings, we present the protocol for CellKine II , a prospective, randomised, double-blind, placebo-controlled, cross-over phase II clinical trial to evaluate the safety and preliminary efficacy of intra-articular delivery of allogeneic, culture-expanded BMSCs in patients with painful LFA. Methods and analysis This study is a single-centre, randomised, double-blind, placebo-controlled, cross-over phase II trial with a target accrual of 40 patients. Eligible patients will be randomly assigned in a 1:1 ratio to receive a single intra-articular injection of either Arm A: BMSC or Arm B: Placebo (dimethyl sulfoxide). At the 6-month follow-up, all patients will be unblinded. Patients initially randomised to Arm A will continue follow-up for up to 12 months. Patients initially randomised to Arm B may cross over to receive the BMSC treatment, followed by an additional 12 months of post-BMSC treatment follow-up, or they may continue observation for another 6 months without BMSC treatment. Overall, each patient will be followed up to 12 months after BMSC treatment. The primary endpoints are back pain, functional outcomes and adverse events at 6 months post-randomisation. Ethics and dissemination This study will be conducted in compliance with U.S. federal regulations and Mayo Clinic institutional research policies and procedures. The protocol has been approved by the Mayo Clinic Institutional Review Board (IRB) and the Data Safety Monitoring Board (IRB No. 21-013278). The results of this trial will be disseminated through publication in peer-reviewed journals and on ClinicalTrials.gov. Trial registration number NCT06001853 .

Introduction: Dysfunction of pelvic floor muscles (PFMs) can limit physical activity and health. Current screening methods are invasive or expensive, although transabdominal ultrasound (TAUS) is gaining popularity. PFM contraction causes bladder base displacement (BBD) and magnitude reflects PFM function; however, manual image analysis is labor-intensive. Artificial intelligence (AI)–assisted TAUS may offer more efficient and feasible solutions. This study compared standard analysis of BBD to measurement with AI-assisted TAUS to assess reliability and validity. Methods: A total of 20 healthy female participants (26.4 ± 8.5 years) performed 6 repeated Kegel contractions after following a bladder-filling protocol. A curvilinear transducer was placed between the anterior superior iliac spines with the probe placed transversely and then angled inferiorly toward the bladder. The Clarius Bladder AI tool automatically measured distance from bladder apex to base. Images were also saved without an AI overlay and measured manually using ImageJ. Results: Average BBD values were equivalent (manual 0.7 ± 0.5 cm; AI-assisted 0.7 ± 0.6 cm, 90% confidence interval: −0.09 to 0.04 [ P < .05]). Measures were highly related ( r = 0.97, intraclass correlation coefficients [ICC 3,k ] = 0.98) and showed good agreement (95% limits of agreement: −0.33 to 0.28 cm). Test–retest reliability was excellent for both manual (ICC 3,k = 0.94, standard error of measurement [SEM]: 0.12 cm) and AI-assisted measures (ICC 3,k = 0.94, SEM = 0.14 cm). Discussion: No systematic differences were shown between methods, which were highly related. The Clarius AI-assisted bladder tool provided valid, reliable BBD measures consistent with traditional methods. The small SEM supports clinical application for PFM assessment. Conclusion: AI-assisted TAUS appears to be an efficient noninvasive method for assessing PFM function and may be applied easily in various clinical and research settings.

Abstract Seizure forecasting has progressed from theoretical aspiration to a rapidly advancing research domain, yet clinical translation remains limited. Over the past decades, advances in algorithm development, chronic electroencephalography (EEG), wearable sensors, and the characterization of seizure cycles have demonstrated that seizure risk is not random but fluctuates according to identifiable biological rhythms and patient‐specific patterns. Forecasting algorithms have shown promising performance across diverse retrospective datasets, including intracranial EEG, subscalp recordings, wearable physiological signals, and even self‐reported diaries. However, the clinical value of forecasting is contentious. Prospective real‐world validation and regulatory approval of patient‐facing forecasting systems remain rare. This review incorporates perspectives presented at the 5th International Congress on Mobile Health and Digital Technology in Epilepsy (2025). We examine barriers impeding clinical translation and current attempts to address them. Crucially, forecasting performance cannot be evaluated in isolation from intended use. Applications range from low‐risk uses, like scheduling diagnostic monitoring or visualization of historical trends, to higher risk interventions, including medication titration and adaptive neuromodulation. Each application entails distinct performance thresholds, ethical considerations, and regulatory requirements. Translational challenges include reliable seizure annotation, nonstationarity dynamics of biological cycles, and practical constraints for real‐time deployment. Ethical concerns center on miscalibrated reliance on low‐risk states, potential anxiety associated with high‐risk advisories, and the heterogeneity of patient preferences and risk tolerance. Regulatory pathways are likely to depend on clearly defined use cases and clinically meaningful endpoints, which may extend beyond seizure counts to include quality of life, anxiety, locus of control, and other patient‐reported outcomes. Ultimately, translation will require rigorous prospective evaluation against transparent benchmarks, sustainable scientific–commercial partnerships, and integration of probabilistic risk information into clinical workflows. With careful implementation, seizure forecasting may evolve from proof‐of‐concept research into a clinically meaningful component of epilepsy management, and we remain cautiously optimistic.

Vascular stents are crucial devices in the treatment of cardiovascular diseases, and their structural design and function critically affect therapeutic efficacy and patient prognosis. Conventional stents can effectively restore vascular patency by providing mechanical support to blood vessels. However, they still face significant challenges including restenosis, thrombosis, and limited adaptability to complex patient-specific lesion characteristics. To address these limitations, drug delivery offers an important strategy to modulate the pathological microenvironment, enhance long-term vascular healing, and reduce systemic side effects. Meanwhile, advances in computational simulations have provided powerful tools for optimizing stent design through structural mechanics, hemodynamics, and drug release modeling. Computational approaches enable the rational design of stent architectures with improved mechanical stability, vascular compatibility, and therapeutic regulation. Consequently, the development of vascular stents is evolving toward the synergistic integration of drug delivery, structural optimization, and intelligent design. This review summarizes the latest advances in functional vascular stents, clinical applications and computational design. This work aims to provide valuable insights for the engineering of efficient, precise, and intelligent vascular stents for cardiovascular therapies.