Accurate classification of peripheral blood cell types from microscopy images can support automated hematology workflows by reducing manual review burden and improving consistency. However, fine-grained recognition remains challenging due to strong inter-class similarity, intra-class morphological variability, and confounding cytoplasmic texture patterns that blur decision boundaries among related cell types. We propose SwinDiNO-AFNet for eight-class classification of basophil, eosinophil, erythroblast, immature granulocyte (IG), lymphocyte, monocyte, neutrophil, and platelet. SwinDiNO-AFNet is a dual-path hybrid that fuses hierarchical window-based features from a Swin Transformer with global representations from a pretrained DINOv2 vision transformer used strictly as a feature extractor. The flattened Swin and pooled DINOv2 features are concatenated and refined through a channel attention module to obtain attention-weighted fusion before the final classifier. To ensure rigorous evaluation, the training pipeline enforces leakage control via perceptual-hash based near-duplicate removal before fold creation, with consistent RGB enforcement, 224 × 224 resizing, ImageNet normalization, and train-only augmentation and robustness noise applied exclusively within training folds. Under leakage-safe stratified fivefold cross-validation, SwinDiNO-AFNet achieves 98.61 ± 0.21% validation accuracy with 98.46% macro-F1, and statistical testing (Welch with Holm adjustment) supports improvements over multiple baseline backbones under the same protocol. Precision-recall analysis further indicates near-ceiling separability (micro-AUPRC 0.9979), while confusion matrix and Grad-CAM based error analysis show that remaining failures concentrate in morphologically ambiguous cases, including IG confusions toward granulocytic patterns. Overall, the results show that attention-guided fusion of complementary Swin and DINOv2 representations, combined with leakage-safe cross-validation and XAI-backed diagnostics, provides a strong solution for eight-class peripheral blood cell classification on the studied dataset.

Sepsis is responsible for approximately 270,000 deaths annually in the United States. Conventional scoring systems, such as SOFA and qSOFA, are largely reactive and do not effectively leverage longitudinal ICU data for early prediction. This study aims to develop a deep learning framework capable of predicting sepsis onset up to 6 h before Sepsis-3 criteria are met while also providing clinically interpretable temporal explanations. The PhysioNet/CinC 2019 Challenge dataset, comprising 1,552,210 patient hours from 40,336 ICU patients, was utilized. A temporal transformer encoder (TTE) was trained using 12-h look-back windows with 92 engineered features. Severe class imbalance (2.6% positive rate) was addressed through weighted random sampling and focal loss. Fivefold patient-level cross-validation was employed to prevent temporal leakage. Platt scaling was applied for probability calibration. Grad-CAM was adapted for temporal explainability, while SHAP was used for feature-level attribution. BiLSTM-Attention and XGBoost models served as baseline comparators. The TTE model achieved a cross-validated AUROC of $0.8320\pm 0.0032$ and an AUPRC of $0.1505\pm 0.0148$, significantly outperforming BiLSTM-Attention (AUROC: 0.7859) and XGBoost (AUROC: 0.7731; DeLong $p<0.0001$). Platt scaling reduced the expected calibration error from 0.3154 to 0.0017. The median alert lead time was 46.5 h (IQR: 21–84 h), with 95.3% of septic patients receiving alerts at least 3 h before onset. Grad-CAM analysis identified time steps $t-10$ and $t-9$ as the most predictive. However, high-severity patients (SOFA proxy $\ge 3$) demonstrated substantially reduced performance (AUROC: 0.257). The proposed TTE framework demonstrated strong and well-calibrated early sepsis prediction with substantial clinical lead time. The concentration of predictive signals 10–11 h prior to alert generation supports the feasibility of continuous automated ICU monitoring from admission onward. Reduced performance in high-severity patients highlights the need for severity-stratified modelling in future research.

Background/Objectives: Sepsis is responsible for approximately 270,000 deaths annually in the United States. Conventional scoring systems, such as SOFA and qSOFA, are largely reactive and do not effectively leverage longitudinal ICU data for early prediction. This study aimed to develop a deep learning framework capable of predicting sepsis onset up to 6 hours before Sepsis-3 criteria are met, while also providing clinically interpretable temporal explanations. Methods: The PhysioNet/CinC 2019 Challenge dataset, comprising 1,552,210 patient-hours from 40,336 ICU patients, was utilized. A Temporal Transformer Encoder (TTE) was trained using 12-hour look-back windows with 92 engineered features. Severe class imbalance (2.6% positive rate) was addressed through weighted random sampling and focal loss. Five-fold patient-level cross-validation was employed to prevent temporal leakage. Platt scaling was applied for probability calibration. Grad-CAM was adapted for temporal explainability, while SHAP was used for feature-level attribution. BiLSTM-Attention and XGBoost models served as baseline comparators. Results: The TTE model achieved a cross-validated AUROC of 0.8320±0.0032 and an AUPRC of 0.1505±0.0148, significantly outperforming BiLSTM Attention (AUROC: 0.7859) and XGBoost (AUROC: 0.7731; DeLong p < 0.0001). Platt scaling reduced the Expected Calibration Error from 0.3154 to 0.0017. The median alert lead time was 46.5 hours (IQR: 21–84 h), with 95.3% of septic patients receiving alerts at least 3 hours before onset. Grad-CAM analysis identified timesteps t − 10 and t −9 as the most predictive. However, high-severity patients (SOFA proxy ≥ 3)demonstrated substantially reduced performance (AUROC: 0.257). Conclusions: The proposed TTE framework demonstrates strong and well-calibrated early sepsis prediction with substantial clinical lead time. The concentration of predictive signals 10–11 hours prior to alert generation supports the feasibility of continuous automated ICU monitoring from admission onward. Reduced performance in high-severity patients highlights the need for severity-stratified modelling in future research.

Diabetic foot ulcers (DFUs) are a major cause of lower-extremity amputation, and image-based decision support may assist screening and preliminary triage where specialist access is limited. We propose DFU-MambaKAN, a hybrid architecture combining a pure-PyTorch selective state-space (Mamba) block with a Gaussian radial-basis-function Kolmogorov–Arnold Network (RBF-KAN) feed-forward layer. The study evaluates two distinct tasks: (1) binary normal-versus-ulcer screening and (2) a dataset-specific four-class Wagner-Meggitt severity-grading task. DFU-MambaKAN contains 1.062 million parameters and was compared with ResNet50, EfficientNet-B0, MobileNetV3-Small, and ViT-Tiny under the manuscript’s common downstream training protocol. The reported values are point estimates from one deterministic 70/15/15 split generated with seed 42; no confidence intervals, repeated-seed averages, or inferential significance tests were obtained. On binary screening, DFU-MambaKAN achieved 97.17% accuracy, macro-F1 0.970, and AUC 0.994. On four-class grading, it achieved 67.75% accuracy, macro-F1 0.677, and AUC 0.895, whereas the baselines achieved 98.14–99.00% accuracy. This large gap means that the present evidence does not establish competitiveness for multiclass severity grading. Possible contributors include training duration and convergence, dataset size and class definitions, label or image-quality uncertainty, hyperparameter selection, global token mixing, and architecture–task mismatch; extended learning-curve analysis and multi-seed evaluation are needed to test these explanations. An automated technical audit identified 342 exact duplicates among 1055 class-folder images in Dataset A (32.4%); no deduplicated-versus-nondeduplicated performance comparison was performed, so this finding is reported as a methodological caution rather than proof of accuracy inflation. The model’s 1.062M parameter count supports parameter-efficient storage, but its measured single-image GPU latency was 90.54–93.14 ms, substantially higher than the baselines. Accordingly, the current evidence supports further investigation for batched or queued screening more strongly than immediate real-time mobile deployment. External clinical validation, multi-seed analysis, statistical testing, convergence studies, and component ablations remain necessary.

Atrial fibrillation (AF) affects more than 37 million people worldwide and substantially raises the risk of stroke, heart failure, and premature death. Wearable single-lead ECG devices offer a natural platform for continuous AF surveillance, yet automated detection from short, noisy recordings remains an unsolved clinical engineering problem. This paper presents a sys tematic benchmark of three deep learning architectures—a one dimensional ResNet-34 (1D-ResNet34), a Temporal Convolutional Network (TCN), and a novel State-Space Model (MambaECG)— evaluated under identical training protocols on two indepen dent public datasets: the PhysioNet Computing in Cardiology Challenge 2017 corpus (8,528 recordings) and the Chapman Shaoxing 12-lead ECG database (45,152 recordings, Lead I used). All models were trained exclusively on single-lead PhysioNet recordings and tested on Chapman–Shaoxing Lead I without fine-tuning, constituting the first cross-institutional generalisation study for this architectural family. The TCN achieves the highest AUROC of 0.957 with only 0.19M parameters, outperforming 1D-ResNet34 (AUROC 0.955, 35.34M parameters) and MambaECG (AUROC 0.928, 0.65M parameters). Notably, all three architectures demonstrate positive zero-shot transfer to Chapman–Shaoxing, with AUROCs of 0.961, 0.963, and 0.948 respectively—without any target-domain exposure during training. MambaECG achieves the fastest in ference at 0.04 ms per recording, enabling real-time deployment on resource-constrained wearable hardware. Gradient-weighted Class Activation Mapping (Grad-CAM) analysis confirms that MambaECG correctly localises clinically meaningful ECG fea tures: QRS complexes and P-waves for normal rhythm, irregular RR intervals and fibrillatory baseline for AF, and premature ventricular complex morphology for Other rhythm. These results provide evidence-based architectural guidance for wearable AF screening systems and demonstrate that lightweight temporal models match or exceed deep residual networks at a fraction of the computational cost.

Background: Cloud computing, remote workforces, and Internet of Things (IoT) deployments have eroded the enterprise network perimeter, exposing the limits of security models that grant implicit trust to internal actors. Zero Trust Architecture (ZTA), governed by the principle of "Never Trust, Always Verify," addresses this through continuous verification, least-privilege access, micro-segmentation, and adaptive monitoring. Objectives: This paper synthesizes reported evidence on ZTA across three research questions: effectiveness relative to perimeter models, the implementation challenges enterprises face, and the security contribution of individual components. Methods: A structured literature review was conducted on a predefined corpus of 37 ZTA-related studies. A rubric-based composite model incorporating relevance, methodology quality, recency, and measurability of outcomes characterized each study. All 37 met the appraisal threshold of S >= 3.0, the lowest composite score being 3.55, so the rubric characterized the corpus rather than functioning as an exclusion filter. Results: Reported findings suggest that ZTA can improve breach detection, reduce unauthorized access, and limit attack propagation in specific empirical and simulation contexts. Headline values, including 75% improvement in breach detection, 66.7% reduction in unauthorized access, up to 80% attack-surface reduction, and 100% repulsion of simulated malicious packet injections, are drawn from a limited number of studies and need replication. Implementation barriers include performance overhead in IoT and edge environments, high transition costs, legacy migration complexity, interoperability gaps, and cultural resistance. Conclusions: ZTA is a promising and structurally well-aligned security framework for modern distributed environments, but the strength of evidence varies by component and deployment context.