6664 research outputs found
Sort by
Machine Learning - driven insights for predicting the impact of nanoparticles on the functionality of biomolecules, Illustrated by the case of DNA Damage-Inducible Transcript 3 (CHOP) inhibitors
Full text link. The presented study contributes to ongoing research that aims to overcome challenges in predicting the bio-applicability of nanoparticles (NPs). The approach explored a variety of combinations of nuclear magnetic resonance (NMR) spectroscopy data derived from the Simplified molecular-input line-entry system (SMILES) notations and small biomolecule features. The resulting datasets were utilised for machine learning (ML) with scikit-learn and deep neural networks (DNN) with PyTorch. Despite the obstacles in predicting how NPs influence biomolecule functionalities, the methodology was reasoned in terms of its applicability to compounds both with and without NPs. The methodology was illustrated through a quantitative high-throughput screening (qHTS) aimed at finding DNA Damage-Inducible Transcript 3 (CHOP) inhibitors. Based on this data, the optimal ML performance was achieved by the Random Forest Classifier, which was trained on 19,184 samples and tested on 4,000, resulting in 81.1% accuracy, 83.4% precision, 77.7% recall, 80.4% F1-score, 81.1% ROC, and a five-fold cross-validation score of 0.821. Complementing the main study, two computational approaches were developed to enhance CHOP inhibitor prediction. The first identifies the most desirable/undesirable functional groups for CHOP inhibition. The second, a CID_SID ML model, achieved 90.1% accuracy in predicting whether compounds designed for other purposes possess CHOP inhibition potential
In an AI-driven world, ‘prompt literacy’, is nursing’s new superpower
Full text linkThe NHS is already using artificial intelligence (AI) in ways that would have seemed impossible just five years ago. Algorithms predict which patients are likely to become frequent A&E users, enabling proactive interventions that cut attendances dramatically. Radiologists work alongside AI that can detect early-stage cancers in scans. Administrative teams use AI to streamline scheduling and reduce bureaucracy
Punching shear of hybrid fiber reinforced concrete flat slabs
No full textThis study investigates the use of polyvinyl alcohol (PVA), steel, and glass fibers to enhance the punching shear capacity of flat slabs. These hybrid fibers in concrete mixes involve considering the unique properties and benefits of each fiber type in concrete. Eight slab specimens, with dimensions of 1100 mm× 1100 mm x 120 mm, were tested to failure under concentric loading conditions. Test results showed that the punching shear capacity of slab elements using three types of hybrid fibers (PVA, steel, and glass) was higher than those containing two types of hybrid fibers by a range of 11 %–32 %. The deflection of slabs containing three hybrid fibers decreased by 2 %–12 % while the stiffness increased by 9 %–23 % compared to those containing two types of fibers. On the other hand, the slabs containing two types of hybrid fibers (PVA and steel fibers) led to a punching shear capacity higher than that using one type of fiber by a range of 4 %–10 %, a reduction in deflection by 2 %–10 %, and an increase in stiffness by 15 %–33 %. The maximum punching capacity and energy absorption of slabs containing three hybrid fibers were higher than those of the control slab without fibers by 71.43 % and 118 %, respectively. The analytical study carried out using the Fib Model showed that the predicted results were in good agreement with the experimental results. However, the ACI-318 (2019) design equations did not provide accurate predictions of the punching shear capacity. It was found that adding the three hybrid fibers in the concrete mixes resulted in benefitting the studied concrete elements from the unique properties of each fiber type in concrete
Consensus-guided evaluation of self-supervised learning in echocardiographic segmentation
No full textBackground:
Left ventricle segmentation is a fundamental task in echocardiography, essential for assessing cardiac function. However, deep learning models for segmentation rely on large labelled datasets, which are expensive and time-consuming to annotate. Self-supervised learning has emerged as a promising approach to leverage unlabelled data, but its effectiveness for left ventricle segmentation remains underexplored.
Methods:
This study investigates self-supervised learning for echocardiographic segmentation, comparing various pretext tasks. The impact of dataset size and distribution on pre-training is examined, revealing that excessive unlabelled data can degrade performance due to redundancy and low variability. A novel multi-expert labelled dataset is introduced to enhance segmentation evaluation, using consensus-based annotations to reduce annotation noise and improve reliability.
Results:
Among the self-supervised learning methods evaluated, contrastive learning consistently outperforms other approaches, particularly in low-label settings. The study demonstrates that AI models pre-trained using self-supervised learning and fine-tuned with only 15% of labelled data achieve stronger alignment with multi-expert consensus than any individual expert.
Conclusion:
The findings suggest that AI models can generalise well across expert annotations, providing more reliable and reproducible assessments
Guided acquisition of high-quality Echocardiogram using deep neural networks
Full text linkObjective:
The quality of echocardiographic image acquisition is vital for precise quantifications and diagnostic accuracy. However, ultrasound equipment is limited in performance throughput and image quality. It is also governed by the operators’ acquisition competence. Although, a subjective quality control process is adopted for standard procedures; to provide optimal quality image, this further introduces major drawbacks in the degree of consistency, quantifications, and diagnostic accuracy.
Materials and Methods:
A deep neural network model was established that used a large data set containing 40 000 echocardiograms and implemented a guided tool for objective optimization of the apical two chamber (A2C), apical four chamber (A4C) and parasternal long axis (PLAX) images, based on clinical protocols. This tool provided real-time quality feedback on image adequacy and gave the operators’ image optimization experience, as they examined patients.
Results:
An average computational speed at 4.24 ms per frame, with 0.032% model error rate, was achieved on apical visibility, anatomical clarity, depth-gain, and foreshortening graded attributes. The novel pipeline was comparable to the operators’ ultrasound guidance system for quality image acquisition and reliable diagnosis in the health care system.
Conclusion:
The result of a guided acquisition provided novel evidence for an objective optimization process, optimal image quality, diagnostic accuracy, and improved users’ acquisition experiences, in clinical practice. A subjective assessment of a sub-optimal image quality has the potential to negatively impact patients’ clinical care
Autoethnography pedagogy and practice: stories of interdisciplinary innovation
No full textAutoethnography Pedagogy and Practice supports and generates new insights into how autoethnography can be taught, supervised and practised by sharing the experiences and reflections of researchers from a wide range of fields and disciplines.
An international cast of leading researchers provide practical examples of how autoethnography can be successfully introduced into health and human sciences curricula, showcasing examples of the power of autoethnography within and beyond academia. By privileging contributors’ experiences within their own field of study as students, teachers, supervisors and researchers, this book explores how autoethnography can be introduced, nurtured and sustained in challenging academic environments. Each chapter considers three interrelated areas: Disciplinary Contexts, which examines autoethnography’s impact across different fields; Relationships, which considers how to successfully manage relational and care dynamics from undergraduate through professor levels; and Ethics, which addresses the many ethical considerations that can arise across a wide range of contexts.
Autoethnography Pedagogy and Practice is a book that encourages readers to engage in autoethnographic practice to create innovative, dialogical and collaborative texts that push the boundaries of polyvocality and diversity within their own disciplines. It will be of interest to researchers in Psychology, Medicine, Pharmacology, Allied Health, Nursing, Mental health, Sport and Exercise Science, Coaching, Sociology, Psychotherapy, Theatre Studies and Communication Studies
A comparative analysis of stochastic models for stock price forecasting: the influence of historical data duration and volatility regimes
Full text linkAccurate stock price forecasting is essential for informed financial decision-making. This study presents a comparative analysis of four foundational stochastic models—Geometric Brownian Motion (GBM), the Heston Stochastic Volatility model, the Merton Jump-Diffusion (MJD) model, and the Stochastic Volatility with Jumps (SVJ) model—each formulated to capture distinct features of financial market dynamics. Utilizing maximum likelihood estimation (MLE) for parameter calibration and Monte Carlo simulation for forecasting, we assessed model performance over varying historical calibration windows (3-month, 6-month, and 1-year) and a 3-months prediction horizon. Empirical findings demonstrate that the SVJ model consistently achieves superior predictive performance, as quantified by root mean square error (RMSE) and mean absolute percentage error (MAPE), across assets with both low and high volatility profiles. Moreover, the analysis reveals that for low-volatility stocks, such as AAPL and MSFT, a 1-year calibration window yields lower forecast errors, whereas for high-volatility stocks, such as TSLA and MRNA, a 6-month calibration window provides improved forecasting accuracy. These results highlight the importance of selecting model structures and estimation periods that align with the underlying volatility characteristics of the asset
Mortality and its predictors amongst patients with advanced dementia receiving psychiatric inpatient care
Full text linkPeople with dementia frequently develop behavioural and psychological symptoms, sometimes necessitating care in specialist dementia mental health wards. There has been little research on their life expectancy following admission or need for palliative care. The work presented here explores the mortality of these patients and whether this can be predicted at their time of admission to the ward
The untapped potential of the student nurse.
Full text linkWith all the demands on the time of community nurses, being allocated a student can often be perceived as a mandate that does not bring any immediate benefit to the supervisor, assessor and wider team. The supervision of students in the community requires one-to-one observation, which can limit the nurse’s feeling of personal autonomy. Students also require documentation to be completed to evidence learning and consideration of their individual learning needs, all of which take time. Although this role is a requirement of registered nurses, it is rarely given any protected time due to workload pressures
Enhanced Non-EEG Multimodal Seizure Detection: A Real-World Model for Identifying Generalised Seizures Across The Ictal State.
Full text linkNon-electroencephalogram seizure detection models hold promise for the early detection of generalised onset seizures. However, these models often experience high false alarm rates and difficulties in distinguishing normal movements from seizure manifestations. To address this, we were granted exclusive access to the newly developed Open Seizure Database, from which a representative dataset of 94 events was selected (42 generalised tonic-clonic seizures, 19 auras/focal seizures, and 33 seizures labelled as Other), with a combined duration of approximately 5 hours and 29 minutes. Each event contains acceleration and heart rate data which have been expertly annotated by a clinician, who labelled each 5-second timestep as Normal, Pre-Ictal, or Ictal. We then introduced the AMBER (Attention-guided Multi-Branching-pipeline with Enhanced Residual fusion) model. AMBER constructs multiple branches to form independent feature extraction pipelines for each sensing modality. The outputs of each branch are passed to our Residual Fusion layer, where the extracted features are combined into a fused representation and propagated through two densely connected blocks. The dataset was split by event, ensuring no overlap between events in the training and testing subsets. The model was trained using -fold cross-validation, where -1 folds were used for training and the remaining fold for validation. The results of these experiments highlight the effectiveness of Ictal-Phase Detection, with the model achieving an accuracy and -score of 0.9027 and 0.9035, respectively, on unseen test data. Notably, the model exhibited consistent generalisation, recording a True Positive Rate of 0.8342, 0.9485 and 0.9118 for the Normal, Pre-Ictal, and Ictal classes respectively, and an average False Positive Rate of 0.0502. In conclusion, this study introduces a new multimodal seizure detection technique and model that reduces the false alarm window and differentiates high and low-amplitude convulsive movements, laying the groundwork for further advancements in non-EEG-based seizure detection research