Generalization challenges in electrocardiogram deep learning: insights from dataset characteristics and attention mechanism.

Deep learning's widespread use prompts heightened scrutiny, particularly in the biomedical fields, with a specific focus on model generalizability. This study delves into the influence of training data characteristics on the generalization performance of models, specifically in cardiac abnormality detection. Leveraging diverse electrocardiogram datasets, models are trained on subsets with varying characteristics and subsequently compared for performance.

A retrospective longitudinal cohort study of the clinical burden in myasthenia gravis.

Background: Patients with generalized myasthenia gravis (MG) often experience debilitating exacerbations, with the possibility of life-threatening respiratory crises requiring hospitalization. Long-term longitudinal studies are needed to understand the burden of MG, including in patients whose disease is refractory to conventional treatment.

Methods: A retrospective, longitudinal, cohort study was conducted of patients in England aged ≥ 18 years with treatment-refractory or non-refractory MG, using data recorded during 1997-2016 in the Clinical Practice Research Datalink and the Hospital Episode Statistics databases.

Structural Landscapes in Geometrically Frustrated Smectics.

A phenomenological free energy model is proposed to describe the behavior of smectic liquid crystals, an intermediate phase that exhibits orientational order and layering at the molecular scale. Advantageous properties render the functional amenable to numerical simulation. The model is applied to a number of scenarios involving geometric frustration, leading to emergent structures such as focal conic domains and oily streaks and enabling detailed elucidation of the very rich energy landscapes that arise in these problems.

Increasing cancer risk over calendar year in people with multiple sclerosis: a case-control study.

Background: Data on cancer prevalence and incidence in multiple sclerosis (MS) patients are controversial. This study is aimed at estimating cancer risk in MS patients.

Methods: Nested case-control study using data collected between 01/01/1987 and 28/02/2016 from the United Kingdom Clinical Practice Research Datalink.