Background: Major Depressive Disorder (MDD) is one of the leading causes of disability worldwide. Deep learning methods have been widely used for MDD detection, with research suggesting that deep models outperform traditional machine learning techniques. However, detecting MDD remains challenging due to data heterogeneity, model complexities and the requirement for discriminative feature representations. Objective: This review outlines recent progress in deep learning methods for MDD detection from Resting-state fMRI (rs-fMRI), with a focus on the model’s generalisability and features that most effectively represent the function/anatomy of the brain to contribute to biomarker identifications and interpretability. Further, the review assesses the applicability of current models to real-world challenges. Methods: This systematic review followed the PRISMA guidelines. Studies involved clinically diagnosed MDD subjects, a control group, and deep learning methods for classification tasks. Results: The cerebellum, thalamus, amygdala, insula, and default mode network are the most frequently reported brain regions associated with depression. Although deep learning has shown impressive results, it has limitations in terms of reliance on labelled data, heterogeneity of data from various hospitals, and model interpretability. A majority of the studies lacked external validation and had a single-site dataset or regionally homogeneous datasets, and did not consider the temporal and dynamic nature of rs-fMRI data. Conclusion: Deep learning offers considerable potential in advancing MDD diagnosis and understanding its mechanisms. Multi-regional data collection, harmonisation techniques, and rigorous testing in real-world workflows should be the primary focus of future research.

BACKGROUND AND OBJECTIVE:Developing multimodal data-driven diagnostic systems has become a key clinical strategy for improving breast cancer outcomes. However, effectively modeling multimodal features remains challenging due to substantial semantic heterogeneity, scale discrepancies, and the inherent difficulty of cross-modal alignment. Although existing studies have proposed various multimodal fusion methods, most rely on direct feature concatenation or shallow integration, which fail to capture fine-grained intra-modality semantics as well as the complex interactions between histopathological and genomic modalities. METHODS:In this study, we propose a multimodal diagnostic framework based on Feature Enhancement and Semantic Collaborative Alignment (FESCA). The method incorporates a semantic-guided modality feature enhancement mechanism that effectively extracts and strengthens diagnostic cues from both pathological images and genomic data. In addition, a contrastive-learning-based cross-modal alignment strategy is introduced to map heterogeneous modalities into a unified semantic space and achieve deep semantic collaboration through contrastive optimization. To ensure robust breast cancer classification under varying modality availability, a multimodal collaborative diagnostic strategy is employed to dynamically adapt the feature representations. RESULTS:We evaluate FESCA on the TCGA-BRCA dataset, and the experimental results demonstrate that it outperforms state-of-the-art methods in breast cancer classification while significantly improving both intra-modality representation quality and cross-modal semantic alignment. CONCLUSION:To enhance accessibility and practical application, we developed a web-based breast cancer pathological staging diagnosis system to visualize and deploy the FESCA model, demonstrating a step toward clinical application and providing a benchmark for other research methods.

The electroencephalogram (EEG) is widely used to study brain activity, including sleep spindles, which are brief neural oscillations occurring during non-rapid eye movement sleep. Rats serve as valuable models for researching sleep disorders and neurological diseases. The manual detection of sleep spindles in EEG recordings is time-consuming and requires expert knowledge, driving the need for automated detection methods. However, most existing methods are designed for humans and cannot be directly applied to rodents due to differences in sleep spindle frequency, morphology, and amplitude. This study presents a gradient boosting classifier approach for detecting sleep spindles in rat EEG recordings. Left and right EEG activities from nine rats were utilized for training and validation, with an additional six rats used for independent testing. EEG recordings were segmented into 1-second epochs with 0.5-second overlap, and 18 features were estimated for classification. The proposed method achieved robust performance and reliably highlighted key predictive features, offering an efficient and reliable method for analyzing sleep spindle oscillations in rat EEG data.

Electroencephalogram (EEG) analysis is important for diagnosing neurological conditions but is often time-consuming and requires expert interpretation. This study focuses on automating the classification of normal and abnormal EEGs using the Temple University Hospital Abnormal EEG Dataset. We employed a ResNet50-based transfer learning method, utilizing 19-channel spectrograms and signal images as input to develop the method. Unlike previous methods that only analyzed specific segments of EEG data, we used entire EEGs to ensure comprehensive coverage, and our method achieved robust performance. Moreover, Gradient-weighted Class Activation Mapping (Grad-CAM) was used to provide explanations by highlighting the regions influencing method decisions. This method offers a promising step toward more efficient and interpretable automated EEG analysis.

OBJECTIVE:Animal models of drug-resistant epilepsy represent an important resource for discovering new drug targets and testing experimental medicines. Intra-amygdala microinjection of kainic acid in mice is one of the most widely regarded models of drug-resistant epilepsy. Mice develop acute status epilepticus, which diminishes after a few hours and then, within a few days, mice display spontaneous seizures. The frequency of spontaneous seizures varies between mice, with some developing low or high seizure rates. METHOD:We developed a feature-based and transfer learning-based approach, for predicting the emergent spontaneous seizure rates in the intra-amygdala kainic acid model based on the acute EEGs recorded in mice during status epilepticus lasting 40 minutes. The method was trained on data from 28 mice and tested on data from 16 mice. RESULTS:The feature-based and transfer learning-based models achieved accuracies of 69% and 75%, respectively on the test set in classifying emergent epilepsy as normal or outlier (i.e. low-frequency or high-frequency seizure rate). CONCLUSION:A limitation of the intra-amygdala kainic acid model has been the loss of time and resources from generating mice with low or high rates of spontaneous seizures. The feature-based and transfer learning-based models will assist researchers in identifying mice with a normal frequency of seizures before the onset of spontaneous seizures. SIGNIFICANCE:We have implemented this approach as a web server, which can potentially reduce the time and resources spent analysing the EEGs of mice who develop low-frequency or high-frequency seizure rates.

Breast cancer is one of the most prevalent cancers among women, with early detection playing a critical role in improving survival rates. This study introduces a novel transformer-based explainable model for breast cancer lesion segmentation (TEBLS), aimed at enhancing the accuracy and interpretability of breast cancer lesion segmentation in medical imaging. TEBLS integrates a multi-scale information fusion approach with a hierarchical vision transformer, capturing both local and global features by leveraging the self-attention mechanism. This model addresses the limitations of existing segmentation methods, such as the inability to effectively capture long-range dependencies and fine-grained semantic information. Additionally, TEBLS incorporates visualization techniques to provide insights into the segmentation process, enhancing the model’s interpretability for clinical use. Experiments demonstrate that TEBLS outperforms traditional and existing deep learning-based methods in segmenting complex breast cancer lesions with variations in size, shape, and texture, achieving a mean DSC of 81.86% and a mean AUC of 97.72% on the CBIS-DDSM test set. Our model not only improves segmentation accuracy but also offers a more explainable framework, which has the potential to be used in clinical settings.

Electroencephalogram (EEG) analysis is essential for diagnosing neurological disorders but typically requires expert interpretation and significant time. Purpose: This study aims to automate the classification of normal and abnormal EEG recordings to support clinical diagnosis and reduce manual workload. Automating the initial screening of EEGs can help clinicians quickly identify potential neurological abnormalities, enabling timely intervention and guiding further diagnostic and treatment strategies. Methodology: We utilized the Temple University Hospital EEG dataset to develop a DenseNet-based deep learning model. To enable a fair comparison of different EEG representations, we used three input types: signal images, spectrograms, and scalograms. To reduce dimensionality and simplify computation, we focused on two channels: T5 and O1. For interpretability, we applied Local Interpretable Model-agnostic Explanations (LIME) and Gradient-weighted Class Activation Mapping (Grad-CAM) to visualize the EEG regions influencing the model’s predictions. Key Findings: Among the input types, spectrogram-based representations achieved the highest classification accuracy, indicating that time-frequency features are especially effective for this task. The model demonstrated strong performance overall, and the integration of LIME and Grad-CAM provided transparent explanations of its decisions, enhancing interpretability. This approach offers a practical and interpretable solution for automated EEG screening, contributing to more efficient clinical workflows and better understanding of complex neurological conditions.

Autism spectrum disorder (ASD) is a neurodevelopmental condition caused by both genetic and environmental factors. Since no single gene variant accounts for more than 1% of the cases, the converging actions of ASD-related genes and other factors, including microRNAs (miRNAs), may contribute to ASD pathogenesis. To date, few studies have simultaneously investigated the mRNA and miRNA profiles in an ASD-relevant model. The BTBR mouse strain displays a range of behaviors with ASD-like features but little is known about the protein-coding and noncoding gene expression landscape that may underlie the ASD-like phenotype. Here we performed parallel mRNA and miRNA profiling using the prefrontal cortex (PFC) of BTBR and C57BL/6 J (B6) mice. This identified 1063 differentially expressed genes and 48 differentially expressed miRNAs. Integration of mRNA and miRNA data identified a strong inverse relationship between upregulated (DEGs) and downregulated miRNAs, and vice versa. Pathway analysis, taking account of the inverse relationship between differentially expressed miRNAs and their target mRNAs highlighted significant shared enrichment in immune signaling, myelination, and neurodevelopmental processes. Notably, miRNA changes were predicted to affect synapse-related functions but we did not find enrichment of protein-coding genes linked to cellular components or biological processes related to synapses in the PFC of BTBR mice, indicating processes may evade miRNA control. In contrast, other miRNAs were predicted to have extensive relationships with DEGs suggesting their role as potential hub coordinators of gene expression. Profiling findings were confirmed via qRT-PCR for representative protein-coding transcripts and miRNAs. Our study underscores the complex interplay between gene expression and miRNA regulation within immune and inflammatory pathways in the BTBR model, offering insights into the neurodevelopmental mechanisms of ASD. These results support the value of the BTBR mouse model and identify strategies that could adjust molecular pathways for therapeutic applications in ASD research.

Electroencephalogram (EEG) is an important tool in diagnosing neurological conditions, with normal abnormal classification often serving as the first step in identifying abnormalities associated with various diseases. However, many previous methods for EEG classification have relied on nonpublic datasets, limiting accessibility and reproducibility. In this study, we propose a lightweight CNN-LSTM-based model for classifying normal and abnormal EEGs using the publicly available TUH abnormal EEG dataset. Unlike previous methods that require multiple EEG channels, our approach uses only the T5-O1 channel, reducing complexity and computational costs. Additionally, our model utilizes convolutional layers for automatic feature selection, allowing it to identify important features without manual intervention, improving efficiency. We also apply Gradient-weighted Class Activation Mapping to visualize the model’s output, enhancing interpretability.

There is considerable evidence implicating maternal immune activation (MIA) and cytokine dysregulation in the pathophysiology of Autism. However, cytokines, due to their lack of specificity are unlikely to translate clinically as prognostic biomarkers. Our aim was to explore the perinatal molecular pathways dysregulated in umbilical cord blood, which precede a diagnosis of childhood Autism, and ascertain whether these putative biomarkers persisted into pre-pubertal childhood. In a cohort of 2137 mother-infant dyads, we conducted a nested case-control study in the BASELINE Birth Cohort. Proteomic and metabolomic analysis was performed on cord blood plasma from 22 children diagnosed with Autism before age 5, and 44 neurotypical controls. In a clinical diagnostic follow-up between 7–10 years in the PiRAMiD Cohort, 24 children with Autism and 48 controls provided blood samples for molecular profiling. In cord blood, proteomics revealed altered glycolysis, selenium metabolism, oxygen transport, and complement signalling. Alterations in these protein pathways persisted into childhood, and dysregulation of GAPDH, SELENBP1, and BLVRB proteins were evident in both cord blood and in serum from pre-pubertal children with Autism. In cord blood, metabolomics analysis indicated Autism outcome was associated with reduced levels of circulating steroids and increased sulfate. We confirmed androstenedione was reduced in cord blood, in Autism cases in comparison to controls, however changes in androstenedione levels were not evident in serum from pre-pubertal children with Autism. Our findings were further corroborated using machine learning approaches, with an AUROC ranging from 0.82 to 0.85 for proteomic and metabolomic cord blood prediction models, respectively. Collectively, these findings confirm a cord blood molecular signature precedes the onset of Autism and has the potential to lead to prognostic biomarkers. Our integrative multi-omics analysis reveals materno-feto-placental molecular processes which potentially underpin Autism aetiology.

MicroRNAs (miRNA) are small non-coding RNAs that are key negative regulators of gene expression. Their roles include shaping the gene expression landscape during and after brain development by defining and maintaining levels of proteins that generate the distinct morphological and functional properties of neurons and other brain cell types. HT22, N2A, and SH-SY5Y are common immortalized neuronal cell lines that offer simple, less expensive, and time-saving options for in vitro modelling to evaluate miRNA functions. The extent to which these lines reflect primary neurons remains, however, unclear. Here, we benchmarked the miRNA profiles of cultured mouse hippocampal neurons against Argonaute-loaded miRNAs in the adult mouse hippocampus and miRNA data from the hippocampus of control human donors. We then compared the miRNA expression landscape in HT22, N2A and SH-SY5Y against mouse hippocampal primary cell cultures. We profiled over 700 miRNAs across the lines and detected 310 miRNAs in all four cell types. This included detection of neuron-enriched miRNAs such as miR-124 and miR-128, although the cell lines typically displayed lower levels of these than in primary neurons and reference adult hippocampal tissue. The miRNA profile in the HT22 cell line showed the highest correlation to the mouse primary neuronal cultures. Together, this study provides a dataset on basal miRNA expression across commonly used cell lines for neuroscience research and evidence for both conserved and distinct profiles that should be used to inform decisions on cell lines for modelling brain and miRNA research.

Cervical cancer is the fourth most prevalent cancer among women, with over 600,000 new cases and 300,000 deaths reported annually. While early detection through Pap smear screening significantly reduces mortality, traditional methods are labor-intensive and heavily reliant on expert cytologists, creating barriers in resource-limited settings. This study proposes a cervical cancer screening through a Resnet-based method and interpretability technique (Inter-ResNet-CxCa) for classifying Pap smear cell images into three categories: unhealthy cells, healthy cells, and rubbish. By using pre-trained models, transfer learning lowers computational demands, mitigates overfitting, and enhances performance on task-specific datasets. To improve interpretability, we incorporate Local Interpretable Model-Agnostic Explanations (LIME), offering visual insights into model predictions. This model has the potential to enhance cervical cancer screening, particularly in resource-limited settings.

One of the most important aspects of students’ academic experiences is their sense of belonging, which supports their motivation, persistence, and overall wellbeing. Previous studies have shown that various complex factors, including gender, race, LGBTQIA+ membership, socioeconomic status, past experiences, and institutional culture, influence students’ sense of belonging in computing. This study compares undergraduate computing students’ sense of belonging in Ireland and Scotland, highlighting key demographic and contextual factors. We found that gender and minoritisation continue to be significant factors, but their impact differs across national settings. Our findings demonstrate that social interaction and clubs/societies membership are consistently associated with higher belonging in both Irish and Scottish contexts. This work supports the development of targeted interventions to improve students’ sense of belonging in computing, particularly across different regional and institutional environments. Future work will expand this comparison to include Wales and England to develop a broader perspective across Ireland and the UK.

A student's sense of belonging refers to their feeling of being accepted, valued, and included within their academic community. Belonging is a critical dimension of academic success and can influence a student's degree of academic adjustment, achievement, aspirations, or even whether they complete their studies. However, minoritised students are more likely to report lower levels of belonging compared to others in their cohorts. In STEM disciplines, mathematical confidence is often a key predictor of performance, even after accounting for actual mathematical skills. Despite its importance, little research has explored how mathematical confidence influences the sense of belonging of undergraduate computing students. This large (N = 738), multi-institutional (N = 14) study investigates the relationship between mathematical confidence and student sense of belonging while testing whether gender and minoritisation moderate this relationship. Our findings revealed a positive correlation (r = 0.3, p < 0.001) between confidence in mathematics and sense of belonging. Although we found no overall statistically significant difference in mathematical confidence between genders or minoritised/non-minoritised students, across all levels of mathematical confidence, men and non-minoritised students reported higher belonging than women and minoritised students. These results emphasise the importance of targeted interventions to increase the confidence of computing students in their mathematical abilities, particularly those who are minoritised in computing. Initiatives such as mentorship programs, confidence-building exercises, and more inclusive teaching strategies could play a role in fostering a stronger sense of belonging and improving academic success.

Effective monitoring of patients’ conditions is crucial in medical practice. Machine learning methods hold promise for automating disease detection, including epilepsy. However, the opacity of these black-box models presents significant challenges. In this study, we propose a LightGBM-based automatic classification model for epilepsy trained using TUH EEG data. Ten channels were employed, and 22 features from both the time and frequency domains were estimated from each channel. The model can distinguish between normal EEG, focal epilepsy, and generalised epilepsy. We trained the model on a dataset comprising 600 adult EEG records. Five-fold cross-validation yielded a mean accuracy of 89.46% and a mean F1 score of 0.8907 on the training set. To assess the model’s generalization performance, we independently tested it on 456 EEGs, achieving an accuracy of 71.49% and a weighted F1-score of 0.7386. Furthermore, we employed permutation feature importance, SHAP and LIME, to provide explanations for the model’s decisions. This model has the potential to gain the trust of clinicians and facilitate its adoption in clinical settings.

About 70 million people globally have been diagnosed with epilepsy. Electroencephalogram (EEG) devices are the primary method for identifying and monitoring seizures. The use of EEG expands the preclinical research involving the long-term recording of neuro-activities in rodent models of epilepsy targeted towards the efficient testing of prospective antiseizure medications. Typically, trained epileptologists visually analyse long-term EEG recordings, which is time-consuming and subject to expert variability. Automated epileptiform discharge detection using machine learning or deep learning methods is an effective approach to tackling these challenges. This systematic review examined and summarised the last 30 years of research on detecting epileptiform discharge in rodent models of epilepsy using machine learning and deep learning methods. A comprehensive literature search was conducted on two databases, PubMed and Google Scholar. Following the PRISMA protocol, the 3021 retrieved articles were filtered to 21 based on inclusion and exclusion criteria. An additional article was obtained through the reference list. Hence, 22 articles were selected for critical analysis in this review. These articles revealed the seizure type, features and feature engineering, machine learning and deep learning methods, training methodologies, evaluation metrics so far explored, and models deployed for real-world validation. Although these studies have advanced the field of epilepsy research, the majority of the models are experimental. Further studies are required to fill in the identified gaps and expedite preclinical research in epilepsy, ultimately leading to translational research.

EEG is a test that helps in the clinical diagnosis of epilepsy. Epilepsy diagnosis is facilitated by establishing the presence of interictal epileptiform abnormalities on EEG, which predict an increased risk of seizure. The identification of interictal epileptiform discharges is a time-consuming task that requires highly-trained experts. A method to assist in the recognition of EEGs with epileptiform abnormalities was developed using transfer learning on multiple channels of paediatric EEGs, without the use of human annotations. The dataset included 350 children with normal EEGs and 597 children with interictal abnormalities, and it was divided into training data (n=452), validation data (n=112), and testing data (n=383). Spectrograms from each EEG signal channel were used as input for five pre-trained transfer learning models (Inception, ResNet, DenseNet, VGG16 and VGG19) and traditional feature-based machine learning methods were developed as a benchmark. A comparison was made between a transfer learning-based method and a traditional feature-based machine learning algorithm. The results revealed that the transfer learning-based method outperformed the feature-based machine learning methods, achieving an accuracy of 77%, an F1 score of 0.85, and a balanced accuracy of 77% on the test set. Our transfer learning-based method can identify interictal abnormalities without the need for feature estimation by domain experts or human annotations. This method can assist in the recognition of EEGs with epileptiform abnormalities in children thereby facilitating the clinical diagnosis of epilepsy.

Background: Epilepsy, a prevalent neurological disorder characterized by recurrent seizures affecting an estimated 70 million people worldwide, poses a significant diagnostic challenge. EEG serves as an important tool in identifying these seizures, but the manual examination of EEGs by experts is time-consuming. To expedite this process, automated seizure detection methods have emerged as powerful aids for expert EEG analysis. It is worth noting that while such methods are well-established for adult EEGs, they have been underdeveloped for pediatric and adolescent EEGs. This study sought to address this gap by devising an automatic seizure detection system tailored for pediatric and adolescent EEG data. Methods: Leveraging publicly available datasets, the TUH pediatric and adolescent EEG and CHB-MIT EEG datasets, the machine learning-based models were constructed. The TUH pediatric and adolescent EEG dataset was divided into training (n = 118), validation (n = 19), and testing (n = 37) subsets, with special attention to ensure a clear demarcation between the individuals in the training and test sets to preserve the test set’s independence. The CHB-MIT EEG dataset was used as an external test set. Age and sex were incorporated as features in the models to investigate their potential influence on seizure detection. Results: By leveraging 20 features extracted from both time and frequency domains, along with age as an additional feature, the method achieved an accuracy of 98.95% on the TUH test set and 64.82% on the CHB-MIT external test set. Our investigation revealed that age is a crucial factor for accurate seizure detection in pediatric and adolescent EEGs. Conclusion: The outcomes of this study hold substantial promise in supporting researchers and clinicians engaged in the automated analysis of seizures in pediatric and adolescent EEGs.

Sense of belonging, or belongingness in academia, is an individual’s personal conviction as to their acceptance as a valued member of an academic community. The importance of belongingness lies in correlations with motivation, persistence, mental health and well-being. Prior work has shown that belongingness can be lower in students who are minoritised due to their gender, race, nationality, socio-economic status, religion and/or disability. However, there is limited research on how a student’s university entry route impacts their sense of belonging in Computer Science and Other Science disciplines. In Ireland’s higher education system, university entry is primarily managed through the Central Applications Office (CAO) process, which predominantly serves students transitioning directly from secondary school to university. Additionally, alternative access routes are available for students from socio-economically disadvantaged backgrounds, those with disabilities, and mature students. At University College Dublin (UCD), entry is facilitated through both the traditional direct-entry school leaving route and combined access routes, each serving distinct student demographics. To investigate how these varying entry routes influence students’ sense of belonging, we utilised a survey adapted from the validated ‘Math Sense of Belonging Scale’ to examine the belongingness of undergraduate Science students, including Computer Science students. We examined how the belongingness of students varies by university entry route, and how the belongingness of students in Computer Science varies by university entry route compared to Other Science students. Our results reveal a significant difference in belongingness between students entering university through the direct-entry school leaving route, compared to those who entered via combined access routes (e.g. university admissions schemes for school leavers from socio-economically disadvantaged backgrounds, with disabilities, and mature students). Specifically, within Computer Science, students entering university through the direct-entry school leaving route had significantly higher belongingness than combined access route students. These results provide insight that may help others improve the belongingness of undergraduate Computer Science students.