Type 1 Diabetes (T1D) is a chronic metabolic disease characterized by elevated blood glucose (BG) concentrations, resulting from the immune-mediated destruction of insulin-producing $\beta$-cells in the pancreas. Effective management of T1D greatly benefits from constant monitoring of BG levels, achievable in real-time using minimally invasive continuous glucose monitoring (CGM) devices. These devices provide data streams that can be leveraged by forecasting algorithms to predict BG levels minutes in advance, enabling timely therapeutic interventions to prevent adverse events, such as hypo/hyperglycemia. With the increasing availability of data, deep learning (DL) algorithms have emerged as the state-of-the-art for BG forecasting, owing to their ability to autonomously learn complex nonlinear relationships, such as those underlying the glucoregulatory system. Despite a growing body of research, a comprehensive review specifically focusing on DL applications for BG prediction is still lacking. To address this gap, a systematic review was conducted following the PRISMA guidelines, involving extensive searches across PubMed, Scopus, and Web of Science databases. A total of 26 studies satisfied the inclusion criteria and were evaluated based on dataset characteristics, model inputs, training paradigm, prediction horizon, model architecture, evaluation metrics, performance, and baseline comparators. While DL models show great promise, several challenges persist—particularly in ensuring physiological fidelity and interpretability, both essential for clinical adoption. To overcome these barriers, future research should prioritize the integration of explainable AI (XAI) techniques to improve model reliability and safety, ultimately supporting the effective deployment of DL models in real-time T1D management.

This study develops an efficient and accurate method for segmenting complex swallowing events in patients with dysphagia. We propose CFSCNet, a novel model that integrates multimodal signals—audio, nasal airflow, and triaxial accelerometry—to enhance swallowing event segmentation. The model employs a dual-stream architecture to capture swallowing information across varying durations and utilizes an exponential moving average-based cost-sensitive weighting method to address data imbalance. To further improve the stability of predictions for clinical applications, a Swallowing State Machine-Driven Post-Processing Method is introduced to smooth the segmentation sequences. CFSCNet achieves state-of-the-art performance on two benchmark datasets, reaching an AUC of 89.83 on the HRCA dataset and 94.10 on the SMSD dataset, significantly outperforming existing methods. Built on a series of tailored methodological innovations, the proposed framework offers a comprehensive and novel solution for swallowing event segmentation. It demonstrates strong accuracy and reliability, as well as great potential for future extension and clinical translation. This work is expected to advance high-precision detection and diagnosis of dysphagia, thereby supporting more effective swallowing rehabilitation and medical interventions.

Automatic pancreas segmentation can facilitate diagnosis and treatment of pancreatic diseases. The combination of non-contrast, arterial, and venous phases of CT imaging can enhance differentiation of the pancreas from its surrounding structures. However, existing multimodal methods, which try to integrate the multimodal information in computer-aided pancreas segmentation, often overlook the inter-modal relationships and have a limited capability for information fusion. In this paper, we propose a multi-phase pancreas segmentation method for incorporating Feature Aggregation Module (FAM) and Modality Adaptive Transformer (MAT). Specifically, we use the venous phase as the primary modality, while the non-contrast and arterial phases serve as supplementary modalities, based on clinical prior knowledge. Our FAM integrates spatial information from the primary and supplementary modalities, while our MAT adaptively enhances feature representation and establishes long-range dependencies among modalities. Our method outperforms state-of-the-art techniques on a large scale dataset. Based on the segmented pancreas region, We further perform a downstream task focused on pancreatic volume calculation. The prediction accuracy is on par with manual segmentation, demonstrating effectiveness and potential application of our proposed method.

Medical image analysis remains challenging due to inherent limitations in imaging modalities, where structural aliasing and noise artifacts persistently compromise diagnostic accuracy. While convolutional neural networks (CNNs) and vision transformers (ViTs) have achieved remarkable progress in feature extraction, their inherent sampling mechanisms and spectral biases often exacerbate these high-frequency distortions, leading to suboptimal lesion characterization. To address this critical limitation, we propose Spec-ViT, a novel wavelet-based anti-aliasing Transformer architecture that synergistically integrates adaptive spectral purification with hierarchical attentive learning. The Wavelet Antialiasing Module (WAM) first implements learnable smoothing factor in the wavelet domain to suppress highfrequency artifacts, while preserving clinically relevant lowfrequency structures and fine diagnostic details. Building upon this spectral foundation, the Lightweight Enhanced Attention (LEA) refines feature representations through a dual-path mechanism, coupling channel-spatial attention with global multi-head self-attention to enhance lesion context modeling. Finally, the Smoothed Convolutional Gate (SCG) further sharpens local discriminability through depthwise convolution and adaptive Swish gating, completing a coherent pipeline from frequency-aware purification to global-local attentive analysis. Extensive experiments on five benchmark medical image classification datasets demonstrate that Spec-ViT consistently outperforms both baseline and state-of-the-art methods, achieving up to 84.04% accuracy on the Pediatric Pneumonia Chest X-rays dataset in particular.

Sleepiness at the wheel is an important factor contributing to road traffic accidents. Based on the characteristic changes in Electroencephalography (EEG) and Electrooculography (EOG) signals, a dozing state is refined into three sub-states: the onset, duration, and end state. Each state is characterized by different physiological indicators such as the EEG alpha waves, the rising edge, and falling edge waveforms in EOG signals. To enable real-time detection of these physiological indicators, we propose a framework integrating three Convolutional Neural Network–Long Short-Term Memory–Self-Attention (CLSA) models, which combine CNN-based local feature extraction with self-attention mechanism for global context capture. The framework is evaluated for performance on continuous test data from 12 subjects. Our results demonstrate that by detecting alpha waves and the rising edge waveform, the alpha wave epoch (AWE) at the onset of the dozing state can be identified with high accuracy and precision. Thus, the onset sub-state is calculated as the period from the start time of the rising edge waveform to the time when the AWE is valid. Subsequently, the duration sub-state corresponds to the sustained presence of alpha waves. Furthermore, the falling edge waveform is detected with high accuracy, enabling the classification of the end state into two distinct phenomena: alpha blocking phenomenon or alpha wave attenuation-disappearance phenomenon, representing the sleepiness level—relaxed wakefulness or sleep onset, respectively. Utilizing three-channel signal processing, this framework provides a promising approach for real-time sleepiness detection in real-world driving scenarios.

Steady-state visual evoked potential-based brain-computer interfaces (SSVEP-BCIs) hold significant promise for enabling high-speed human-computer interaction in real-world scenarios. However, existing frequency-domain decoding methods treat frequency spectrum features (the real and imaginary spectrum features) as a single feature without considering their unique spatial and spectral characteristics, resulting in insufficient generalizable features and limited classification accuracy in cross-subject scenarios. To address this issue, we propose a Dual-Branch Attention-Based Frequency Domain Network (DB-AFDNet) to independently decode real and imaginary spectral components, aiming to acquire more discriminative and generalizable features for cross-subject applications. Specifically, we construct inter-branch attention similarity constraints to encourage the two branches to have similar attention properties, promoting to learn the consensus characteristics in the dual branches. Furthermore, we propose intra-branch orthogonality constraints to explore branch-specific discriminative features to learn generalizable features. Experimental studies on two public datasets, the Benchmark and Beta datasets, demonstrate that DB-AFDNet outperforms state-of-the-art methods in cross-subject classification, achieving a relative improvement of 1.36$\%$ and 1.45$\%$, respectively.

Cellular responses to genetic perturbations are prevalent in wide contexts from the fundamental understandings on pathology to the development of clinical therapies and the discovery of novel drug targets. Nonetheless, the substantial amount of possible perturbation combinations renders wet-lab experiments prohibitively expensive and time-consuming. To address it, the BRNET model is proposed for predicting non-linear transcriptional outcomes where multiple perturbations exist. BRNET integrates prior knowledge with advanced embeddings into a non-stacked neural structure to predict transcriptional responses to both individual and multiple genetic perturbations. For unseen scenarios, BRNET also generalizes well under the corresponding perturbations. Experimental results highlight the capabilities of BRNET, demonstrating promising performance as compared to established deep learning models.

Medical image saliency prediction plays a pivotal role in emulating clinician visual attention to prioritize diagnostically critical regions. Current methods remain constrained by their spatial-domain dependency and limited cross-modality generalizability, neglecting frequency-domain patterns critical for subtle pathology detection while suffering from over-specialization in specific imaging modalities. Therefore, we propose a dual-domain visual prompt network (DVPNet) that integrates cross-modality generalization with spectral pattern awareness. On the one hand, DVPNet establishes a dataset prompt branch that dynamically modulates spatial feature encoding through modality-specific priors, allowing adaptive interpretation of heterogeneous medical imaging domains. On the other hand, a spatial-frequency hybrid prompt module employs learnable wavelet filters to decompose images into multi-scale spectral components, preserving low-frequency anatomical context while enhancing discriminative high-frequency biomarkers that are typically obscured in previous pixel-level analysis. By seamlessly integrating these complementary representations, DVPNet optimally synthesizes spatial and spectral evidence, enabling robust generalization across diverse medical imaging modalities while sustaining computational efficiency. Extensive experimental results on two distinct datasets demonstrate that the proposed method outperforms state-of-the-art approaches, showing superior saliency prediction performance and enhanced generalizability across medical contexts.

This study introduces a novel electrocardiogram (ECG) arrhythmia classification model, PathFusion-Net, which integrates Rough Path Theory with deep learning technologies. The model combines Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), Path Signatures, and Path Development to extract spatial morphological features from ECG images and multi-order temporal representations from ECG signals. By adopting an inter-patient split paradigm, our approach more closely reflects real-world clinical diagnostic settings compared to intra-patient methods. The model demonstrates state-of-the-art overall classification performance on both the MIT-BIH Arrhythmia Database and a private clinical dataset, achieving 94.7% and 95.1% accuracy, respectively, under the AAMI four-class standard with an inter-patient split paradigm. On the MIT-BIH dataset, the proposed method attains competitive precision and recall across multiple arrhythmia types, including 95.2% /87.9% for ventricular ectopic beats (V) and 75.7% /92.3% for supraventricular ectopic beats (S), indicating balanced performance across clinically diverse categories. This research highlights the potential of Rough Path Theory in time-series analysis and offers a novel deep learning framework for automated early detection and monitoring of ECG arrhythmias. The code used in this study is available at: https://github.com/Rand2AI/PathFusion-Net.

Radar sensing has emerged as a promising solution for the contactless monitoring of Heart Rate Variability (HRV), a crucial indicator of the cardiovascular and autonomic nervous systems. However, due to signal noise and interference that easily obscure heartbeat details, along with variations in heartbeat across different physiological conditions, existing methods remain restricted to laboratory settings with healthy subjects and fail in real-world scenarios involving more complex physiological conditions. In this study, we propose a physiological prior-inspired deep learning framework for robust radar-based HRV monitoring. Specifically, we leverage the prior that internal heartbeats drive movements across the entire torso surface and design a hybrid deep neural network to model the spatio-temporal relationship between full-body radio reflections and heartbeats, effectively mitigating interference. Then, we incorporate the cardiac motion's self-similarity prior to establish a signal augmentation strategy, effectively remodeling the HRV distribution and enhancing performance across diverse physiological conditions. We build and validate our method on a large-scale dataset comprising 7,150 outpatients with complex physiological conditions in real-world scenarios. The experimental results demonstrate that our method achieves a mean IBI error of 19.21 ms, an RMSSD error of 16.23 ms, an SDSD error of 16.70 ms, and a pNN50 error of 7.28%. We further validate the performance by classifying five common cardiac conditions based on HRV results, demonstrating performance comparable to ECG-based methods. These results highlight the great potential of our approach for accurate, contactless HRV monitoring in real-world applications.

Many deep learning methods represented by graph-based approaches achieve significant progress in drug repositioning. However, these graph-based methods face a critical limitation: they often fail in cold-start scenarios because the graph structure relies heavily on known association information from both the drug and disease sides. To address this challenge, we propose a bidirectional behavior learning strategy for drug repositioning, BiBLDR, an innovative framework that reformulates drug repositioning as a behavior sequence learning task. First, we construct bidirectional behavioral sequences based on drug and disease sides. Bidirectional behavior sequences ensure sufficient information for model learning in both drug and disease cold-start scenarios, while providing more precise feature representations for association prediction tasks. Subsequently, we propose a two-stage strategy for drug repositioning. In the first stage, we construct prototype spaces to characterise the representational attributes of drugs and diseases. In the second stage, these refined prototypes and bidirectional behavior sequence data are leveraged to predict potential drug-disease associations. This design allows BiBLDR to more robustly capture hidden pharmacological relationships from bidirectional behavioral sequences, delivering significant benefits in cold-start scenarios. Extensive experiments demonstrate that our method achieves state-of-the-art performance on benchmark datasets. Meanwhile, BiBLDR demonstrates significantly superior performance compared to previous methods in cold-start scenarios.

Inferring gene regulatory networks (GRNs) is critical for understanding the mechanisms that govern cellular behavior. Advances in single-cell RNA sequencing (scRNA-seq) have enabled GRN analysis at single-cell resolution and stimulated the development of many computational methods. However, most existing approaches depend heavily on extensive prior regulatory information, which limits their effectiveness in few-shot settings where such data for the target cell line are scarce or unavailable. To address this challenge, we propose TransGRN, a transfer learning–based method for inferring gene regulatory networks (GRNs) across cell lines. TransGRN adopts a cross-cell-line pre-training strategy that combines scRNA-seq data from multiple source cell lines with biological knowledge obtained from large language models. In addition, it includes a regulatory interaction extraction module that integrates gene expression profiles with semantic information. By transferring generalizable gene–gene regulatory patterns from source to target cell lines, TransGRN achieves state-of-the-art performance in both benchmark tests and few-shot GRN inference tasks.

Accurate sleep staging is vital for evaluating sleep quality and diagnosing sleep disorders. Yet most automated sleep staging methods rely on large datasets labeled by experts. However, clinical annotation is both time-consuming and subjective, making it difficult to obtain sufficient high-quality data for automated sleep staging research. To address this bottleneck, we propose a few-shot, dual-branch contrastive pre-training framework for single-channel electroencephalogram (EEG)–based sleep staging. The framework first conducts fully self-supervised pre-training on unlabeled data, then performs fine-tuning that requires only a small set of labeled samples. We developed and evaluated our solution with the public Sleep-EDF-v2 EEG dataset, achieving state-of-the-art results despite using limited labeled data. Specifically, with only 1% labeled data, our method delivers an accuracy of 76.10% and Macro F1-score of 61.34%, comparable to supervised models trained on 100% labeled data. We further validated our approach on the ISRUC-1 and ISRUC-3 datasets, where similar robust results were consistently observed. The ability to effectively develop sleep classification models using minimal labeled data demonstrates the potential value of our framework across diverse clinical settings.

The combination of positron emission tomography (PET) and computed tomography (CT) can accurately reflect the metabolic and anatomical information of a variety of tumors, including nasopharyngeal carcinoma, lymphoma and lung cancer, which plays an important role in the diagnosis, staging and efficacy evaluation of tumors. Accurate and automatic delineation of target tumors is crucial for radiotherapy, however, the tumor segmentation task is extremely challenging due to the fuzzy tumor boundaries, uncertain locations, and the scattered distribution of multiple tumors throughout the body. To this end, this study extended the STUNet pre-trained on the TotalSegmentator dataset and proposed M-AECA, which integrates a Mamba-based auxiliary encoder (M-AE) to provide multi-scale global features for enhanced feature extraction. In addition, an Inter-Branch Feature Fusion Module (IBFFM) is designed to achieve more comprehensive global and local feature fusion through cross attention (CA) and feature subspace projection. The method was evaluated on Hecktor and AutoPET datasets, demonstrating superior performance compared to other comparison methods. In the test sets of these two datasets, the average Dice similarity coefficients of the proposed method were 70.86% and 64.91%, respectively. In addition, the results of the ablation experiment show that the proposed M-AE and IBFFM demonstrate strong performance and significant advantages.

Chronic Heart Failure (CHF) represents a significant global health concern due to its high morbidity and mortality rates. Effectively addressing this challenge requires scalable technology solutions to shift Heart Failure (HF) care from episodic reactive treatment to continuous personalized management. As digital health technologies advance, integrating Artificial Intelligence (AI) and the Internet of Things (IoT) into CHF care enables the development of scalable monitoring, prevention, and management strategies and real-time Clinical Decision Support Systems (CDSSs). This Systematic Literature Review (SLR) analyzes 67 peer-reviewed studies published between January 2021 and May 2024, selected using Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines to evaluate the technological and clinical impacts of AI-enabled systems in CHF and broader HF care. The review identifies emerging trends, discusses dataset characteristics and clinical relevance, identifies IoT integration patterns, gaps, and deployment barriers, and highlights opportunities for improving the integration of AI/IoT systems into HF care workflows. The studies are organized into four clinical application domains: HF detection, phenotyping and classification, risk stratification, and other miscellaneous applications. Our findings highlight the progress in AI/IoT synergy; however, challenges remain in dataset heterogeneity and coverage, reproducibility, benchmarking practices, and clinical workflow integration, particularly as IoT integration is often limited or insufficiently explored. Our primary recommendations emphasize the use of multimodal datasets, the adoption of interpretable modeling approaches, and stronger interdisciplinary collaboration to improve clinical applicability and support integration into real-world settings.

Chronic pain is a widespread phenomenon affecting over 21% of the United States population. Despite the significant impact of pain on a patient's quality of life, the detection and identification of pain relies on subjective methods such as self-reporting. To address the challenges in identifying and treating chronic pain, a quantifiable biomarker for pain is needed. Here we present novel AI-driven method for the identification and isolation of localized pain signals in the brain during both acute and chronic pain. By using Matching Pursuit (MP) to decompose Local Field Potential (LFP) recordings from the Anterior Cingulate Cortex, the Nucleus Accumbens, and the Prelimbic Cortex, we can learn a latent representation with a conditional variational autoencoder (CVAE) and track changes in latent signal components in response to acute, sub-chronic, and chronic pain after injury. This method allows for both the identification of LFP signal components which are the primary drivers of observed aggregate changes in brain activity during pain, as well as for the tracking of said components over time. The model achieves an average per-feature RMSE of 0.130 on validation data and produces functionally separable latent representations of input MP atoms. The combination of MP for feature extraction and CVAE for latent space development allows for the extraction of both generalized and subject-specific pain motifs involved in chronic pain. These AI-driven biomarkers provide a basis for precision identification and quantitative monitoring of pain over time.

Accurate cancer subtype identification is crucial for personalized medicine, as it enables precise diagnosis based on molecular characteristics. With the advent of large-scale multi-omics data from various resources, researchers now have unprecedented opportunities to explore cancer subtypes comprehensively. However, the inherent complexity, high dimensionality, and heterogeneity of these datasets present significant statistical and computational challenges, often leading to suboptimal clustering performance when inter-omics heterogeneity is overlooked. To address these challenges, we propose a novel method called the Hierarchical Multi-view Graph Diffusion Weighted (HMGDW) model for cancer subtype identification. Our approach begins with the generation of multiple base clusterings through random feature sampling, effectively mitigating the impact of high dimensionality. These base clusterings are subsequently integrated via a late integration strategy to yield the consensus clustering result. Then, we introduce a graph diffusion weighted mechanism that prioritizes views with the most significant contributions to the unified graph representation. Lastly, we conducted extensive experiments on both generic multi-view datasets and multi-omics cancer multi-omics datasets. The experimental results demonstrate that HMGDW consistently outperforms several state-of-the-art methods, achieving robust and accurate clustering. Additionally, a case study on the acute myeloid leukemia (AML) dataset validates the practical efficacy of our model in identifying clinically relevant subtypes.

The integration of multi-omics data is crucial for cancer patient classification and biomarker identification. While this integration presents significant potential, it necessitates the development of sophisticated methodological frameworks. There remains considerable opportunity for enhancement in existing approaches to simultaneously fulfill the demands of omics-specific feature extraction and cross-omics association modeling. Consequently, in this study, a deep learning framework based on improved autoencoders and attention mechanism, named MOAEAM, is proposed to address this issue. Specifically, a novel composite loss facilitates the extraction of omics-specific features, and a multi-omics integration module incorporates capture cross-omics information, collectively enhancing classification performance. Systematic evaluations across multiple cancer datasets show MOAEAM achieves consistently higher classification performance than current mainstream multi-omics integration methods. Ablation studies reveal that the auxiliary classifier introduced in the improved autoencoder plays a key role in performance improvements. The feature importance scores computed by the model identify potential clinically significant biomarkers, which are further validated through literature analysis and enrichment analysis.

Anxiety disorders (AD) are prevalent psychiatric conditions that profoundly impact adolescent neural development. Abnormal delta–beta cross-frequency coupling (CFC) has been identified as a key electrophysiological marker of altered neural dynamics in individuals with AD. However, most existing studies focus on static analysis within restricted brain regions and predefined frequency bands, which limits the understanding of large-scale dynamic neural communication. Therefore, we propose a novel cross-frequency coupling directed brain network (CFCDBN) framework, which integrates personalized CFC estimation and causal information flow modeling to capture the dynamic interactions of the brain network in AD. Personalized CFC significantly improves the precise representation of AD-related neural dynamics by adaptive frequency band division and individualized oscillation feature extraction, overcoming the limitations of traditional CFC methods. The analysis reveals significant delta-beta coupling abnormalities in the left hemisphere of AD, accompanied by disrupted directional pathways involving the thalamus, precuneus, and insula. These findings suggest impaired emotional and cognitive communication from the subcortical to cortical regions. To validate the efficacy of CFCDBN in distinguishing AD patients from healthy individuals, we developed a direction-aware graph neural network (DA-GNN) model that uses CFCDBN representations as input to capture dynamic neural patterns in causal brain connectivity. Experimental results show that the model consistently outperforms traditional machine learning methods and undirected GNN baselines in automatic AD identification, achieving a classification accuracy of 77.8%, and confirming the value of CFCDBN as a robust biomarker for AD-related network dysfunction. These findings not only deepen our understanding of the neural dynamics underlying AD, but also lay the foundation for personalized and mechanism-driven neuromodulation strategies. The core implementation of the CFCDBN framework is available on GitHub: https://github.com/wdxcjnb6/CFCDBN.

Morphological changes in oral mucosal microvasculature serve as early diagnostic markers for various diseases. However, existing dermoscopy image analysis relies heavily on physician expertise, leading to high subjectivity and low efficiency. This paper proposes a zero-shot capillary segmentation method for oral mucosa based on Segment Anything Model 2 (SAM2), which effectively handles reflection artifacts and highlights minute vascular structures through a multi-scale adaptive enhancement algorithm. The method employs a morphology-aware automatic prompt annotation strategy to generate composite guidance containing bounding boxes, foreground points, and background points for SAM2. Without requiring annotated data or model training, this approach achieves precise instance segmentation of capillaries through an “enhancement-annotation-segmentation” collaborative paradigm. On a clinical dataset comprising 212 dermoscopy images from 106 subjects, our method achieved a Dice coefficient of 0.7278 and an IoU of 0.5721, representing improvements of 17.12% and 26.9% respectively compared to the medical-specific baseline MedSAM. This provides an objective auxiliary diagnostic method for oral mucosal diseases that depend on capillary morphology analysis.