搜索结果：precision

Opportunistic fungal infections are an escalating global challenge, impacting over a billion people worldwide, particularly immunocompromised individuals. Effective management strategies are urgently needed. While laboratory culture remains the presumptive identification of fungal infections, it is time-consuming, requires expensive equipment, and suffers from observer-dependent variability. Innovative automated tools integrating artificial intelligence (AI) offer promising solutions for timely diagnosis and improved treatment outcomes. This study aims to develop and validate a self-supervised deep learning (SSL) model (DINOv2) to automatically identify clinically relevant Candida species from microscopic images. To begin with, appropriate data preparation, combining detection and segmentation techniques were conducted. Microscopic images were processed using the YOLOv4 tiny model to locate the organisms of interest, followed by image segmentation using the UNet algorithm. Study outcomes showed that YOLOv4 tiny and SSL DINOv2 detection models demonstrated high performance, achieving a mean average precision (mAP@50) of 0.908 and an AUC under the precision-recall (PR) curve of 0.949. Additionally, the semantic segmentation model yielded a Dice score of 0.930, an Intersection over Union (IOU) score of 0.874, and a Hausdorff Distance score of 67.101, confirming the reliability of the dataset for further analysis. Binary classification of two fungal species, Candida albicans and C. krusei, was performed. The SSL models achieved accuracies ranging from 0.980 to 0.988, outperforming baseline models such as Vision Transformer. Similarly, the SSL models achieved an F1 score of 0.981, significantly higher than the baseline models' score of 0.938. For four-class classification, the small version of DINOv2 trained on 224 × 224-pixel images achieved the highest recall, precision, and F1 score values of 0.977, demonstrating superior performance. Additionally, five-fold cross-validation (accuracy: 94.8-98.3%; AUC-ROC: 0.990-0.998) and an almost perfect level of agreement analysis (κ = 0.845) demonstrate robust potential as a proof-of-concept for clinical applications, pending multi-center validation. In conclusion, the hybrid models exhibited exceptional performance, highlighting their potential as a research prototype for healthcare applications, requiring rigorous prospective validation. However, further validation with diverse datasets is essential to enhance and confirm the feasibility of deploying AI models in the healthcare sector.

The ability to predict pathological complete response (pCR) prior to neoadjuvant therapy (NAT) would inform and facilitate tailored therapeutic regimens for patients with breast cancer. This study aimed to develop and internally validate an interpretable machine-learning model for pretreatment pCR prediction using routinely available clinicopathologic variables, hematologic indices, and ultrasound (US)/contrast-enhanced ultrasound (CEUS) descriptors. This retrospective analysis included 309 sequential breast cancer cases managed with neoadjuvant therapy followed by surgical intervention. Patients were randomly divided into training and testing cohorts at a ratio of 8:2 using stratified sampling according to pCR status. Feature selection was performed exclusively in the training cohort using least absolute shrinkage and selection operator (LASSO) regression. Five machine-learning classifiers, including CatBoost, LightGBM, XGBoost, SVM, and CART, were developed and evaluated. Model performance was assessed using the area under the receiver operating characteristic curve (AUC, with 95% CI), accuracy, specificity, sensitivity, precision, recall, F1-score, G-mean, Brier score, decision curve analysis (DCA) and calibration analysis. Shapley Additive Explanations (SHAP) were used for model interpretation. pCR was achieved in 29.8% (92/309) patients. LASSO regression selected eight predictors, including CEA, clinical T stage, ER, PR, HER2, Ki-67 index, Hyperechoic halo on US, and Range expansion on CEUS. In the testing cohort, CatBoost achieved an AUC of 0.8295 (95% CI: 0.7166-0.9231) and showed the highest overall accuracy (0.8065), specificity (0.8182), and precision (0.6364). CatBoost also demonstrated acceptable calibration, with a Brier score of 0.1595, and favorable net benefit across clinically relevant threshold probabilities. SHAP analysis identified HER2 status, CEA, clinical T stage, and hyperechoic halo as the leading contributors to the final model. An interpretable pretreatment CatBoost model showed good discrimination, acceptable calibration, and favorable clinical net benefit for predicting pCR after NAT in breast cancer. This approach may help support pretreatment risk stratification and inform multidisciplinary discussions.

Precision oncology aims to tailor treatment according to tumor-specific molecular alterations, but the success of aberration-guided therapies has been limited in clinical trials. Here, we develop an integrated whole-genome and transcriptome workflow to systematically distinguish functionally credible, predictive driver aberrations from non-functional alterations across all classes of genomic events. We applied the integrated omics workflow to 335 patients with ovarian high-grade serous carcinoma (HGSC) enrolled in the observational DECIDER study. Tumor samples were collected from multiple cancer sites as part of the standard cancer care. DNA and RNA were extracted together from snap-frozen tumor samples and sent to whole-genome and transcriptome sequencing. Sequencing data were processed with the Anduril 2 pipeline for detection and validation of short somatic changes and with the HMW toolkit and the nf-core/rnafusion pipeline for assessment of structural changes. Aberration-specific drug sensitivity was tested in patient-derived organoids with a drug screen combining targeted agents and chemotherapy. Using an agnostic integrated omics analysis, we identified clinically relevant ESCAT Tier II-III alterations in more than 40% of the patients, even though 58% of all nominally pathogenic variants proved to be false positives. Credible aberrations were predominantly clonal, detected across anatomical sites, and preserved from diagnosis to relapse, indicating early establishment during tumor evolution. The most recurrent actionable event was NF1 deficiency, which was associated with a robust transcriptional footprint and marked sensitivity to KRAS- and MEK-inhibition in patient-derived organoids. Notably, integrated DNA-RNA analysis enabled discrimination of treatment-guiding aberrations from false-positive findings that would otherwise misinform treatment selection and confound clinical trial outcomes. Our findings provide a strategy for more reliable biomarker detection in precision oncology, inform biomarker-guided clinical trial design, and reveal unexploited therapeutic vulnerabilities in HGSC.

A new simple spectrofluorimetric method was developed and validated to determine apixaban in both its authentic form and drug formulations. It depends on the enhancement of the native fluorescence of apixaban using sodium dodecyl sulfate (SDS) without requiring pH adjustment. Different surfactants and factors affecting apixaban's fluorescence were studied in order to identify the optimal conditions with the highest sensitivity. The proposed approach was validated according to ICH guidelines, showing high accuracy and precision. The calibration curve exhibited a linear range of (0.1-2.0 μg/mL) demonstrating a competitive sensitivity with LOD of 0.03 μg/mL and LOQ of 0.09 μg/mL. The presented method was adapted for content uniformity testing (according to the USP) and for drug determination in spiked plasma samples, which were not applied in previously reported spectrofluorimetric methods. Compared with conventional HPLC techniques, the method reduces organic solvent consumption by over 80%, as distilled water is the primary diluting solvent while maintaining comparable sensitivity and precision. The method's greenness and blueness were also evaluated using the Eco-scale, AGREE, and BAGI tools, yielding higher scores (Eco-scale = 86, AGREE = 0.76) than typical HPLC methods (Eco-scale ≈70-75, AGREE ≈0.60), highlighting the environmental friendliness and practical applicability of the proposed procedure.

Weed pressure causes global crop yield losses of 10-34%, while the deployment of deep learning-based weed detection systems at scale remains constrained by the high cost of bounding-box annotation across diverse field environments. This study addresses this annotation bottleneck in precision agriculture by proposing WEEDINO-YOLOv12, a label-efficient weed detection framework that transfers global-average-pooled feature distributions from a frozen DINOv3 ViT-B/16 teacher into a lightweight YOLOv12n backbone through feature-distribution distillation on unlabeled agricultural imagery, followed by supervised fine-tuning on a limited labeled subset. To rigorously evaluate the proposed framework, we present a controlled empirical benchmark comparing four training regimes: fully supervised YOLOv12n, semi-supervised Soft Teacher, self-supervised BYOL, and the proposed DINOv3 distillation approach. All methods are assessed using a common YOLOv12n backbone, consistent evaluation metrics, matched controls, and multi-seed reporting. External validation on the multi-class CottonWeedDet12 dataset further examines whether the observed label-efficient behaviour generalises beyond the single-class Roboflow Weeds benchmark. Across matched 20%-label settings, WEEDINO-YOLOv12 improved mAP@0.5:0.95 from 0.6402 ± 0.0271 to 0.6517 ± 0.0087 on the Roboflow fixed split and from 0.7987 ± 0.0154 to 0.8083 ± 0.0078 on CottonWeedDet12. Full-label supervision remained the strongest overall setting, indicating that the proposed method provides modest but consistent annotation-efficiency gains rather than replacing fully supervised training. High-resolution fine-tuning at 896 × 896 pixels is analysed separately because it can improve localisation independently of the distillation stage. A Streamlit-based deployment prototype further demonstrates the practical accessibility of the framework for agronomists and precision-agriculture users without requiring direct interaction with deep learning code.

Brain metastases (BrM) affect up to 30% of patients with solid tumors, yet durable intracranial control remains rare, and the biological drivers of this poor prognosis are incompletely understood. Patient-derived resources, such as clinical cohorts, biobanks, functional ex vivo models, and multi-omic platforms, are central to closing this gap, but their generation and integration face substantial logistical and technical hurdles. Drawing on the RISEbrain consortium's experience, this Perspective examines BrM-focused cohorts and biobanks, highlighting the underused potential of rapid autopsy programs to capture early metastatic seeding. We discuss patient-derived organotypic cultures and emerging organoid-based "avatar" systems as functional platforms for therapeutic profiling, alongside the complementary strengths of bulk and single-cell/-nucleus transcriptomics. We outline how spatial transcriptomics and proteomics are resolving the architecture of the BrM microenvironment, and assess liquid biopsy approaches, including emerging photonic biosensors, for non-invasive monitoring. Together, these resources form an interdependent toolkit whose coordinated deployment will advance early detection, prevention, and precision treatment of BrM.

To provide an overview of diagnostic tests for Stage B heart failure (SBHF), synthesizing evidence from guidelines and clinical studies. Advances in diagnostic technologies have expanded the ability to identify subclinical myocardial remodelling and early myocardial injury before symptom onset. We highlight the central role of transthoracic echocardiography as the cornerstone diagnostic modality for detecting subclinical myocardial remodelling and dysfunction, including the use of speckle tracking echocardiography. In parallel, circulating biomarkers, especially natriuretic peptides and high-sensitivity cardiac troponins, can play important roles in the detection and risk stratification of SBHF. Additional diagnostic approaches, including electrocardiography, chest X-ray, cardiac magnetic resonance imaging, cardiac computed tomography, nuclear imaging, and exercise stress testing, are reviewed for their adjunctive roles in selected clinical contexts. Emerging applications of artificial intelligence are explored as promising strategies to increase the diagnostic precision, scalability, and early detection of SBHF in clinical practice. SBHF - representing a subclinical phase of HF characterized by structural cardiac abnormalities, functional impairment, or persistently abnormal cardiac biomarkers in individuals - has historically been difficult to recognize in the community. Advances in imaging, biomarkers, and AI may improve the feasibility of detecting this entity, creating a crucial window for intervention, because timely risk stratification and preventive strategies during SBHF may attenuate progression to symptomatic HF and reduce its long-term clinical and economic burden.

Major depressive disorder is among the most prevalent and disabling conditions in global medicine, yet its biological underpinnings remain incompletely understood, and current pharmacological treatments fail to produce adequate responses in approximately one-third of patients. A rapidly accumulating body of evidence has not simply challenged the long-dominant monoamine deficiency hypothesis but has provided a mechanistic framework that may explain many of the observed monoaminergic alterations in depressive cohorts, including IDO1-driven serotonin depletion, cytokine-mediated AMPA receptor internalization, and HPA-immune feedback dysregulation. Neuroinflammation, particularly glial activation across microglial, astrocytic, and oligodendrocyte lineages, and its downstream consequences for tryptophan metabolism, glutamatergic transmission, synaptic plasticity, and neurotrophic signaling, represents a central pathophysiological mechanism in a substantial subgroup of depressed patients. Peripheral inflammatory markers, including C-reactive protein, interleukin-6, and tumor necrosis factor-alpha, are elevated in a significant proportion of individuals with major depressive disorder, and these elevations predict poor response to conventional antidepressants while identifying patients who may respond preferentially to anti-inflammatory strategies. Post-mortem studies, positron emission tomography imaging of translocator protein density, and transcriptomic analyses of brain tissue have collectively provided consistent, though not yet fully definitive, evidence that microglial activation is a neurobiological feature of depression rather than a consequence of comorbid physical illness. This review synthesizes current mechanistic understanding of the neuroinflammatory hypothesis of depression, examines the evidence base from epidemiological, biomarker, neuroimaging, and interventional studies, including null findings and methodological limitations, evaluates emerging therapeutic strategies targeting the immune-brain interface, and identifies the critical questions that will determine whether immunopsychiatry fulfills its promise as a precision medicine framework for treatment-resistant depression.

Psychotic experiences are commonly reported in population-based surveys, yet brief self-report screening instruments often yield high false-positive rates by capturing normative, culturally sanctioned, or other non-psychotic phenomena. Large surveys and administrative datasets increasingly include free-text responses that could contextualize these endorsements; however, qualitative data are under-analyzed due to resource constraints. Using data from a US adult online survey, respondents who endorsed at least one item on the abbreviated World Health Organization Composite International Diagnostic Interview (WHO CIDI) psychosis screen were asked whether they could describe their experiences in an open-ended format. Using reflexive thematic analysis, three coders developed codes that were grouped into categories, and ultimately classified responses as probably psychotic, probably not psychotic, or unclear. Content analysis identified 13 thematic categories, including hallucinations, paranoia, emotional distress, interpersonal conflict, spiritual or paranormal beliefs, sleep-related experiences, and unintelligible responses. Nearly half of the responses were classified as unlikely to reflect psychotic experiences, while approximately 9% were deemed probably psychotic and 41% remained unclear due to insufficient context. Many responses reflected affective distress, stress-related interpretations, grief, or ambiguous experiences rather than clear psychotic phenomena. Descriptions capturing distress, functional impact, and insight were clinically informative but were inconsistently reported. Brief qualitative descriptions accompanying psychosis screening items provide valuable context that can clarify whether psychotic experiences reflect clinically meaningful psychotic phenomena or normative experiences. Integrating qualitative free-response fields may improve the interpretability and precision of psychosis screening, inform early detection efforts, and reduce the risk of over-pathologization.

Gut microbiota-derived extracellular vesicles have emerged as crucial mediators in microbe-host communication, not only facilitating intracellular communication, quorum sensing, and horizontal gene transfer among bacteria but also playing a central role in cross-kingdom dialogue. In recent years, bacterial extracellular vesicles (BEVs) have attracted widespread attention due to their ability to carry a diverse array of bioactive molecules-such as proteins, lipids, and nucleic acids-and deliver them to host cells, thereby precisely regulating host metabolic and immune homeostasis. This review systematically elaborates the entire biological process of BEVs, from their biogenesis to functional interactions with host cells, with a specific emphasis on revealing their roles in the pathogenesis of various metabolic diseases-including obesity, type 2 diabetes (T2DM), metabolic dysfunction-associated steatotic liver disease (MASLD), atherosclerosis, and hypertension-at both molecular and cellular levels. Furthermore, leveraging their inherent stability, biocompatibility, and targeting capabilities, we discuss the translational potential and challenges of BEVs in the diagnosis and treatment of metabolic disorders. Beyond summarizing the latest research advances on BEVs in metabolic disorders, this review provides a critical analysis of current mechanistic insights and clinical translation pathways, aiming to establish a theoretical framework for developing novel microbiome-based metabolic interventions. Deciphering the BEV-mediated microbiota-host interaction network holds promise for pioneering new strategies for the precision prevention and treatment of metabolic disease.

Postoperative recurrence remains a major obstacle to durable remission in patients with solid tumors, even after complete macroscopic resection. Growing evidence suggests that surgery creates a transient yet highly permissive biological window characterized by inflammatory signaling, coagulation activation, endothelial disruption, and systemic immune suppression. Together, these processes foster a protective niche that enables microscopic residual disease to evade immune surveillance and initiate metastatic outgrowth. Although modern adjuvant therapies have improved outcomes, their effectiveness is often limited by inadequate tumor-site specificity, systemic toxicity, poor immune cell trafficking, and tumor heterogeneity. Consequently, a critical unmet clinical need persists for biologically precise strategies capable of eliminating residual tumor cells at their point of vulnerability. Platelets, traditionally viewed as mediators of hemostasis, are now recognized as active regulators of tumor progression. By facilitating fibrin deposition, shielding circulating tumor cells from immune attack, and shaping inflammatory networks, platelets inadvertently support the survival of postoperative tumors. Paradoxically, these same wound-targeting properties create a compelling therapeutic opportunity: leveraging platelet-driven homing mechanisms to direct immunotherapy precisely to fibrin-rich surgical beds where recurrence often originates. In this review, we propose a platelet-guided CAR-T platform that leverages endogenous wound biology to create a precision immunotherapeutic delivery system. This strategy integrates platelet membrane cloaking or platelet-CAR-T conjugation with thrombin-responsive biomaterial depots to enhance local effector retention, amplify effector-to-target ratios, and prolong functional persistence. Programmable safety features, including affinity tuning, logic-gated activation, and inducible suicide switches, are used to reduce thrombo-inflammatory risk while preserving therapeutic efficacy. These mechanisms restrict activity to appropriate contexts and allow controlled shutdown in case of adverse events, improving overall safety. When coupled with minimal residual disease-guided patient selection using circulating biomarkers, this approach establishes a clinically actionable framework for perioperative intervention. Emerging preclinical evidence suggests that localized platelet-assisted delivery can reduce circulating tumor cell burden, enhance antigen presentation when combined with immune adjuvants, and suppress recurrence more effectively than systemic therapies. With rigorous safety validation, scalable manufacturing, and biomarker-enriched clinical trials, platelet-guided CAR-T therapy has the potential to transform the postoperative microenvironment from a sanctuary of tumor survival into a targeted domain for durable immune-mediated eradication.Clinical trial numberNot applicable.

Pathogenic mutations in the DNA polymerase ε (POLE) exonuclease domain define a rare but clinically distinct subset of microsatellite-stable (MSS) colorectal cancers (CRCs) characterized by hypermutation and exceptional immune checkpoint blockade sensitivity. Yet POLE testing is not routinely performed, leaving immunotherapy-eligible patients undetected. Because most diagnostic multigene panels do not include POLE, strategies enabling its recognition from routine molecular data are needed. We analyzed 675 CRC cases sequenced using a small targeted NGS panel. Tumors with ≥6 non-synonymous SNVs were flagged as potentially hypermutated. Confirmatory POLE sequencing and comprehensive genomic profiling (CGP) were performed in preselected cases. Findings were validated using two external POLE-mutant CRCs and TCGA-COAD/READ cohorts (>1000 CRCs in total). All POLE-mutant CRCs (n = 5 of 15 flagged cases; two external validation cases) showed exonuclease domain hotspot mutations, pMMR/MSS status, yet MSI-like histopathology. These tumors exhibited predominantly ultra-high TMB, low dbSNP overlap, C>T transition bias, and disrupted co-mutation patterns - canonical POLE-driven hypermutation features. In TCGA, 41/43 POLE-mutant CRCs carried panel-detectable co-mutations. Routine small-panel NGS data can flag candidate POLE-mutant MSS CRCs for confirmatory testing, enabling detection of immunotherapy-responsive tumors otherwise missed. Integrated with AI-based POLE/MSI prediction from H&E slides, this supports multimodal diagnostic workflows enhancing precision immuno-oncology in CRC. Trial registration: NA.

To investigate the expression level of hypoxia-inducible factor-1&#x3b1; (HIF-1&#x3b1;)/vascular endothelial growth factor (VEGF) signaling pathway in elderly patients with acute myocardial infarction (AMI), and to analyze its relationship with the prognosis of AMI. A total of 160 elderly patients with AMI were selected from October 2023 to September 2024 in Nanjing Medical University Affiliated Wuxi People's Hospital as the study group, 160 individuals who underwent physical examination during the same period were selected as the control group according to the 1:1 matching principle. The levels of serum HIF-1&#x3b1; and VEGF were compared between the two groups. The study group was treated with percutaneous coronary intervention (PCI). According to the prognosis of patients at 12 months after operation, they were divided into good subgroup and poor subgroup. The clinical data, serum HIF-1&#x3b1; and VEGF levels of the two subgroups were compared. The effect of serum HIF-1&#x3b1; and VEGF on prognosis and its predictive value were analyzed. The accuracy-recall rate (PR) curve was drawn to evaluate the performance of combined prediction. The area under the curve (AUC) was calculated, the DeLong test was used to compare AUCs, and the precision-recall (PR) curve was drawn to evaluate the performance of combined prediction. The levels of serum HIF-1&#x3b1; and VEGF in the study group were higher than those in the control group (P&#x2009;<&#x2009;0.05). The time from onset to admission, the proportion of Killip class IV, the proportion of multivessel disease, N-terminal pro-brain natriuretic peptide (NT-proBNP), HIF-1&#x3b1; and VEGF in the poor subgroup were higher than those in the good subgroup, while left ventricular ejection fraction was lower (P&#x2009;<&#x2009;0.05). Before and after correction of multivessel disease and NT-proBNP, serum HIF-1&#x3b1; and VEGF levels were the influencing factors of prognosis in patients with AMI (P&#x2009;<&#x2009;0.05). The AUC of serum HIF-1&#x3b1; and VEGF levels alone and combined prediction were 0.717,0.748 and 0.900, respectively, the AUC of combined prediction was significantly higher than that of the two alone (Z&#x2009;=&#x2009;3.315,2.832, P&#x2009;=&#x2009;0.001,0.005), and the best sensitivity and specificity were 83.33% and 84.87%. The DCA curve showed that the combined prediction of serum HIF-1&#x3b1; and VEGF levels in the probability range of 5%-70% could obtain significant positive net benefits. The PR curve showed that the PR-AUC value of serum HIF-1&#x3b1; and VEGF levels in evaluating prognosis was 0.767, which had a high recall rate and accuracy rate. Serum HIF-1&#x3b1; and VEGF levels are highly expressed in elderly patients with AMI, and are closely related to the prognosis of patients. The combination of the two can be used to predict the prognosis of patients, provide a reference for the assessment and prognosis of elderly patients with AMI, facilitate early identification of high-risk groups, and provide a reference for subsequent personalized program development.

Deciphering how developmental gene-regulatory programs interface with lineage-specific reproductive novelties in Bemisia tabaci demands a temporally resolved, integrative molecular framework. Using deep miRNA sequencing, high-throughput transcriptomics, quantitative proteomics, two-dimensional proteo-mapping, and spatially multiplexed FISH, we delineate the circuitry governing whitefly ontogeny. This atlas uncovers a previously unrecognized morphological innovation: a transporter-enriched egg stalk operating as an active, metabolically competent nutrient-harvesting appendage. Integrated temporal profiling indicates that the strongest molecular reprogramming in B. tabaci occurs at hatching and adult emergence, consistent with conserved developmental transition points described in other insects. Conserved miRNAs (Btab-mir-34 and Btab-mir-2944b) were associated with early developmental transitions, whereas clade-restricted miRNAs (Btab-mir-307a, Btab-mir-352a, and Btab-mir-107a) were predicted to regulate metabolic, detoxification, and chemosensory networks. Spatial FISH supports that Btab-novel-mir-018a represses vitellogenin during late-nymphal stages, establishing a developmentally gated post-transcriptional module relieved at adult emergence. Comparison of egg versus isolated stalk tissue proteomics revealed a transporter-enriched molecular signature for the B. tabaci egg stalk, extending earlier physiological evidence for pedicel-mediated water and solute uptake. Localization of NaPi-III and RNAi-associated egg viability phenotypes further supports a transport-related role for this structure during embryogenesis. Collectively, this study provides a stage-wise molecular reconfiguration framework toward understanding the developmental and evolutionary architecture of B. tabaci. Furthermore, preliminary evidence for egg-stalk-associated transport identifies a candidate ontogeny-specific process that may inform future precision pest-management strategies.

Sleep profoundly impacts health, yet current gold-standard Polysomnography (PSG) is constrained by cost, discomfort, and limited scalability for longitudinal monitoring. Ballistocardiography (BCG) offers a non-invasive and user-friendly alternative but often lacks the precision needed for reliable real-world applications. To address this gap, we propose BCGNet, a two-stage transfer learning model that is first pre-trained on 580,865&#x2009;h of PSG and then fine-tuned and validated on 15,081&#x2009;h of BCG (total 595,946&#x2009;h of recordings). Across multiple validation cohorts, BCGNet achieves strong performance in 4-class sleep staging (F1: 0.710-0.817), Apnea-Hypopnea Index (AHI3%) estimation (Pearson's r&#x2009;>&#x2009;0.95), and robust quantification of sleep continuity and architecture (ICC and Pearson's r generally >0.8). Notably, BCGNet maintains strong performance even on short daytime naps and demonstrates excellent generalizability across diverse external datasets. Deployed as a portable, contactless sleep tracking mat, BCGNet represents a major step towards scalable, user-friendly solutions for longitudinal home sleep monitoring, with important implications for population screening and personalized sleep medicine.

Long non-coding RNAs and N6-methyladenosine RNA methylation represent two pivotal layers of gene regulation. Their extensive crosstalk forms a sophisticated bidirectional network that is fundamentally rewired in cancer. This review synthesizes current knowledge to elucidate the principles and consequences of this synergistic axis. We detail how m6A modification dictates long non-coding RNA stability, splicing, localization, and function through recruitment of distinct "reader" proteins, with one "reader" family primarily mediating decay while another promotes stabilization. Conversely, we examine how long non-coding RNAs act as scaffolds, guides, and decoys to modulate the activity and specificity of the m6A machinery, establishing powerful feedforward and feedback loops. This reciprocal regulation converges on multiple cancer hallmarks, including proliferation, metabolic reprogramming, immune evasion, stemness, and therapeutic resistance. We critically discuss experimental strategies to establish causal relationships, including site-directed mutagenesis, CRISPR-based editing, and rescue assays. We also evaluate current methodological limitations in m6A detection, from antibody-dependent approaches to emerging nanopore sequencing, and highlight how single-cell and spatial transcriptomic technologies can resolve cell-state-specific networks within the tumor microenvironment. From a translational perspective, we compare small molecule inhibitors targeting m6A "writers" with RNA-based therapies, addressing their respective delivery challenges and toxicity concerns. Finally, we outline how m6A-related long non-coding RNA signatures serve as prognostic biomarkers and liquid biopsy tools for non-invasive cancer monitoring. By integrating molecular mechanisms with clinical perspectives, this review charts a roadmap for targeting the epitranscriptomic-long non-coding RNA circuit in precision oncology.

Periprosthetic joint infection (PJI) is one of the most serious complications following artificial joint replacement, posing significant treatment challenges and a heavy clinical burden. Debridement, antibiotics, and implant retention (DAIR) has become the preferred treatment strategy for acute PJI due to its advantages of minimal invasiveness, functional preservation, and cost-effectiveness. This article reviews the latest advances in optimizing the indications, timing of intervention, key technical details, and multimodal anti-infection strategies for DAIR. It focuses on discussing patient selection models (e.g., KLIC score), the impact of modular component exchange on biofilm eradication, the synergistic role of chemical debridement, and the central position of biofilm-active therapy in penetrating biofilms. Furthermore, this article proposes that future efforts should integrate molecular diagnostics, intelligent antimicrobial materials, and artificial intelligence prediction models to achieve individualized precision therapy for PJI, providing a theoretical basis for clinical decision-making.

Immune checkpoint inhibitors (ICIs) are a standard treatment across cancers, yet most patients do not respond, and existing biomarkers generalize poorly across tumor types and therapies. Here we present COMPASS, a pan-cancer foundation model that predicts immunotherapy response from bulk tumor transcriptomes using a concept bottleneck transformer. COMPASS encodes gene expression through 44 biologically grounded immune concepts representing immune cell states, tumor-microenvironment interaction and signaling pathways. Trained on 10,184 tumors across 33 cancer types, COMPASS achieves better average performance than 22 methods across 16 clinical cohorts spanning seven cancers and six ICIs, improving accuracy by 8.5% and area under the precision-recall curve by 15.7% on average across cohorts. COMPASS generalizes to cancer types and treatments not represented during fine-tuning and may inform indication selection and patient stratification. In survival analyses, patients classified by COMPASS as responders had longer overall survival (hazard ratio&#x2009;=&#x2009;4.7, P&#x2009;<&#x2009;0.0001). Personalized response maps connect gene expression to immune concepts, identifying programs associated with response and resistance; in immune-inflamed non-responders, COMPASS highlights programs including TGF&#x3b2; signaling, endothelial exclusion, CD4+ T cell dysfunction and B cell deficiency. COMPASS predicts immunotherapy response and provides hypothesis-generating mechanistic insight for trial design and translational studies.