Osteoarthritis (OA) is a whole-joint disease characterized by progressive structural degeneration and chronic low-grade inflammation affecting the cartilage, synovium, subchondral bone, and immune compartments. It is a complex degenerative disorder associated with substantial morbidity, heterogeneous clinical trajectories, and limited disease-modifying treatment options. Extracellular vesicles (EVs) have emerged as important mediators of intercellular communication within the OA joint microenvironment, and are implicated in pathophysiological responses to mechanical stress, inflammatory cues, and metabolic dysfunction. Through the transfer of context-dependent nucleic acid, protein, and lipid cargoes, EVs can amplify pathogenic processes in OA such as synovitis, cartilage catabolism, and cellular senescence, while also supporting reparative pathways. Understanding the mechanisms governing EV biogenesis, cargo selection, tissue targeting, and functional heterogeneity offers opportunities to identify mechanistically informed biomarkers and therapeutic strategies. This review discusses emerging concepts in EV-mediated joint communication, highlights translational potential and limitations, and outlines key priorities for advancing EV-based diagnostics and therapies in OA. However, EV-based strategies remain largely at the preclinical or experimental stage, and rigorous validation is required before they can enter routine clinical practice.
Chronic non-healing wounds remain a major clinical challenge, with limited therapeutic efficacy and a huge burden on patients. Nanozymes have emerged as promising bioactive platforms for wound repair by modulating reactive oxygen species, immune responses, angiogenesis, and antimicrobial defense. However, traditional nanozyme development primarily relies on empirical approaches, resulting in limited functional specificity and suboptimal therapeutic outcomes. Recent advances in artificial intelligence (AI) have provided promising alternatives to overcome these limitations and accelerate the rational engineering of nanozymes. This review proposes a pathology-informed design framework that links wound-specific therapeutic needs to nanozyme screening, function optimization, and iterative refinement. We synthesize current progress in AI-enabled material discovery and structure–activity prediction, function-oriented and multiobjective optimization, development of stimuli-responsive systems, and elucidation of biological mechanisms. We further examine how these approaches support pathology-guided functional matching, coordination of multifunctional interventions, treatment monitoring, and closed-loop feedback optimization. Key challenges include heterogeneous and biased datasets, limited model generalizability, weak correlation between in vitro catalytic metrics and in vivo efficacy, insufficient wound-relevant endpoints, and incomplete long-term safety assessments. Future work should strengthen the links among material properties, catalytic behavior, and tissue-repair outcomes, while incorporating pathological conditions, delivery systems, and therapeutic feedback into model optimization.
Effective therapies that enable precise drug delivery and integrated monitoring are urgently needed for fertility-sparing management of endometrial cancer (EC). Although calcium channel blockers (CCBs) exhibit anti-proliferative potential in EC, their poor tumor selectivity and systemic toxicity limit clinical translation. To overcome these limitations, a nanoparticle-mediated sonodynamic therapy (SDT) strategy is employed to deliver CCBs while integrating real-time imaging. This study develops MPA-RSP/Ce6/MDP, medroxyprogesterone (MPA)-modified nanoparticles based on a reactive oxygen species (ROS)-sensitive polymer (RSP), co-incorporating the sonosensitizer chlorin e6 (Ce6) and CCB manidipine (MDP). Upon ultrasound activation, Ce6 generates cytotoxic ROS, which trigger degradation of the ROS-responsive shell and release of MDP. The combination of ROS generation and calcium influx inhibition synergistically induces severe endoplasmic reticulum stress, thereby triggering immunogenic cell death and subsequent activation of antitumor immunity. This SDT approach further synergizes with PD-1 checkpoint inhibition, significantly enhancing therapeutic outcomes in vivo. Additionally, single-photon emission computed tomography (SPECT) imaging with co-assembled MPA-RSP/99mTc-RSP confirms efficient tumor accumulation. In summary, this study introduces an ultrasound-responsive nanoparticle that integrates calcium-homeostasis disruption, sonodynamic therapy, and immune activation for non-surgical treatment of endometrial cancer. This strategy shifts from sustained hormonal exposure to spatiotemporally controlled physical intervention, thereby establishing a multimodal foundation for EC fertility-preservation treatment.
Metabolic reprogramming is a fundamental hallmark distinguishing tumor cells from their normal counterparts. This process leads to pronounced alterations in the types and concentrations of metabolites present in the bodily fluids of cancer patients via liquid biopsy approaches. Plasma and serum, owing to their minimally invasive collection, repeatability, and ability to reflect systemic metabolic status, have emerged as optimal sample types for clinical metabolomics. These metabolic changes serve as valuable indicators for inferring disease progression and predicting patient prognosis. Lung cancer, particularly non-small cell lung cancer (NSCLC), with its high global incidence and mortality, represents a critical area where plasma metabolomics can address unmet clinical needs in prognostic prediction and therapeutic stratification. This review focuses primarily on lung cancer. However, the scarcity of studies investigating the prognostic value of metabolite alterations in lung cancer promoted the inclusion of research from other malignancies as well. This review first summarizes the current liquid biopsy metabolomics detection technologies and associated biological materials, followed by an overview of tumor-related metabolic pathway alterations. It then discusses the clinical applications of these principles in prognostic prediction and therapeutic evaluation. The aim is to provide a comprehensive overview and inspire future research directions in this field.
Extubation failure in ICU patients is associated with poor outcomes. Existing prediction models often rely on static data, missing dynamic disease fluctuations. This study introduces TrAcE, a deep learning-based model integrating static and temporal data for improved extubation failure prediction with explainable results. The model was trained and validated using MIMIC-III (Medical Information Mart for Intensive Care-III) data and tested on the LOCAL-Ext (a local database for extubation) dataset. A Transformer-based neural network with temporal fusion was used to screen extubation records (Patients planned for post-extubation non-invasive ventilation or tracheostomy were excluded). Model performance was assessed using AUROC (area under the receiver operating curve) and AUPRC (area under the precision recall curve). Explainability was ensured via Captum’s occlusion method, identifying feature attributions at both population and individual levels. TrAcE’s extubation timing was compared to the spontaneous breathing test (SBT) in LOCAL-Ext. From MIMIC-III, 5,895 patients (4,126 for training, 1,729 for validation) were selected. LOCAL-Ext included 6,765 test patients. TrAcE outperformed other models, achieving AUROCs of 0.823 (95
