Search Results (388)

Alzheimer’s disease (AD) classification from structural magnetic resonance imaging (MRI) may benefit from weak supervision that uses clinically meaningful but imperfect supervisory signals. We evaluated a weakly supervised framework in which a multilayer perceptron (MLP) trained on age, sex, and Mini-Mental State Examination (MMSE) scores generated clinical pseudo-labels to initialize a patch-based fully convolutional network (FCN). For 260 Alzheimer’s Disease Neuroimaging Initiative (ADNI) training participants, subsequent refinement combined 80% of the preceding MRI-model probability with 20% of the participant’s ground-truth diagnostic label. This design preserves a dominant pseudo-label/self-training component while using partial diagnostic guidance to stabilize refinement. The FCN generated whole-brain probability maps, and selected voxel probabilities were classified by a second MLP. The framework was developed using ADNI (n = 417). Using ADNI validation data only, iteration 3 and a classification threshold of 0.5 were selected and then applied unchanged to the held-out ADNI test set and the external AIBL (n = 182), FHS (n = 102), and NACC (n = 265) cohorts. The selected model achieved F1 scores of 0.853 in ADNI, 0.707 in AIBL, 0.765 in FHS, and 0.807 in NACC. These results support the feasibility and cross-cohort transferability of clinical pseudo-label-based weak supervision for MRI classification. The framework is not intended to be label-free; rather, it provides a transparent strategy for integrating imperfect clinical pseudo-labels with partially weighted diagnostic guidance during training. Full article

Philadelphia-negative (Ph-negative) myeloproliferative neoplasms (MPNs) include polycythemia vera (PV), essential thrombocythemia (ET), and primary myelofibrosis (PMF), which are clonal hematopoietic disorders caused by somatic gene mutations in the JAK2, CALR, or MPL genes. Mutations activate the JAK–STAT pathway and disrupt NF-κB signaling, leading to a chronic inflammatory state caused by pro-inflammatory cytokines and reactive oxygen species (ROS). This altered microenvironment causes serious clinical features of the disease, such as bone marrow fibrosis, splenomegaly, vascular niche remodeling, and a greater probability of thrombosis or secondary leukemic transformation. Concurrently, MPNs cause both severe immune dysregulation and tumor evasion, as evidenced by progressive lymphopenia, T and B cell exhaustion, Natural Killer cell maturation arrest, and the accumulation of myeloid-derived suppressor cells. Although FDA-approved JAK1/JAK2 inhibitors ruxolitinib, fedratinib pacritinib and momelotinib effectively reduce splenomegaly and symptom burden and have demonstrated survival benefits in clinical trials, their ability to eliminate malignant clones or induce durable disease modification remains limited, and disease progression continues to occur in most patients. Finally, this review assesses the complex immunological dysfunction and chronic inflammatory dysregulation that characterize Ph-negative MPNs, as well as emerging therapeutic strategies, emphasizing the importance of fully understanding these intricate microenvironmental mechanisms for the identification and development of novel precision treatment targets. Full article

Triple-negative breast cancer (TNBC) is characterized by marked immune microenvironment heterogeneity and variable chemotherapy response, yet the epithelial transcriptional programs governing cytotoxic immune activation remain poorly understood. We performed an exploratory, integrative analysis using single-cell RNA sequencing of 31,962 cells from eight TNBC patients operationally stratified into Good and Bad Prognosis groups based on pathological lymphoid infiltration, a discovery grouping subsequently validated against pathological complete response (pCR) in three independent bulk RNA-seq cohorts. This analysis identified four epithelial transcriptional states. The G5 DNA damage subpopulation—predominantly restricted to Good Prognosis tumors (29.2% vs. 0%)—and the G4 Metabolism subpopulation—2.4-fold enriched in Bad Prognosis—were the primary prognostic signatures. Machine learning validation using nested leave-one-cohort-out (LOCO) cross-validation across 614 samples demonstrated that G4 + G5 raw genes with random forest yielded the largest observed mean AUC of 0.653, though these results are exploratory and do not establish a validated clinical classifier. CellChat ligand–receptor interaction analysis revealed that G5 DNA-damage epithelial cells are the dominant immune activators in Good Prognosis TNBC, predominantly engaging CD8 cytotoxic T cells through MHC-I antigen presentation via HLA-A/B/C/E/F → CD8A/CD8B interactions, the highest-probability signaling pathway identified. Spatial transcriptomics independently validated significantly higher DNA damage and CD8 T-cell scores in Good Prognosis tissue. Together, these exploratory findings suggest a framework in which tumor-intrinsic DNA damage signaling is associated with MHC-I antigen presentation upregulation and CD8 cytotoxic T-cell engagement, supporting further investigation of this axis and its potential implications for combining DNA-damaging chemotherapy with immune checkpoint blockade in TNBC. Full article

Automatic modulation recognition (AMR) identifies the modulation scheme of received radio frequency (RF) signals under unknown channel conditions and underpins spectrum monitoring and signal demodulation in wireless systems. Under low signal-to-noise ratio (SNR), multipath fading, and limited observation length, however, the discriminative features of modulated signals are severely attenuated, degrading recognition robustness. We propose MSTFFNet, a multi-scale time-frequency fusion network that addresses these challenges with two designs. First, it fuses the raw in-phase/quadrature (I/Q) signal with its short-time Fourier transform (STFT) time-frequency map at the token level through dual-stream heterogeneous encoding, capturing complementary temporal and spectral features. Second, rather than relying on external SNR ground truth, the network self-estimates an SNR-bin probability from the I/Q features and generates a channel-quality embedding that conditions the classifier, requiring no SNR label at inference. On the RadioML2016.10a and 10b benchmark datasets, MSTFFNet achieves overall accuracies of 67.33% and 70.87%, outperforming state-of-the-art methods by 3.53% and 5.33%, with improvements of 5.91% and 9.52% in the low-SNR regime. These results demonstrate improved recognition performance across the SNR conditions represented in the two synthetic RadioML2016 benchmarks, particularly at low SNR. Full article

Non-intrusive load monitoring (NILM) provides a cost-effective way to obtain appliance-level electricity information from aggregate smart-meter measurements and is therefore important for energy management, demand-side response, and sustainable operation in smart buildings. However, accurate appliance-level power disaggregation remains challenging because residential load signals usually involve overlapping appliance signatures, sparse activations, heterogeneous temporal patterns, and transient switching events. To address these challenges, this paper proposes a State-Event-Guided Multi-Scale Gated Network (SEMG-Net) for NILM. The proposed framework integrates a residual temporal encoder, multi-scale dilated convolutional blocks, and a state-event-guided gating mechanism within a unified multi-task learning architecture. The shared encoder extracts hierarchical temporal representations from aggregate mains windows, while task-specific branches jointly estimate appliance power, on/off state, and switching event type. The predicted state probability, three-class event probability distribution, and shared temporal representation are jointly used to construct a continuous gate that modulates the raw power estimate, thereby directly incorporating behavioral predictions into final power estimation. Experimental results on public datasets show that SEMG-Net achieves competitive overall performance, with clear advantages in power estimation, energy consistency, and state identification, particularly for appliances with complex operating stages or transient switching behavior. The ablation results further demonstrate the benefits of multi-scale feature extraction and auxiliary supervision, as well as the effectiveness of the proposed state-event-guided power modulation mechanism. Full article

Hybrid artificial intelligence optimization models combine learned behavioral signals with constrained decision-making, but their transition from prototype calibration to operational use is rarely evaluated across changing platform regimes. The Adaptive Calibration and Operational Transferability (ACOT) framework separates the parameterized decision utility from a fixed external evaluation instrument, validates calibrated configurations on unseen stochastic scenarios, and audits directional transfer regret. The framework is evaluated on a de-identified pilot dataset from a digital group-buying platform comprising 150 users, 200 products, 150 lots, 4000 behavioral events, and 500 orders. Because the observed lot records predominantly represent completed or expired states, the assignment experiments use counterfactually reconstructed pre-activation lot states rather than a complete historical replay. A corrected objective-aware heuristic restores identifiability of the relevance–completion parameter. Across 20 matched optimizer seeds, Tree-structured Parzen Estimation achieved mean validation $J=0.734518$, compared with $0.733899$ for random search and 0.730085 for expert weights, and won 15 of 20 seed-paired comparisons. The run-level BCa 95% confidence interval for the mean TPE–random difference was $[-0.000001, 0.001200]$, indicating a modest, seed-sensitive advantage. Continuous stress testing showed a non-monotonic calibration value, with the largest sampled gain occurring at an available-user fraction of 0.90. Tight-budget calibration had the lowest point-estimate worst-case transfer regret $\left(0.000720\right)$. However, hierarchical bootstrap assigned it a 55.1% probability of being the minimax source, compared with 40.1% for the base regime. The results support uncertainty-aware transfer auditing rather than assuming a universally robust calibration regime. Full article

Endometriosis is a chronic, oestrogen-responsive inflammatory disease characterised by endometrial-like tissue outside the uterine cavity. Because retrograde menstruation is common, lesion establishment probably requires cellular competence and a permissive ectopic microenvironment. This hypothesis-driven narrative review evaluates fascin (FSCN1) as a candidate cytoskeletal effector and considers antecedent eutopic priming versus induction after ectopic adhesion. Functional evidence was integrated with a targeted public-data screen. Donor-level reanalysis of GSE179640 found no conclusive overall eutopic case–control difference and predominantly non-epithelial expression. Exploratory analysis of GSE203191 suggested higher FSCN1 expression within a HSPA6+ stromal subcluster in diagnosed cases, without a comparable epithelial signal or detectable increase in subcluster abundance. This small post hoc analysis remains hypothesis-generating. FSCN1 was absent from the published HECA stromal/macrophage differential-expression lists and was not prioritised by the 2023 endometriosis GWAS. The current evidence therefore argues against uniform epithelial or whole-eutopic overexpression but permits a lineage-restricted stromal state. Fascin participates in autophagy- and miR-145-sensitive invasion networks, although these pathways are pleiotropic. Validation requires cycle- and lineage-resolved tissue mapping, compositional controls, matched lesions, and direct FSCN1 perturbation. Fascin should currently be regarded as a candidate multi-marker component and preclinical target, not a validated biomarker or systemic therapeutic target. Full article

Background/Objectives: Immune exclusion in head and neck squamous cell carcinoma (HNSCC) limits immunotherapy efficacy, yet the molecular determinants of stromal-immune organization remain incompletely characterized. The desmosomal cadherin DSG2 is highly expressed in squamous epithelium; its role in shaping the tumor microenvironment (TME) is unknown. Methods: We integrated bulk RNA-seq from 836 HNSCC patients (TCGA-HNSC n = 566, GSE65858 n = 270), single-cell RNA-seq (GSE139324, n = 26 patients, 133,308 cells), spatial transcriptomics (GSE208253, n = 12), proteomics (CPTAC-HNSCC, n = 108), and external validation cohorts (GSE41613, n = 97). CellChat ligand-receptor analysis, mediation analysis, Mendelian randomization (MR), LASSO-penalized Cox regression, HPV-stratified sensitivity analysis, and transcription factor (TF) correlation analysis were employed. Results: DSG2 exhibited epithelial-specific expression and showed consistent positive correlation with CXCL8 (IL-8; TCGA ρ = 0.228, p = 4.4 × 10⁻⁸) and myCAF activation across independent cohorts. Single-cell analysis revealed that 99.5% of CXCL8-producing cells have zero DSG2 expression, establishing the bulk correlation as compositional rather than cell-intrinsic. CellChat identified CXCL8-CXCR2 as the strongest tumor-stroma interaction in DSG2-high regions (probability = 0.821, 1.80-fold enrichment). Mediation analysis demonstrated 43.6% (95% CI [34.3–53.6%]) of DSG2’s tissue-level association with myCAF activation is mediated through CXCL8 (compositional mediation). Multi-instrument MR (IVW: Beta = −0.028, p = 0.028; I² = 0.0%) corroborated the compositional model. Protein-level validation in CPTAC-HNSCC confirmed DSG2-CD8A inverse correlation (Spearman ρ = −0.35, p = 2.2 × 10⁻⁴). Pan-squamous meta-analysis confirmed negative DSG2-cytolytic activity correlations (pooled ρ = −0.213, 95% CI [−0.296, −0.128], I² = 58.6%, 4 cohorts). DSG2 correlated with TIDE score (ρ = 0.176) and TGF-β exclusion subscore (ρ = 0.428). DepMap analysis identified CXCR2 inhibitor collateral sensitivity (ρ = −0.408, p < 0.0001). An eight-gene co-expression module was validated in two independent cohorts (GSE41613: HR = 3.09, p = 0.003; GSE65858: HR = 1.57, p = 0.032). Conclusions: DSG2 marks a stromal-immune organizational state characterized by CXCL8-CXCR2 paracrine signaling, myCAF activation, and immune exclusion, conserved across squamous malignancies. DSG2-high/PD-L1-high tumors (30.4% prevalence) exhibit the worst predicted ICI response and represent a candidate population for biomarker-selected CXCR2 inhibitor trials in combination with anti-PD-1 therapy. Full article

Motivated by the increasing security requirements of next-generation satellite-terrestrial communication systems and the emergence of Open Radio Access Network (O-RAN) architectures, this paper presents a secrecy analysis of a novel reconfigurable intelligent surface (RIS)-assisted mixed free-space optical (FSO) and radio frequency (RF) satellite downlink transmission system within an O-RAN-enabled non-terrestrial network (NTN) framework. The inherent broadcast nature of RF transmissions presents significant eavesdropping risks, which serves as the primary impetus for this study. We analyze the combined effects of imperfect channel state information (CSI) and random link blockage within such integrated networks. The impact of discrete phase shift constraints at the RIS is also investigated. Closed-form expressions are derived for three key performance metrics: connection outage probability (COP), secrecy outage probability (SOP), and the probability of positive secrecy capacity (PPSC). Through high signal-to-noise ratio (SNR) asymptotic analysis, corresponding asymptotic expressions are obtained, and all analytical results are validated via extensive Monte Carlo simulations. Our findings demonstrate that: (i) Link blockage probability and channel estimation accuracy jointly govern the secrecy performance floor. (ii) Increasing the number of RIS elements enhances physical-layer security by driving both the COP and SOP toward their theoretical lower bounds. (iii) Improving channel estimation accuracy diminishes the eavesdropper’s channel advantage and improves the overall system security. These results offer valuable insights for designing secure mixed FSO/RF satellite-terrestrial systems within O-RAN-enabled NTN architectures that effectively balance connectivity and confidentiality. Full article

Guidelines in early breast cancer place adjuvant chemotherapy before radiotherapy, permit radiotherapy concurrently with endocrine and anti-HER2 agents, but discourage it before or during cytotoxic chemotherapy. This narrative review asks whether that convention remains defensible. Its historical basis, a single randomised trial whose distant-metastasis signal did not survive long-term follow-up, is weaker than is commonly assumed. Its contemporary basis is stronger and is stated here explicitly: an asymmetry of absolute benefit, in which a two-point gain in locoregional control yields roughly half a percentage point of mortality benefit once the EBCTCG four-to-one relationship is applied, whereas degradation of systemic therapy reaches mortality undiscounted. We specify three conditions under which a sequencing change could be justified and note that the absolute loss from a short chemotherapy delay cannot presently be quantified from randomised data. Meanwhile, chemotherapy has lengthened while radiotherapy has contracted to a one- to three-week whole-breast schedule, potentially extended when a sequential tumour-bed boost is required. The retrospective evidence is limited: one cohort offers a hypothesis-generating locoregional signal qualified by an unexpectedly high control-arm event rate and an unplanned subgroup analysis, while another supports only short-term feasibility and tolerability; neither demonstrates benefit in distant control or survival. We further argue that the population in which the question remains clinically live is narrow and probably contracting, since genomic de-escalation withdraws from chemotherapy the phenotypes with the most favourable arithmetic. Adjuvant sequencing is best regarded as an open, testable question within a defined population rather than a general case for change. Full article

In signalized urban road networks, periodic signal phase switching is a key factor influencing traffic-flow stability and operational efficiency. With the rapid development of Connected and Automated Vehicle (CAV) technologies, exploiting their enhanced perception, communication, and cooperative control capabilities has become an important research topic. To characterize the acceleration, deceleration, queueing, and discharge disturbances induced by signal phase transitions, this study proposes a Signal-Constrained Optimal Velocity Model (SC-OVM). By introducing a continuous signal decision function, the proposed model dynamically couples traffic signal states with vehicle-following behavior, including preceding-vehicle following and stop-line tracking within a unified optimal-velocity framework. Furthermore, linear stability analysis, boundary critical condition analysis, and disturbance probability modeling are integrated to reveal the instability mechanism caused by abrupt signal phase transitions, with extensions to stochastic prediction errors and adaptive Signal Phase and Timing (SPaT) inputs. Numerical simulations show that SC-OVM-controlled CAVs can smooth vehicle trajectories, reduce average delay, improve end-of-green passing performance, and achieve a balanced performance in efficiency, stability, and safety compared with the Full Velocity Difference Model (FVDM), Virtual Leading Vehicle model (VLV), and Intelligent Driver Model (IDM). The findings provide theoretical support and practical insights for stability modeling and cooperative control of mixed-traffic flow at signalized intersections. Full article

Astrobiological observations usually return atmospheric, mineralogical, molecular, isotopic, or textural states rather than organisms. The interval between environmental habitability and confirmed life detection is therefore an evidential problem. This article develops planetary geochemical metabolism (PGM) as a scale-explicit description of abiotic water–rock–fluid reactions, including atmospheric or ice-shell boundary conditions where relevant, that sustain redox disequilibria, catalytic mineral interfaces, prebiotic molecular fluxes, and preservable mineral products. Proto-biosignatures are defined as contextual, non-diagnostic signatures of this interval: signals that do not demonstrate life, but increase the plausibility of prebiotic network organization relative to low-organization abiotic chemistry. A nondimensional Geochemical Metabolic Potential, ${\Phi }_{\mathrm{PGM}}$, is formalized as a target-specific screening index describing redox exergy, catalytic-interface density, reaction-network closure, environmental cycling efficacy, and preservation potential. The index is not calibrated as a probability of life. A reproducible Latin-hypercube experiment across $M_{\mathrm{sim}}=5000$ synthetic environments examines internal model behavior rather than ranking planets. Higher ${\Phi }_{\mathrm{PGM}}$ values mark hypotheses to be tested under declared scale, uncertainty, and observability constraints. Applications to early Earth, Noachian Mars, ocean worlds, and terrestrial exoplanets illustrate that habitability, prebiotic organization, biological inference, and preservation are distinct quantities. Full article

Heart failure (HF) remains a major global cause of morbidity and mortality, where early diagnosis is critical for improving patient outcomes. Conventional single-modality approaches often fail to capture the complex and multifactorial nature of HF. This study investigates the feasibility of a late-fusion framework that integrates modality-specific predictions derived independently from cine-MRI, electrocardiographic signals, biomarkers and demographic data for HF prediction. Independent cine-MRI data from 281 patients, ECG recordings from the PTB-XL PhysioNet database and biomarker profiles from 157 patients were retrospectively analyzed as separate modality-specific cohorts. Twenty-five features were extracted and processed. Modality-specific models (Attention U-Net, MLP, XGBoost) were trained separately on pre-extracted features to preserve predictive accuracy while minimizing computational cost. Their outputs were combined through ensemble meta-learning (XGBoost, LightGBM, Random Forest) with sample weighting to handle missing data. The final HF prediction probability was obtained by averaging the outputs across the three meta-learners. The proposed framework achieved competitive diagnostic performance, with 98.00% (95% CI: 94.96–99.45%) accuracy, 97.80% (95% CI: 92.28–99.73%) sensitivity, 98.17% (95% CI: 93.53–99.78%) specificity, an F1-score of 97.80% (95% CI: 93.6–99.8%) and an AUC of 0.978 (95% CI: 0.945–0.996) when evaluated against state-of-the-art methods. The results highlight the potential of late-fusion strategies for integrating independently trained modality-specific predictions, offering a feasible approach for HF risk assessment under heterogeneous data availability. Full article

This paper presents a retrospective proof-of-concept development and preliminary evaluation of an interpretable decision-support system for daily fatigue-risk management in one elite women’s football club. The system integrates morning wellness, GPS-derived external load, and a daily wellness-duration internal-load proxy at the player-day level. It combines a player-day integration layer, an interpretable predictive layer, and a recommendation layer that returns one of three staff-facing actions: Reduce, Maintain, or Progress. The predictive model outputs a calibrated probability of elevated next-morning self-reported soreness. The target is a subjective questionnaire outcome, not an injury, medical diagnosis, or objective marker of recovery. The decision-support layer maps this probability to a three-state recommendation, informed by a review threshold, operational guardrails, and staff oversight. Using retrospective monitoring data from two competitive seasons (2024/25 and 2025/26) in a single professional team, we evaluated the proposed approach using rolling-origin temporal validation, leave-one-player-out cross-validation, and between-season validation. To separate genuine predictive signal from the day-to-day persistence of soreness, we report a baseline ladder ranging from a trivial persistence rule to the full model, with bootstrap confidence intervals for performance differences. Under rolling-origin validation across 16 monthly folds, the final logistic regression model achieved a mean ROC-AUC of $0.759$ ($SD=0.089$). Critically, a model excluding current soreness still outperformed the persistence baseline (ROC-AUC $0.738$ vs. $0.721$), and the isolated contribution of current soreness was modest but reliable ($\Delta$ROC-AUC $=+0.044$, 95% CI $[+0.027,+0.059]$). Between-season validation (train: 2024/25; test: 2025/26) yielded an ROC-AUC of 0.801. The three-state recommendation layer separated outcomes monotonically, with observed next-morning soreness rates of $0.054$ for Progress, $0.217$ for Maintain, and $0.326$ for Reduce ($p<0.001$ for the Progress-versus-Maintain contrast). These preliminary findings support the feasibility of the proposed approach within the club studied. However, because the model requires complete wellness, GPS, and proxy data, it operates only on the fully monitored on-pitch stratum (3386 of 17,703 player-days); the reported performance therefore applies to this stratum rather than to a typical player-day, and prospective evaluation and external validation by independent teams are required before broader implementation can be considered. Full article

Industrial process safety systems are predominantly reactive: alarms activate after limits are crossed, faults are diagnosed after deviations develop, and HAZOP knowledge remains offline during operation. This paper proposes Predictive Operational Safety Engineering (POSE) as an emerging research paradigm in which operational safety is treated as a continuously forecastable state rather than a post-event classification, shifting the operational question from what has gone wrong? to how much safe operating time remains, and which intervention is most urgent? Four integrated predictive safety metrics anchor the framework: Remaining Safety Margin (RSM), quantifying the normalized distance between the predicted process trajectory and the nearest safety boundary; Remaining Safe Operating Time (RSOT), estimating when that boundary will be crossed under the current trajectory; the Operational Vulnerability Index (OVI), combining margin depletion rate, safeguard availability, and consequence severity into a single intervention-urgency signal; and Predictive Safety Confidence (PSC), the probability that a specific named operator intervention can be executed to completion before the predicted safety boundary is crossed, coupling prediction uncertainty with action execution time. The Predictive Operational Safety Twin (POST) is proposed as a three-layer reference architecture implementing POSE through predictive process intelligence, predictive safety intelligence, and human safety intelligence. The paper synthesizes six research streams, positions POSE against seven adjacent disciplines, states ten guiding principles, and formulates a research agenda. As a conceptual narrative review, the paper does not claim empirical validation of POSE. Instead, it establishes the foundational vocabulary, reference architecture, and research agenda required to advance predictive operational safety from an emerging concept toward benchmarked and industrially validated practice. This article constitutes the conceptual and evidence-synthesis phase of a staged research program; subsequent work must test the proposed constructs through benchmark simulation, uncertainty calibration, baseline comparison, operator studies, and industrial case studies. Full article