Base modifications are fundamental epigenetic marks widely present in both deoxyribonucleic acid (DNA) and ribonucleic acid (RNA), playing crucial roles in regulating gene expression, development, aging, and the pathogenesis of various diseases. Conventional detection methods often depend on harsh chemical treatments or amplification, which can introduce biases, damage biomolecules, and limit detection accuracy. The advent of third-generation long-read sequencing technologies, such as PacBio Single-Molecule Real-Time (SMRT) and Oxford Nanopore sequencing, has revolutionized the field by enabling direct, single-molecule resolution detection of base modifications without complex preprocessing. This review presents a comprehensive overview of the principles and advantages of these technologies, highlighting recent advances in the detection of DNA modifications (e.g., 5mC, 6mA, and 5hmC) and RNA modifications (e.g., m6A and $\Psi$). In contrast to prior reviews that typically focus on either DNA or RNA alone, our work provides an integrated perspective on modification detection across both molecular types, covering underlying mechanisms, computational tools, and technological innovations. We further provide a comparative analysis of representative detection methods, especially those that are compatible with the latest sequencing platforms and capable of supporting complex biological contexts. We also discuss emerging applications in epigenetics, disease research, and clinical diagnostics, alongside major challenges such as enhancing detection accuracy and enabling multi-omics integration. By unifying DNA and RNA modification analysis, this review offers a more holistic framework for advancing epigenetic research and its applications.

Federated medical image segmentation holds potential for training a global model to facilitate collaborative multi-site medical image analysis. However, the challenge of non-independent and non-identically distributed multi-site medical data complicates the task for Federated Learning (FL) methods. Heterogeneous feature distributions, resulting from varying imaging acquisition protocols or scanner vendors, often hinder the global model’s performance. While some previous studies have tackled the non-iid issue, they primarily focus on image classification tasks and cannot be readily applied to segmentation. In this paper, we present a novel method, Federated Attention-based Map Ensemble (FedAME), tailored to overcome the unique challenges posed by federated image segmentation. Firstly, FedAME introduces a novel knowledge type called attention-based structural knowledge, effectively harnessing the underlying semantic region while filtering out noisy regions. Secondly, to align local models before aggregation, we incorporate structure knowledge ensembles into local training, considering the drift of semantic information between global and local models. Thirdly, we propose a dynamic policy for adaptively generating consensus structural knowledge. It enhances collaboration among sites sharing similar feature distributions and effectively mitigates the negative contributions of dissimilar sites. Comprehensive evaluations and ablation studies conducted across multiple datasets show that FedAME outperforms state-of-the-art methods.

Unsupervised rationale extraction aims to identify concise text snippets supporting model predictions without annotated rationales. However, existing two-phase approaches often suffer from interlocking (where predictors overfit to poorly generated rationales) and spurious correlations (where models capture non-causal relationships between text and labels). To address these challenges, we propose a novel single-phase framework called You Only Forward Once (YOFO) that simultaneously explains and predicts. YOFO utilizes a Pre-trained Language Model (PLM) to gradually remove unimportant tokens, with the remaining tokens considered as rationales. We introduce a length configuration list to flexibly control the proportion of remaining tokens in each layer. Importantly, we replace the traditional zero-out operation for token deletion with Attention Mask Deletion (AMD), which significantly improves model performance. Experiments on the BeerAdvocate and Hotel Review datasets demonstrate that YOFO achieves state-of-the-art rationalization performance, improving token-level F1 by up to 14.5% compared to previous methods. Our approach effectively mitigates the interlocking and spurious correlation issues prevalent in two-phase models by eliminating the need for separate generation and prediction phases, offering a relatively robust and efficient approach for explaining Artificial Intelligence (AI) model predictions.

The short video market is currently growing rapidly. As a typical bandwidth-intensive application, short video streaming can easily cause bandwidth bottlenecks in servers. Hence, we propose a prefetching strategy to improve the quality of experience (QoE) and reduce the wastage of bandwidth, utilizing computility for performance optimization. We reveal that the state-of-the-art Dashlet prefetching strategy fails to maximize the weighted sum of QoE and bandwidth usage (i.e., utility). To overcome this limitation, the proposed comprehensive multistep prefetching strategy (CMPS) for short video streaming computes the prefetching urgency of chunks and combines it with expected utility to comprehensively evaluate multistep decision sequences, achieving higher QoE and bandwidth utilization than Dashlet. The extensive evaluations show that the proposed CMPS improves the average utility across diverse scenarios by 16.1% compared with Dashlet.

Deep learning offers notable promise for computational pathology, but its performance is constrained by the need for extensively annotated datasets, which are costly and laborious to produce. Self-Supervised Learning (SSL) provides an effective paradigm for learning discriminative representations from unannotated pathological images. However, existing SSL methods often overlook domain-specific characteristics of pathological images and suffer from the adverse effects of low-quality negative samples, leading to sub-optimal feature representations for downstream tasks. To overcome these limitations, we propose a novel Domain-Specific Self-supervised Contrastive Learning (DSSCL) framework, which incorporates two novel components: (1) a Stain-Separation Based Data Augmentation (SSDA) module that enhances stain-aware representation learning by fusing stain-separated components with original hematoxylin and eosin images, and (2) a Contrast-Aware Pair Refinement (CAPR) module that improves feature discriminability by filtering potential positives and mining hard negatives, thereby mitigating the influence of low-quality negatives. Extensive experiments demonstrate that DSSCL achieves comparable accuracy in classification tasks using only 0.1% labeled data compared to a network fine-tuned from ImageNet with 10% labeled data, while also delivering competitive performance in detection and segmentation tasks, underscoring its effectiveness in learning transferable and robust feature representations across diverse downstream tasks. The code is available at https://github.com/junjianli106/DSSCL.

The increase in cancer drug resistance poses an enormous challenge in implementing effective therapeutic interventions. Combination therapy has emerged as an effective method to combat this resistance, but traditional methods for identifying viable drug combinations are often cumbersome and resource intensive. Recently, computational models have been developed to simplify the prediction of viable drug combinations, thereby improving the efficiency of this field of research. However, many existing models treat drug combinations independently, ignoring the crucial interaction dynamics between them. Moreover, these models fail to exploit the complementary insights provided by cell line multiomics data. In this work, we propose MVCASyn, an innovative deep learning model that predicts synergistic drug combinations. Compared with existing models, MVCASyn combines a dual-view representation learning module to precisely extract the multilevel features of atomic interactions, and adopts a cross-attention mechanism to fuse cell line multiomics data. Our experimental results show that MVCASyn consistently outperforms the current advanced models across all the evaluation metrics. Visualization experiments of drug atomic importance scores further emphasize the ability of MVCASyn to identify key drug substructures. A case study experiment also confirms that MVCASyn is effective in practical applications. The code of MVCASyn is publicly accessible at https://doi.org/10.57760/sciencedb.31476.

Accurately predicting the survival of patients with esophageal cancer after esophagectomy is crucial for clinical precision treatment. However, the existing methods of predicting Overall Survival time (OStime) mostly build supervised learning with the uncensored data, ignoring the potential information hidden in the censored data. To utilize the information hidden in the clinically abundant censored data, we propose a Semi-Supervised Learning with Adaptive pseudo-label Selection and Correction (SSLASC) to predict the OStime of esophageal cancer using both uncensored and censored data. Specifically, we first transform the OStime regression problem to a classification task followed by Softmax Expected Value Refinement (SEVR) and train a Transformer network using the uncensored data, which is then used to predict the OStime for the censored data. Secondly, we design an adaptive pseudo-label selection strategy to dynamically select more classes and more balanced samples from the predicted censored data by allocating adaptive thresholds for different classes of samples when performing pseudo-label selection. Finally, a distribution correction and a meta label correction modules are proposed to make the selected pseudo-labels closer to the real overall OStime. We test SSLASC on an internal dataset and two external datasets with sample sizes of 327, 104, and 16, respectively. The experimental results demonstrate that SSLASC achieves Mean Absolute Error (MAE) of 12.23, 12.64, and 12.47 months on the three test datasets. Compared to the optimal State-Of-The-Art (SOTA) method, SSLASC improves performance by 1.09, 1.07, and 1.09 months, respectively. In addition, SSLASC also achieves the best performance in dichotomized survival analysis.

MicroRNAs (miRNAs) play a key role in the prevention, diagnosis, and treatment of complex diseases. However, identifying miRNA-disease associations (MDAs) through traditional methods is costly and time-consuming. Recent studies have reported numerous validated MDAs, forming the basis for the prediction of new MDAs using computational methods. In this study, we propose SAETNMDA, a computational method that applies fast kernel learning (FKL) and variant triplet networks to predict MDAs. First, miRNA and disease similarities are integrated into two kernels via FKL to enrich biological data. Next, feature representations are obtained by applying stacked autoencoders (SAEs) and triplet networks, enabling the identification of associated pairs by mapping them to nearby locations in the embedding space, while unassociated ones are mapped distantly. Finally, we utilize XGBoost (Extreme Gradient Boosting) to obtain predictive scores for MDAs from these features. SAETNMDA’s performance is evaluated with 5-fold cross-validation (5-fold-CV) and compared with other methods. It achieves the highest AUC and AUPR (0.9419, 0.4749 for HMDD v2.0; 0.9496, 0.5355 for HMDD v3.2, respectively). The performance is also validated on an independent dataset and de novo miRNAs, with SAETNMDA achieving the highest AUC and AUPR in all validations. Case studies also demonstrate the robust predictive capability of our method, with the top 50 predicted miRNAs validated for each of the three diseases. These results highlight SAETNMDA as an efficient model for MDA prediction. SAETNMDA’s source code is available at https://github.com/npxquynhdhsp/SAETNMDA.

A novel hybrid model combining a convolutional neural network (CNN) and a low-complexity Transformer network is introduced for predicting lung cancer response to neoadjuvant chemoimmunotherapy using computed tomography scans. This approach is crucial as it assists clinicians in identifying patients likely to benefit from treatment and in assessing their prognosis. The model employs channel splitting to minimize parameter count. It then leverages both CNN for local feature extraction and a streamlined Transformer for global feature comprehension. To enhance efficiency, a novel self-attention mechanism is implemented, focusing on feature aggregation and element-wise multiplication. To address the different semantic meanings of features, an attention-based module is designed to seamlessly integrate features from both networks, employing a process of coarse fusion, attention computation, and fine fusion. When evaluated with data from 232 lung cancer patients who have undergone neoadjuvant chemoimmunotherapy, the model demonstrates exceptional performance, achieving a Dice score of 47.04% and a 95.00% Hausdorff distance of 25.12 mm, outperforming existing methods. Additionally, it has only 2.91×10⁶ parameters and 52.95×10⁹ floating point operations. Moreover, the model’s predictive accuracy in tumor diameter estimation is beneficial for treatment planning. Its robustness is further validated through its application in stroke lesion prediction, indicating its broad applicability.

Predicting mortality risk in the Intensive Care Unit (ICU) using Electronic Medical Records (EMR) is crucial for identifying patients in need of immediate attention. However, the incompleteness and the variability of EMR features for each patient make mortality prediction challenging. This study proposes a multimodal representation learning framework based on a novel personalized graph-based fusion approach to address these challenges. The proposed approach involves constructing patient-specific modality aggregation graphs to provide information about the features associated with each patient from incomplete multimodal data, enabling the effective and explainable fusion of the incomplete features. Modality-specific encoders are employed to encode each modality feature separately. To tackle the variability and incompleteness of input features among patients, a novel personalized graph-based fusion method is proposed to fuse patient-specific multimodal feature representations based on the constructed modality aggregation graphs. Furthermore, a MultiModal Gated Contrastive Representation Learning (MMGCRL) method is proposed to facilitate capturing adequate complementary information from multimodal representations and improve model performance. We evaluate the proposed framework using the large-scale ICU dataset, MIMIC-III. Experimental results demonstrate its effectiveness in mortality prediction, outperforming several state-of-the-art methods.