Deep learning for psychiatric genomics: from tools to applications.

IKDP: Implicit Knowledge Enhanced Disease Prediction via heterogeneous admission sequence graphs.

Application of artificial intelligence in colonoscopy imaging for polyp analysis-A systematic review.

Explainable Convolutional Neural Networks for the identification of the Ampullariidae genus.

DrowsyDG-Phys: Generalizable driver drowsiness estimation in conditional automated vehicles using physiological signals.

UltraMN: Advancing Real-Time Median Nerve Ultrasound Monitoring With a Multitask Deep Learning Framework.

Artificial intelligence (AI) has been transforming many aspects of medical care. In prostate cancer, ongoing progress in AI has improved research and patient care. Recent advances in machine learning and deep learning have produced tools that help diagnose cancer, assess risk, and predict outcomes. In screening, AI-based risk calculators improve detection and help avoid unnecessary biopsies. Deep learning algorithms, particularly convolutional neural networks, have demonstrated expert-level performance in pathology, identifying malignancy and assigning Gleason grades with high accuracy. These tools also streamline workflow, flagging challenging cases for review and quantifying prognostic markers, such as Ki-67 and cribriform patterns. In addition, AI-based models can predict molecular alterations, microsatellite instability, and lymph node metastasis directly from histology images, providing cost-effective alternatives to traditional assays. The development of multimodal models integrates digital pathology and clinical parameters, enabling personalized treatment recommendations and improved outcome prediction. Natural language processing and large language models further expand AI's potential, facilitating information extraction from clinical notes and enhancing patient education. Despite these advances, most studies remain retrospective with heterogeneous endpoints. Performance often drops when models are tested at new sites because of differences in patient populations and slide preparation. Access to large, well-annotated datasets is limited, and technical variation hampers reproducibility. To move toward clinical use, the field needs prospective, multicenter validation, preanalytical and analytical standardization, and clear reporting of failure modes and human oversight. Emerging approaches, including self-supervised pretraining, transformer-based image models, and language-vision systems, are likely to improve generalization and support more personalized care.

Gestational age (GA) is essential for assessing fetal development, but conventional methods such as last menstrual period and ultrasound are often inaccurate, particularly in late pregnancy. Recent advances in deep learning (DL) and MRI offer more reliable and consistent GA estimation by capturing detailed fetal brain development. This study aimed to develop deep learning models for GA prediction using multi-view fetal brain MRI and to compare their performance with conventional biometric regression techniques. A total of 1,321 fetal MRI scans were used to train and evaluate various DL models, while an additional 80 publicly available MRI scans served as an external test set. Two training strategies were explored: transfer learning versus training from scratch, and single-view versus multi-modality input. The pre-trained ResNet-101 model achieved a mean absolute error (MAE) of 4.47days and a coefficient of determination (R2) of 0.96 on the internal test set. On the external test set, the model yielded an MAE of 6.57days, outperforming the biometric regression method, which achieved an MAE of 9.42days. Explainability analysis revealed that the model predominantly focused on the lateral ventricles, cerebellum, and surrounding brain regions for GA prediction. The integration of multi-view MRI and transfer learning significantly enhanced the predictive accuracy of DL models for GA estimation. The proposed approach outperformed conventional biometric regression and highlighted clinically relevant anatomical regions, demonstrating its potential for use in prenatal diagnostic applications.

Bridging Clinical Microbiology and Artificial Intelligence: An Image-Based Deep Learning Framework for Automated Antimicrobial Susceptibility Testing.

Vascular Endothelial Growth Factor (VEGF) plays a central role in angiogenesis, regulating both physiological processes such as wound healing, tissue repair, and bone formation, and pathological events including tumor progression, metastasis, and diabetic retinopathy. Due to its crucial role in vascular biology, VEGF serves as an important therapeutic target in anti-angiogenic drug development and precision medicine. However, conventional experimental methods for VEGF identification are costly and time-consuming, emphasizing the need for efficient computational approaches. To address this challenge, we introduce DeepStack-VEGF, an advanced deep learning framework designed for accurate and robust VEGF prediction. The model integrates diverse sequence-derived features, including physicochemical descriptors, sequential patterns, evolutionary information, and secondary structure motifs, further enhanced by pretrained embeddings from UniProt and ProtBert. Feature optimization was achieved using Support Vector Machine-Recursive Feature Elimination. DeepStack-VEGF employs a stacking ensemble of three architectures including Feedback Generative Adversarial Network Gated Recurrent Unit and Capsule Convolutional Neural Network each contributing distinct representational capabilities. Comprehensive evaluations demonstrate that the fused feature set and stacking ensemble substantially outperform individual models, achieving superior accuracy, robustness, and generalization. By combining deep learning with biological insight, DeepStack-VEGF provides a reliable and scalable computational framework for VEGF identification, supporting rational drug discovery, anti-angiogenic therapy design, and precision medicine applications.

The three-dimensional structure of a protein underpins its biological function, making structure determination and prediction central challenges in structural biology. Although experimental techniques such as X-ray crystallography, nuclear magnetic resonance (NMR), and cryo-electron microscopy (cryo-EM) can yield high-resolution structures, they are limited by low throughput, high cost, and demanding sample preparation. Likewise, traditional computational methods often perform poorly in the absence of homologous templates or under complex folding dynamics. Recent advances in deep learning and large-scale protein language models have transformed protein structure prediction. Models such as AlphaFold3 and RoseTTAFold achieve near-experimental accuracy by integrating evolutionary information, geometric constraints, and end-to-end neural architectures, while single-sequence approaches such as ESMFold offer substantial gains in speed and scalability. This review summarizes the biochemical foundations of protein folding, recent AI-driven methodological advances, and representative applications in drug discovery, enzyme engineering, and disease research, and discusses current challenges and future directions.

A systematic review of machine and deep learning techniques for acute lymphoblastic leukemia diagnosis.

Vehicular communication networks demand highly efficient and accurate channel estimation to ensure reliable data exchange in high mobility scenarios. The IEEE 802.11p standard is widely regarded as the foundation of the Vehicle-to-Vehicle (V2V) communication channel; however, it is constrained by limited pilot resources and a fixed pilot structure, which degrade the performance and effectiveness of traditional estimation techniques, particularly in dynamic environments. Recent advances in deep learning offer significant potential for addressing these issues by improving estimation accuracy and modelling complex channel dynamics. Though deep learning-based methods introduce trade-offs in computational complexity and accuracy, these are crucial constraints in latency-sensitive V2V scenarios. This article presents a comprehensive review of deep learning-based channel estimation techniques, analysing methods for the IEEE 802.11p standard and critically examining their limitations in both classical and deep learning-based approaches. Additionally, the article highlights improvements introduced by IEEE 802.11bd, which features an enhanced pilot structure and advanced modulation schemes, providing a more robust framework for adaptive, efficient channel estimation. By identifying future research pathways that balance delay, complexity, and accuracy, an intelligent and effective transportation system can be established.

As a critical step in industrial quality control, surface defect detection in aluminum materials remains challenging for minor defects despite advances in deep learning. To address this, this paper proposes an enhanced YOLOv8-based model, BFI-YOLO, that incorporates a Bidirectional Multi-scale Residual Network. Specifically, we design a Bidirectional Multi-scale Feature Pyramid Network (BM-FPN) based on BiFPN to strengthen cross-scale feature fusion. The parameter-free SimAM attention module is embedded to enhance subtle defect responses while suppressing background texture interference, without introducing additional computational overhead.Furthermore, we develop a Multi-scale Residual Convolution (MSRConv) module to capture defects of varying sizes on aluminum surfaces comprehensively. MSRConv utilizes multi-scale convolutional kernels to adapt to cross-scale defect features and retains shallow details via residual connections, thereby strengthening the model’s representation of fine defects. Extensive experiments on the public TAPSDD dataset show that BFI-YOLO achieves a precision of 91.3%, a recall of 89.8%, and mAP@0.5 of 92.1%, with only 1.8 M parameters. Compared to the baseline, BFI-YOLO reduces parameters by 40% while increasing mAP@0.5 by 4.2%, effectively balancing detection accuracy and lightweight performance. Optimized for resource-constrained industrial platforms such as embedded systems and mobile robots, BFI-YOLO meets real-time monitoring requirements while achieving competitive detection accuracy, providing an efficient and practical solution for metal surface defect detection.

Seed germination and early seedling development are critical determinants of crop establishment, stress tolerance, and yield stability, yet these stages remain insufficiently integrated into contemporary crop improvement strategies. Recent advances across genome editing, microbiome-assisted seed treatments, nanotechnology-enabled priming, and artificial intelligence-guided phenotyping have generated substantial but fragmented insights into early developmental regulation. This review synthesizes recent advances across early plant development research. It demonstrates that seemingly diverse technologies converge on a limited set of regulatory control nodes, including abscisic acid-gibberellin balance, redox homeostasis, and root system architectural plasticity. By integrating evidence from molecular, microbial, physicochemical, and computational studies, early plant ontogeny is presented as a tunable regulatory state governed by quantitative thresholds rather than as a strictly predetermined genetic process. Advances in deep learning, reinforcement learning, and high-throughput phenotyping further enable the modeling and optimization of early developmental trajectories across genotype by environment contexts. Together, these insights establish early development as a programmable target for crop improvement and provide a mechanistic foundation for designing integrated interventions that enhance developmental uniformity, stress resilience, and yield stability across diverse agroecological systems.

Sea turtle species identification is vital for marine biodiversity conservation, as sea turtles impact marine ecosystem balance by consuming dead seagrass and maintaining coral reefs. They help preserve the health of seagrass beds and coral reefs that benefit commercially valuable species. Therefore, to sustain sea turtle populations, detection systems that facilitate conservation efforts are essential. In developing underwater detection models, researchers must address several challenges specific to the underwater environment, including low illumination conditions, complex backgrounds, and underwater blur effects. In addition, YOLOv10-nano has emerged as the most efficient object detector in its family, though improving its performance remains a challenge. To overcome this issue, we propose an advanced deep learning approach using modified YOLOv10-nano with a new Parallel Fusion Module (PFM) integrated into the backbone alongside self-attention to enhance detection performance, named TurtleNet. The Parallel Fusion Module enhances detection performance by capturing channel-wise representational features. It emphasizes channels with relevant information through a dual-scaling process, improving feature quality. PFM is integrated into the untouched branch of the Partial Self-Attention mechanism to enrich the split half of the feature channels. Our model uses 48,302 images from Bunaken National Marine Park containing Green, Hawksbill, and Olive Ridley turtles with data augmentation applied. The method leverages YOLOv10-nano's real-time detection capabilities while the PFM optimizes feature fusion and localization accuracy. Experimental results show our model achieves an mAP50 score of 0.856 and runs at 28 FPS on CPU devices, outperforming existing approaches in precision, recall, and efficiency. This research combines computer vision with marine biology, creating an automated system that helps researchers and conservationists monitor endangered turtles.

Thyroid nodules are common, and accurate classification into benign or malignant types is essential for effective clinical management. Although high-resolution ultrasound is the primary diagnostic tool, its accuracy is limited by operator dependency. Recent advances in deep learning have shown promise for automated and objective assessment, but many existing methods lack focus on lesion-specific regions, compromising model robustness. To overcome these limitations, we propose a novel dual-branch deep learning framework that combines lesion segmentation and classification. A key feature of this framework is a nodule mask-guided feature enhancement module, which leverages probability masks from the segmentation branch to guide the classification branch toward diagnostically relevant regions while suppressing irrelevant information. Evaluated on ultrasound datasets from three medical centers, our approach demonstrates superior classification accuracy compared to baseline methods, highlighting its potential as a reliable computer-aided diagnosis tool for thyroid nodules.

Cryptocurrency markets are characterized by extreme volatility, rapid price fluctuations, and complex nonlinear behavior,making accurate forecasting a significant challenge for investors, analysts, and researchers. This study investigates the application of Time Series Analysis techniques to model and predict cryptocurrency prices using historical market data. Both traditional statistical approaches, such as the AutoRegressive Integrated Moving Average (ARIMA) model, and advanced deep learning methods, including Long Short-Term Memory (LSTM) networks, are implemented to capture underlying temporal patterns. The dataset consists of daily open, high, low, close prices, and trading volume obtained from reliable financial data sources. Data preprocessing steps such as handling missing values, normalization, stationarity testing using the Augmented Dickey-Fuller test, and time series decomposition are performed to ensure model efficiency and accuracy. Exploratory Data Analysis (EDA) is conducted to identify trends, seasonality, and volatility characteristics. Model performance is evaluated using statistical metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE). The comparative analysis demonstrates that while ARIMA performs adequately for short-term forecasting, LSTM models provide superior performance in capturing nonlinear and long-term dependencies within cryptocurrency price movements. However, external factors such as market sentiment and regulatory changes continue to influence prediction accuracy. This research contributes to a better understanding of cryptocurrency forecasting techniques and highlights the effectiveness of deep learning approaches in financial time series analysis. Keywords: Cryptocurrency, Time Series Analysis, ARIMA, LSTM, Price Prediction

Dementia is a progressive neurodegenerative disorder that severely impacts cognitive functions and daily living, especially in aging populations. Among its subtypes, Alzheimer's disease (AD) and frontotemporal dementia (FTD) exhibit overlapping clinical symptoms, making early and accurate differentiation a critical challenge. Electroencephalography (EEG), as a non-invasive and cost-effective modality, provides valuable insights into the neurophysiological disruptions associated with these conditions. This study aims to develop a robust EEG-based diagnostic framework capable of accurately classifying AD, FTD, and healthy controls (HC) by integrating domain-specific signal processing with advanced deep learning techniques. This study employed a publicly accessible dataset consisting of resting-state EEG recordings from a total of 88 participants, comprising 29 individuals with AD, 23 diagnosed with FTD, and 36 age-matched HC. The proposed model integrates Common Spatial Pattern (CSP) filtering with a sequential modified hybrid architecture that combines Convolutional Neural Networks (CNNs) and a Vision Transformer (ViT). By fusing domain-informed spatial filtering with deep hierarchical feature learning, the model captures both local signal characteristics and global contextual dependencies. A 10-fold cross-validation approach was employed to assess model performance and generalizability. The proposed model achieved notable classification accuracies of 95.86%, 94.76%, 94%, and 92.14% for the AD/HC, FTD/HC, AD/FTD, and AD/FTD/HC classification tasks, respectively. These results underscore the diagnostic potential of EEG-based deep learning frameworks in distinguishing among neurodegenerative conditions and highlight their promise in supporting more precise and individualized clinical interventions. This study presents a novel end-to-end EEG classification pipeline that fuses domain-guided spatial filtering with deep neural feature learning. The promising results suggest that the proposed method could serve as a valuable component in future clinical decision support systems for dementia, contingent upon further validation in real-world clinical settings.

Integrating multimodal data, such as unstructured clinical narratives and quantitative blood biomarkers, remains a major challenge in modern healthcare due to heterogeneous formats, complex semantics, and limited inter-modal correlations. Despite significant advances in deep learning, effective fusion of clinical text and biomarkers is still unresolved, restricting the full potential of precision medicine. We propose PSA-1DCNN, a novel Parallel Self-Attention 1D Convolutional Neural Network designed for multimodal integration in lung cancer detection. By combining self-attention mechanisms with 1D convolutional layers, PSA-1DCNN captures global semantic relationships from clinical text while learning local discriminative patterns from biomarker data. We further investigate four fusion strategies to optimize cross-modal information integration. Experiments conducted on MIMIC-III and MIMIC-IV demonstrate that PSA-1DCNN outperforms state-of-the-art baselines, including ClinicalBERT, LSTM, and 1D-CNN. Our best-performing configuration achieves an F1-score of 98.4% on MIMIC-IV, with strong cross-version generalization to MIMIC-III. SHAP-based interpretability further highlights the clinical relevance of key biomarkers such as WBC and RBC, alongside critical textual features. This study presents a scalable and interpretable framework that bridges heterogeneous modalities, advancing precision oncology and offering promising opportunities for personalized diagnostics.