Explore the vast range of titles available.

A brain computer interface (BCI) framework uses computer algorithms to detect mental activity patterns and manipulate external devices. Because of its simplicity and non-invasiveness, one of the most commonly used imaging technologies is electroencephalography (EEG). The evaluative method used in assessing the output of an EEG-based BCI system is classifying EEG signals for particular applications. The growth of artificial intelligence technology inspired researchers to use machine learning (ML) techniques and deep learning (DL) approaches to classify EEG-based BCI. Machine learning techniques enable the brain computer interface to learn from the subject's brain with each new session, adapting the generated rules for classifying thoughts and thus improving the system's efficiency. The authors present a concentrated survey on the use of various ML/DL techniques in EEG-based BCI. Three EEG paradigms for classification are used: motor imagery, p300, and steady state evoked potential. In addition, the challenges that recent EEG-based BCI systems face are addressed based on ideal signal processing methods, BCI functioning, performance assessment and commercialization. The authors hope that the information gathered would aid in application of suitable machine learning techniques, as well as provide a foundation for BCI researchers to enhance future BCI system.

Cross-lingual transfer learning using multilingual models has shown promise for improving performance on natural language processing tasks with limited training data. However, translation can introduce superficial patterns that negatively impact model generalization. This paper evaluates two state-of-the-art multilingual models, Cross-Lingual Model-Robustly Optimized BERT Pretraining Approach (XLM-Roberta) and Multilingual Bi-directional Auto-Regressive Transformer (mBART), on the cross-lingual natural language inference (XNLI) natural language inference task using both original and machine-translated evaluation sets. Our analysis demonstrates that translation can facilitate cross-lingual transfer learning, but maintaining linguistic patterns is critical. The results provide insights into the strengths and limitations of state-of-the-art multilingual natural language processing architectures for cross-lingual understanding.

The identification and classification of human brain activities are essential for many medical and Brain-Computer Interface (BCI) systems, saving human lives and time. Electroencephalogram (EEG) proves to be an efficient, non-invasive, and cost-effective means of recording electrical signals in the human brain. With the advancement in Artificial Intelligence, various techniques have emerged that provide efficient ways of classifying EEG signals to solve real-life challenges. One such method is Convolutional Neural Network (CNN), which has received considerable research attention. This paper presents a systematic review of CNN techniques for the identification and classification of EEG signals and their main achievements. The review has considered the most reliable studies from various fields and application domains where CNN has been used for EEG signal classification or identification. The review also highlights the approaches taken so far. While there are many available survey types of research, none has provided a comprehensive view of a particular model for EGG-signal analysis. This survey focuses on the successful deployment of CNN models in various application domains that use EEG signals. Additionally, this paper attempts to answer research questions and discusses current challenges. The presented detailed review strengthens the belief in the futuristic potential that CNN has in solving real-world problems using EEG signals. All of this indicates that CNN-based EEG-signal analysis is a promising field with exciting opportunities for research enthusiasts.

Human single-strand-selective monofunctional uracil DNA glycosylase 1 (hSMUG1) removes uracil, 5-hydroxymethyluracil (5hmU) and 5-fluorouracil (5FU) from DNA, thereby initiating the base excision repair (BER) process. hSMUG1 is important for maintaining genomic integrity and plays a significant role in cancer biology. Here, we present the structures of hSMUG1, including complexes with products (uracil and 5FU) and an enzyme-product complex of hSMUG1 with double-stranded DNA (dsDNA). Analysis of our hSMUG1-dsDNA complex reveals how uracil is flipped out of the dsDNA for excision and identifies key residues that we confirm to be critical for both DNA binding and enzymatic activity. Furthermore, our hSMUG1 substrate complexes, molecular dynamics simulations and neutron diffraction data suggest a mechanism by which the substrate uracil rotates following base excision. The structural and functional information presented here will be highly useful for the future development of inhibitors and/or activators targeting hSMUG1. DNA repair enzymes such as SMUG1 are essential for maintaining genomic integrity and have important implications in cancer biology. Here, the authors present the structures of human SMUG1 and reveal how it recognises and excises mutagenic uracil from DNA, identifying key residues and proposing a post excision rotation mechanism.

In a pilot study, lithium treatment slowed progression of amyotrophic lateral sclerosis (ALS). We aimed to confirm or disprove these findings by assessing the safety and efficacy of lithium in combination with riluzole in patients with ALS. We did a double-blind, placebo-controlled trial with a time-to-event design. Between January and June, 2009, patients with ALS who were taking a stable dose of riluzole for at least 30 days were randomly assigned (1:1) by a centralised computer to receive either lithium or placebo. Patients, caregivers, investigators, and all site study staff with the exception of site pharmacists were masked to treatment assignment. The primary endpoint was the time to an event, defined as a decrease of at least six points on the revised ALS functional rating scale score or death. Interim analyses were planned for when 84 patients had been allocated treatment, 6 months later or after 55 events, and after 100 events. Analysis was by intention to treat. The stopping boundary for futility at the first interim analysis was a p value of at least 0·68. We used a log-rank test to compare the distributions of the time to an event between the lithium and placebo groups. This trial is registered with ClinicalTrials.gov, NCT00818389. At the first interim analysis, 22 of 40 patients in the lithium group had an event compared with 20 of 44 patients in the placebo group (log rank p=0·51). The hazard ratio of reaching the primary endpoint was 1·13 (95% CI 0·61–2·07). The study was stopped at the first interim analysis because criterion for futility was met (p=0·78). The difference in mean decline in the ALS functional rating scale score between the lithium group and the placebo group was 0·15 (95% CI −0·43 to 0·73, p=0·61). There were no major safety concerns. Falls (p=0·04) and back pain (p=0·05) were more common in the lithium group than in the placebo group. We found no evidence that lithium in combination with riluzole slows progression of ALS more than riluzole alone. The time-to-event endpoint and use of prespecified interim analyses enabled a clear result to be obtained rapidly. This design should be considered for future trials testing the therapeutic efficacy of drugs that are easily accessible to people with ALS. National Institute of Neurological Disorders and Stroke, ALS Association, and ALS Society of Canada.

Recent studies have exposed the vulnerabilities of deep neural networks to some carefully perturbed input data. We propose a novel untargeted white box adversarial attack, the dynamic programming-based sub-pixel score method (SPSM) attack (DPSPSM), which is a variation of the traditional gradient-based white box adversarial approach that is limited by a fixed hamming distance using a dynamic programming-based structure. It is stimulated using a pixel score metric technique, the SPSM, which is introduced in this paper. In contrast to the conventional gradient-based adversarial attacks, which alter entire images almost imperceptibly, the DPSPSM is swift and offers the robustness of manipulating only a small number of input pixels. The presented algorithm quantizes the gradient update with a score generated for each pixel, incorporating contributions from each channel. The results show that the DPSPSM deceives the model with a success rate of 30.45% in the CIFAR-10 test set and 29.30% in the CIFAR-100 test set.

Introduction: This study introduces a fully differentiable, end-to-end audio transformation network designed to overcome these limitations by operating directly on acoustic features. Methods: The proposed method employs an encoder–decoder architecture with a global conditioning mechanism. It eliminates the need for parallel utterances, intermediate phonetic representations, and speaker-independent ASR systems. The system is evaluated on tasks of voice conversion and musical style transfer using subjective and objective metrics. Results: Experimental results demonstrate the model’s efficacy, achieving competitive performance in both seen and unseen target scenarios. The proposed framework outperforms seven existing systems for audio transformation and aligns closely with state-of-the-art methods. Conclusion: This approach simplifies feature engineering, ensures vocabulary independence, and broadens the applicability of audio transformations across diverse domains, such as personalized voice assistants and musical experimentation.

The first multi-bend achromat based synchrotron MAX IV operates two protein crystallography beamlines, BioMAX and MicroMAX. BioMAX is designed as a versatile, stable, high-throughput beamline catering for most protein crystallography experiments. MicroMAX is a more ambitious beamline dedicated to serial crystallography including time-resolved experiments. Both beamlines exploit the special characteristics of fourth-generation beamlines provided by the 3 GeV ring of MAX IV. In addition, the fragment-based drug discovery platform, FragMAX, is hosted and, at the FemtoMAX beamline, protein diffraction experiments exploring ultrafast time resolution can be performed. A technical and operational overview of the different beamlines and the platform is given as well as an outlook for protein crystallography embedded in the wider possibilities that MAX IV offers to users in the life sciences.

The MySteth is an intelligent medical tool designed for cardiac disease screening, utilizing either a stethoscope or smartphone to record heart sounds. Normal heart sounds in healthy individuals consist of \"lub\" and \"dub\" noises, while murmurs-additional sounds during heartbeats-can indicate cardiac anomalies such as valve dysfunctions and rapid blood flow, categorized as systolic or diastolic. MySteth was developed and tested using heart sounds recorded via smartphone and digital stethoscope. For ensuring the clinical validity of the data, all heart sound samples were meticulously annotated by human experts-super-specialized cardiologists with extensive experience in cardiac diagnostics. To achieve high classification accuracy, MySteth employs a hybrid CNN-LSTM model combined with Linear Predictive Coding (LPC) for preprocessing. The study involves classifying recorded heart sounds into normal heartbeats and murmurs, with murmurs further divided into systolic and diastolic categories. The tool demonstrated an accuracy of 92% in distinguishing normal heartbeats from murmurs, 91% in classifying murmurs into systolic and diastolic types, and 90% in further categorizing systolic murmurs into Ejection Systolic Murmurs (ESM) and Pansystolic Murmurs (PSM). MySteth is accessible and affordable, requiring minimal equipment, as most individuals already own a smartphone, and digital stethoscopes are commonly available. This ease of use facilitates both professional and home-based heart monitoring, especially beneficial in remote areas with limited healthcare access. MySteth is an at-home heart diagnostic tool that leverages deep learning to classify heart sounds into normal, ESM, PSM, and diastolic murmurs. Its user-friendly design and minimal hardware requirements ensure broad adoption across various healthcare settings, facilitating timely and accurate preliminary heart investigations. This capability is crucial in combating the global burden of cardiovascular diseases. MySteth's scalability and ease of deployment underscore its potential in early cardiovascular disease diagnosis, particularly in underserved regions, thereby promoting preventive healthcare.

Purpose: To report clinical characteristics, predisposing factors, and treatment outcome of Curvularia keratitis. Methods: Retrospective chart review of consecutive culture-proven Curvularia keratitis patients who presented to a tertiary eye care center in north India. Patients with mixed infections with Curvularia as one of the pathogens were also included. Standard case report form was developed to capture demographic information, clinical features, etiology, treatment, and outcome. Binary logistic regression was done to ascertain the effect of identified variables on final visual acuity. Results: Medical records of 97 patients of Curvularia keratitis were reviewed. Median age of patients was 45.3 years. Seventy-nine (79.4%) patients presented during the months of September to November. History of corneal trauma was present in 69.1%. Trauma from sugarcane leaf was identified in 66.1% of cases with corneal trauma with vegetative matter. Presenting visual acuity was worse than 20/60 in 57.8% of patients. Hypopyon and pigmented plaque-like infiltrate was present in 16.5% and 28.8% of patients, respectively. Mixed infection was reported in 14.4% of cases. Median time of antifungal therapy was 24.5 days. Surgical intervention was required in 18.5% cases. Of all, 11.1% patients achieved final VA of more than 20/200 who were managed surgically as compared to 68 (86%) patients who were managed medically. Younger age, absence of comorbidities, and lesser infiltrate size were found associated with good final visual acuity. Conclusion: Working males were most affected by Curvularia keratitis. Corneal trauma with sugarcane leave was the most common predisposing factor in the study area. Most of the cases presented with worse visual acuity but could be managed medically.