8383 research outputs found
Sort by
Hybridized artificial intelligence system for reducing neonatal mortality in Nigeria
Get PDFBackground
Neonatal diseases represent the leading cause of death in Nigeria, ranking the country second globally in neonatal mortality rates. Early and accurate diagnosis remains challenging, leading to delayed interventions and increased mortality.
Aim
To develop an artificial intelligence system capable of detecting multiple neonatal diseases using local datasets and advanced machine learning techniques to facilitate early intervention and reduce neonatal mortality in Southwest Nigeria.
Methods
Clinical records from 4,027 previously treated neonatal patients were collected from five tertiary hospitals across three Southwest Nigerian states. The dataset underwent comprehensive analysis, balancing using Synthetic Minority Over-sampling Technique (SMOTE), and preprocessing before training three deep learning models: Artificial Neural Networks (ANN), Long Short-Term Memory (LSTM), and a novel hybrid LSTM-ANN architecture. Model performance was evaluated using accuracy, precision, recall, and F1-score metrics with rigorous subject-wise validation and statistical testing.
Results
The hybrid LSTM-ANN model demonstrated superior performance with 82 % accuracy, 88 % precision, 82 % recall, and 86 % F1-score, significantly outperforming both standalone ANN (80 % accuracy) and LSTM (77 % accuracy). Disease-specific classification revealed exceptional performance for sepsis (precision: 0.90, F1-score: 0.88), birth asphyxia (0.88, 0.85), jaundice (0.86, 0.83), and prematurity (0.82, 0.80). McNemar’s test confirmed significant hybrid superiority over ANN (χ2 = 12.45, p < 0.001) and LSTM (χ2 = 15.67, p < 0.001), whilst Friedman test (χ2 = 18.42, p < 0.001) validated the 5–6 % accuracy improvement.
Conclusion
The hybrid LSTM-ANN model establishes a valuable diagnostic tool for early neonatal disease detection. However, external validation and prospective clinical trials are necessary before clinical deployment
Artificial intelligence in clinical trials: A comprehensive review of opportunities, challenges, and future directions
Get PDFBackground
Clinical trials face unprecedented challenges including recruitment delays affecting 80% of studies, escalating costs exceeding $200 billion annually in pharmaceutical R&D, success rates below 12%, and data quality issues affecting 50% of datasets. Artificial intelligence (AI) offers transformative solutions to address these systemic inefficiencies across the clinical trial lifecycle.
Objective
To evaluate the current state, future potential, and implementation challenges of AI technologies in clinical trials, providing evidence-based guidance for responsible AI integration while maintaining patient safety and scientific integrity.
Method
Comprehensive narrative review following established guidelines for literature synthesis. Systematic search of PubMed, Embase, IEEE Xplore, and Google Scholar databases from January 2015 to December 2024. Data extraction and narrative synthesis organized thematically according to clinical trial lifecycle stages.
Results
Analysis of relevant studies demonstrated substantial AI benefits: patient recruitment tools improved enrollment rates by 65%, predictive analytics models achieved 85% accuracy in forecasting trial outcomes, and AI integration accelerated trial timelines by 30–50% while reducing costs by up to 40%. Digital biomarkers enabled continuous monitoring with 90% sensitivity for adverse event detection. However, significant implementation barriers emerged, including data interoperability challenges, regulatory uncertainty, algorithmic bias concerns, and limited stakeholder trust.
Conclusion
AI represents a transformative force in clinical research with proven capabilities to enhance efficiency, reduce costs, and improve patient outcomes. Realizing this potential requires addressing technical infrastructure limitations, developing explainable AI systems, establishing comprehensive regulatory frameworks, and fostering collaborative efforts between technology developers, clinical researchers, and regulatory agencies to ensure responsible implementation
Comparative Genomic Hybridization (CGH) in Genotoxicology: From the Basics to Modern Approaches.
No full textOver the past two decades, comparative genomic hybridization (CGH) and array CGH have become essential tools in clinical diagnostics, oncology, and toxicological risk assessment. Initially developed to identify chromosomal imbalances like copy number variations (CNVs) in tumor cells, these technologies have expanded into genotoxicology and toxicogenomics, exploring gene responses to toxic agents and their molecular mechanisms. As of 2024, new developments include integrating array CGH with next-generation sequencing (NGS), machine learning, and CRISPR-Cas9 genome editing, greatly improving precision. High-density CGH arrays now offer single-cell resolution, enabling the detection of cellular heterogeneity in toxic responses, while long-read sequencing facilitates the identification of complex genomic rearrangements. Recent innovations include combining CGH and toxicogenomics with organ-on-chip models for real-time, tissue-specific toxicological assessment. This has significantly improved the relevance of toxicological data for human health. However, while these advances are promising, array CGH remains costly and requires substantial data processing, driving the need for advanced bioinformatics tools. AI-driven predictive toxicology models are also gaining traction, correlating toxicogenomic profiles with clinical outcomes. Despite these advancements, the field still faces challenges, such as evolving regulatory guidelines and complex data interpretation, which hinder broader adoption and the full realization of CGH's potential in toxicology and risk assessment. [Abstract copyright: © 2026. The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature.
Evaluating the Wellbeing of ENT Trainees in the UK: Survey Findings
No full textObjectives
The Association of Otolaryngologists in Training wanted to assess trainee wellbeing.
Methods
A survey was developed, incorporating the Copenhagen Burnout Inventory, short Warwick–Edinburgh Mental Wellbeing Scale and Brief Resilience Scale plus questions on working conditions.
Results
There were 190 responses and while most respondents had low or moderate levels of burnout, 15% had high personal burnout and 13% had high work-related burnout. The mean wellbeing score for respondents was lower than the whole population mean. 39% reported mental wellbeing has been slightly affected in a negative way by their working environment and conditions in the last six months, and 26% reported it being significantly affected negatively. Of these, 43 respondents reported an impact on patient safety.
Conclusions
This first ever survey of ENT trainees in the UK identified several areas of concern including how the working environment and conditions affect trainee wellbeing and impact on patient safety
Women and Pro-environmental Initiatives in Tourism: The Intersection of Gender Dynamics and Environment Issues
No full textWomen’s involvement, gender equality, and environmental challenges are critical concepts in tourism development, particularly in patriarchal systems where hierarchical value systems create challenges for both nature and women. However, the theoretical framework addressing these issues remains underexplored. This study examined how women interpret their motivations and challenges in these contexts while contributing to tourism-related environmental conservation and navigating patriarchal gender norms. It investigated the lived experiences of Iranian women actively engaged in tourism-related conservational initiatives. The findings revealed that their commitment to pro-environmental initiatives is driven by an intrinsic connection to nature and a desire to resist gender inequalities and challenge male-dominated structures. Despite facing challenges stemming from patriarchal structures, they demonstrate collective resilience against gender stereotypes. Additionally, their participation in tourism pro-environmental projects fosters trust and social recognition, driving incremental social change. This study makes a strong theoretical contribution by exploring gender issues through the lens of ecofeminism theory
Adipocytes in tumour microenvironment promote chemoresistance in triple negative breast 3 cancer through oxysterols
Get PDFObjective: Triple-negative breast cancer (TNBC) patients with excess adipose tissue experience poorer disease-free survival than those with a healthy body mass index.
Adipocytes store and release cholesterol, which can be hydroxylated to form oxysterols. These cholesterol derivatives activate the liver X receptor (LXR) pathway. This study tested the hypothesis that adipocytes contribute to an imbalanced tumour-microenvironment by exposing cancer cells to elevated oxysterols, mimicking chemotherapy-exposure conditions and priming for chemoresistance.
Methods: Tumour tissue microarray from 148 TNBC patients was assessed using immunohistochemistry for CH25H, CYP46A1, CYP27A1 and P-glycoprotein (Pgp), expression and survival outcomes assessed. Gene expression was compared between tumours from patients (GSE78958) and mouse models (GSE151866) with high versus low adiposity. In vitro, cell lines from lineages fount in the tumour-microenvironment were evaluated for oxysterol content, secretion, expression of relevant enzymes, and ability to induce Pgp expression and drug resistance in TNBC cells.
Results: In patients, stromal expression of oxysterol-synthesizing enzymes correlated with Pgp expression in cancer epithelial cells and was associated with shorter disease-free survival. Adipocytes conditioned media contained significantly higher oxysterols levels than that conditioned by other cell types and induced Pgp expression and drug resistance in MDA.MB.468 cells. Obese mice had elevated levels of Pgp in tumours compared to lean
counterparts.
Conclusions: Adipocytes secrete oxysterols that promote drug resistance in vitro and correlate with oxysterol:Pgp axis and survival in vivo.
Significance: This study reveals a mechanism by which adipose tissue contributes to drug resistance in ER-negative breast cancers, identifying the oxysterol-Pgp axis as potential therapeutic target
Experiences of bullying and harassment, including sexual harassment, amongst ENT trainees in the UK: survey findings
No full textObjectives
The Association of Otolaryngologists in Training (AOT) wanted to assess the experiences of bullying, harassment and raising concerns in their Ear, Nose and Throat (ENT) posts.
Methods
An online survey of ENT trainees, with 190 responses out of 350 targeted, included questions on bullying and harassment.
Results
Many respondents had experienced or witnessed a range of bullying, harassment and sexual harassment behaviours, including: unrealistic expectations about workload, responsibilities or level of competence; inadequate or absent supervision; and undervaluing someone’s contribution (in their presence or otherwise). However, very few (5% or less) had reported them. 21% would not feel confident in reporting bullying/harassment or sexual harassment problems and 40% do not feel safe raising concerns. Just 10% said the existing reporting mechanisms are sufficient.
Conclusions
A number of initiatives have been introduced recently in the UK to address bullying and harassment within the medical workplace, but there is still potential for further development
Enhancing Cardiovascular Disease Prediction: Optimised Feature Selection and Machine Learning Techniques for Improved Accuracy
No full textCardiovascular disorders (CVD) make a notable contribution to the global death toll, signifying the urgent need for precise prediction and proactive management tools. This study investigates the incorporation of advanced feature selection techniques with machine learning models to provide better predictions regarding cardiovascular disease in terms of accuracy and clarity. A merged healthcare dataset was used to address the common challenges such as small data size, incomplete or missing data and high dimensionality. Feature selection and dimensionality reduction through PCA and SHAP were used according to how much variance and importance they applied to features. The result demonstrates that SHAP-extracted features, which are fewer, obtained better performance compared to PCA and full-feature models. In addition, the fewer features from SHAP offered a more computationally efficient and interpretable solution. These results underscore the potential of incorporating explainable AI into clinical decision-making processes and the early diagnosis of cardiovascular disease
Mitigating Data Scarcity in Healthcare through Wasserstein Generative Adversarial Network- A Case Study in Medical Application
No full textBreast cancer represents a significant public health challenge, necessitating accurate predictive models for timely diagnosis and effective treatment. However, the scarcity and privacy constraints of medical datasets present substantial obstacles to developing robust predictive models. This study explores the application of Wasserstein Generative Adversarial Network (WGAN) to mitigate these challenges by generating synthetic breast cancer data. Using a comprehensive methodology that includes feature engineering, WGAN training, and model evaluation techniques, this research demonstrates the potential and effectiveness of integrating GAN-generated synthetic data into predictive modeling tasks. Evaluation metrics like Mean Squared Error (MSE) and Wasserstein distance (WD) are used to evaluate the quality of the generated data. Additionally, Random Forest is employed to evaluate model performance through accuracy scores and Area Under Curve (AUC) scores, which help assess the effectiveness of the predictive model for real, synthetic and combined data. The findings highlight the potential of WGAN to effectively enhance data availability and diversity, thereby improving predictive model performance in applications like breast cancer diagnosis and prognosis. In the future, ongoing advancements in GAN technology offer promising opportunities to refine data-driven methodologies in healthcare and advance patient care outcomes