Parahoric Lusztig induction gives a broad class of virtual smooth representations of parahoric subgroups in a p-adic group, serving as a natural generalization of classical Lusztig induction. This construction has useful applications in the representation theory of p-adic groups. In this paper, we prove a scalar product formula for parahoric Lusztig induction under the (L, G)-generic condition, which can be viewed as a parahoric analogue of a classical result of Lusztig in 1976. As an application, we describe the irreducible decomposition of the parahoric Deligne-Lusztig representation in the case of elliptic tori. (c) 2026 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
Accurate prediction of treatment response and selection of optimal treatments remain challenging in epilepsy management. With no reliable surrogate biomarkers for treatment response, the current process of selecting an antiseizure medication remains largely a trial-and-error approach. Other non-pharmacological treatment options, such as epilepsy surgery, are viable alternatives for patients with drug-resistant epilepsy. Statistical and machine learning techniques have been used to predict seizure outcomes associated with antiseizure medications and epilepsy surgery. Recent breakthroughs in deep learning have unveiled new pathways and opportunities, potentially revolutionising personalised treatment selection in health care. In this Review, we explore a broad range of studies that have used various statistical and machine learning methodologies, with particular emphasis on state-of-the-art deep learning techniques to predict the outcomes of both pharmaceutical and surgical treatments for epilepsy. We also review potential future research trajectories and address the inherent challenges of incorporating machine learning into the clinical management of epilepsy.
Neutrophil dysfunction and activation of the absent in melanoma 2 (AIM2) inflammasome have been implicated in ulcerative colitis (UC) pathogenesis. However, whether these two immune responses interact synergistically and their precise role in UC remains unclear. Here, we performed bioinformatics and clinical analyses to assess neutrophil extracellular traps (NETs) and AIM2 inflammasome cascades in UC, including their correlation. Dual immunofluorescence revealed the predominant cell types expressing AIM2. Neutrophil activation and NETs release were examined in Fusobacterium nucleatum ( Fn )-stimulated cell models, while in vitro and in vivo studies investigated the effect of NETs on AIM2 inflammasome activation, UC severity, and therapeutic potential. Bioinformatics revealed co-up-regulation and significant co-expression of AIM2 inflammasome cascade-related genes and NETs-related genes in active UC. Confocal imaging showed that AIM2 primarily localized to intestinal epithelial cells in intestinal biopsies from UC patients and colon tissues from UC model mice. Fn infection partially promoted neutrophil activation and the release of NETs via neutrophil extracellular vesicles (nEVs). NETs were carried by nEVs and internalized by intestinal epithelial cells, subsequently triggering AIM2 inflammasome-dependent pyroptosis. In vivo studies demonstrated that elevated AIM2 inflammasome-related indicators triggered by Fn -stimulated nEVs-NETs were related to UC severity. Therapeutically, blockade of the nEVs-NETs/AIM2 inflammasome cascade using the NET-degrading agent DNase I or the AIM2 inhibitor A151 effectively attenuated colitis. Collectively, AIM2 inflammasome cascade activation by NETs-laden nEVs triggered by Fn exacerbates ulcerative colitis, highlighting the nEVs-NETs/AIM2 inflammasome cascade as a promising therapeutic target.
Oral-image decision support requires reliable perception, calibrated uncertainty, auditable explanations, and controlled workflow integration. We present X-ODA, a clinician-supervised framework for 2D RGB intraoral photographs and standardized 2D views rendered from optical scans. The system combines automatic dentition localization, 10-class dominant visible-phenotype classification, auxiliary multi-label coexistence analysis, probability calibration, model-specific explainability, schema-constrained LLM reporting, deterministic routing, and dentist confirmation. On a locked group-disjoint test set (n = 1,275), EfficientNet-B4 achieved the highest observed macro-F1 (94.6%), while the prespecified VGG16 audit configuration achieved 92.4%. Automatic ROI processing increased VGG16 macro-F1 from 0.898 to 0.924, and temperature scaling reduced expected calibration error from 0.084 to 0.021. A same-supervision comparison showed similar joint localization-classification success for modular and single-stage architectures, with lower latency for the joint model. Grad-CAM produced an 88.9% confidence decrease versus 18.5% for matched random masks, with expertlocalization Dice of 0.72; hard-case C1-C9 macro-F1 was 82.6%. In a locked 500-event simulated workflow bank, routing accuracy was 98.2%, and evidence-link validation reduced the conditional report false-finding rate from 10.6% to 2.8%. Against deterministic-template reporting with identical structured evidence, the constrained LLM improved documentation usability and reduced editing burden and report-finalization time, while deterministic generation remained faster and produced no observed unsupported narrative findings. In 2,160 reviews by 12 dentists, full X-ODA reduced mean total review time to 3.0 min, while agreement did not differ significantly from visual assistance alone. X-ODA supports bounded optical-image review and documentation rather than comprehensive multimodal diagnosis.
Introduction: Psoriasis is a chronic immune-mediated inflammatory skin disease with a complex genetic background, susceptible to psoriatic arthritis as a complication. However, interpretation of genetic associations from bulk tissue analyses is limited by cellular heterogeneity, hindering mechanistic inference. This study aimed to explore cell-type-specific causal regulatory mechanisms between psoriasis and periodontitis. Methods: We integrated psoriasis and periodontitis genome-wide association data with three single-cell expression quantitative trait locus (sc-eQTL) datasets (OneK1K, sc-eQTLGen, and dynamic CD4+ T-cell activation). Two-sample Mendelian randomization (MR) was supplemented by bulk eQTL, plasma protein QTL, and Steiger directionality analyses. Prioritized genes were further examined using single-cell transcriptomics, network-based in silico perturbation, functional enrichment, drug-signature analysis, and molecular docking. Results: Cell-type-resolved MR analyses revealed heterogeneous and context-dependent genetic effects across immune cell subsets, with both concordant and discordant associations between psoriasis and periodontitis. Among the prioritised candidates, LRRC25 showed nominally significant MR associations with both diseases, although these associations were not corrected for multiple testing. Bulk blood-derived eQTL analyses showed consistent inverse associations with both diseases, whereas naive B cell-specific MR analyses showed associations in the opposite direction. Single-cell RNA sequencing showed that LRRC25 expression was predominantly detected in monocytes, whereas only a small proportion of naive B cells expressed LRRC25 . In silico LRRC25 perturbation in monocytes produced concordant transcriptional changes across psoriasis and periodontitis, with FCGR3A identified as a central network hub. Functional enrichment analyses implicated innate immune, type I interferon and Fc receptor-related pathways. Conclusions: LRRC25 emerged as a putative shared immunoregulatory candidate, although evidence varied across molecular levels and immune cell contexts. No statistically significant shared protein-level association was identified. Our findings suggest that bulk eQTL inference may obscure cell-type-specific regulatory signals and highlight the value of single-cell-informed gene prioritization. Further experimental validation is needed.
