Wireless Mouth Motion Recognition System Based on EEG-EMG Sensors for Severe Speech Impairments

Sensors, Год журнала: 2024, Номер 24(13), С. 4125 - 4125

Опубликована: Июнь 25, 2024

This study aims to demonstrate the feasibility of using a new wireless electroencephalography (EEG)–electromyography (EMG) wearable approach generate characteristic EEG-EMG mixed patterns with mouth movements in order detect distinct movement for severe speech impairments. paper describes method detecting based on signal processing technology suitable sensor integration and machine learning applications. examines relationship between motion brainwave an effort develop nonverbal interfacing people who have lost ability communicate, such as paralysis. A set experiments were conducted assess efficacy proposed feature selection. It was determined that classification meaningful. signals also collected during silent mouthing phonemes. few-shot neural network trained classify phonemes from signals, yielding accuracy 95%. technique data collection bioelectrical phoneme recognition proves promising avenue future communication aids.

Язык: Английский

---

OxcarNet: Sinc convolutional network with temporal and channel attention for prediction of Oxcarbazepine monotherapy responses in patients with newly diagnosed epilepsy

Journal of Neural Engineering, Год журнала: 2024, Номер 21(5), С. 056019 - 056019

Опубликована: Сен. 9, 2024

Monotherapy with antiepileptic drugs (AEDs) is the preferred strategy for initial treatment of epilepsy. However, an inadequate response to initially prescribed AED a significant indicator poor long-term prognosis, emphasizing importance precise prediction outcomes regimen in patients

Язык: Английский

---

Construction 5.0 and Sustainable Neuro-Responsive Habitats: Integrating the Brain–Computer Interface and Building Information Modeling in Smart Residential Spaces

Sustainability, Год журнала: 2024, Номер 16(21), С. 9393 - 9393

Опубликована: Окт. 29, 2024

This study takes a unique approach by investigating the integration of Brain–Computer Interfaces (BCIs) and Building Information Modeling (BIM) within residential architecture. It explores their combined potential to foster neuro-responsive, sustainable environments framework Construction 5.0. The methodological involves real-time BCI data subjective evaluations occupants’ experiences elucidate cognitive emotional states. These inform BIM-driven alterations that facilitate adaptable, customized, sustainability-oriented architectural solutions. results highlight ability BCI–BIM create dynamic, occupant-responsive enhance well-being, promote energy efficiency, minimize environmental impact. primary contribution this work is demonstration viability neuro-responsive architecture, wherein input from enables modifications designs. technique enhances built environments’ flexibility user-centered quality integrating occupant preferences mental states into design process. Furthermore, BIM technologies has significant implications for advancing sustainability facilitating energy-efficient ecologically responsible areas. offers practical insights architects, engineers, construction professionals, providing method implementing systems user experience practices. research examines ethical issues concerning privacy, security, informed permission, ensuring these adhere moral legal requirements. underscores transformational while acknowledging challenges related interoperability, integrity, scalability. As result, ongoing innovation rigorous supervision are crucial effectively technologies. findings provide industry offering roadmap developing intelligent ethically sound

Язык: Английский

---

Unlocking Security for Comprehensive Electroencephalogram-Based User Authentication Systems

Sensors, Год журнала: 2024, Номер 24(24), С. 7919 - 7919

Опубликована: Дек. 11, 2024

With recent significant advancements in artificial intelligence, the necessity for more reliable recognition systems has rapidly increased to safeguard individual assets. The use of brain signals authentication gained substantial interest within scientific community over past decade. Most previous efforts have focused on identifying distinctive information electroencephalogram (EEG) recordings. In this study, an EEG-based user scheme is presented, employing a multi-layer perceptron feedforward neural network (MLP FFNN). utilizes P300 potentials derived from EEG signals, focusing user’s intent select specific characters. This approach involves two phases: identification and authentication. Both phases utilize recordings data preprocessing, database store manage these efficient retrieval organization, feature extraction using mutual (MI) selected segments, specifically targeting power spectral density (PSD) across five frequency bands. phase employs multi-class classifiers predict identity set enrolled users. associates predicted identities with labels probability assessments, verifying claimed as either genuine or impostor. combines segments mapping, confidence calculations, verification robust It also accommodates new users by transforming into vectors without need retraining. model extracts features identify classify input based authenticate user. experiments show that proposed can achieve 97% accuracy

Язык: Английский

---

Tuning VGG19 hyperparameters for improved pneumonia classification

THE SCIENTIFIC TEMPER, Год журнала: 2024, Номер 15(02), С. 2231 - 2237

Опубликована: Июнь 15, 2024

This research focuses on the classification of chest X-ray (CXR) images using powerful VGG19 convolutional neural network (CNN)architecture. The task involves distinguishing between various conditions present in images, with aim assisting medical professionals achieving accurate and efficient diagnoses. work explores use model for classifying CXR three optimization algorithms: Stochastic gradient descent momentum (SGDM), root mean square propagation (RMSprop), adaptive moment estimation (Adam). study investigates impact factors hyperparameter adjustments, including a learning rate (LR), mini-batch size (MBS) training epochs. Additionally, two dropout layers are introduced weight decay an L2 factor, data augmentation techniques applied activation functions. not only helps optimize image analysis but also offers valuable insights into comparative efficacy popular algorithms deep (DL) applications

Язык: Английский

---

Identifying EEG Biomarkers of Depression with Novel Explainable Deep Learning Architectures

Communications in computer and information science, Год журнала: 2024, Номер unknown, С. 102 - 124

Опубликована: Янв. 1, 2024

Язык: Английский

---

Classification of Known and Unknown Study Items in a Memory Task Using Single-Trial Event-Related Potentials and Convolutional Neural Networks

Brain Sciences, Год журнала: 2024, Номер 14(9), С. 860 - 860

Опубликована: Авг. 26, 2024

This study examines the feasibility of using event-related potentials (ERPs) obtained from electroencephalographic (EEG) recordings as biomarkers for long-term memory item classification. Previous studies have identified old/new effects in paradigms associated with explicit and familiarity. Recent advancements convolutional neural networks (CNNs) enabled classification ERP trials under different conditions identification features related to processes at single-trial level. We employed this approach compare three CNN models distinct architectures experimental data. Participants (N = 25) performed an association task while recording ERPs that were used training validation models. The EEGNET-based model achieved most reliable performance terms precision, recall, specificity compared shallow deep approaches. accuracy reached 62% known items 66% unknown items. Good overall requires a trade-off between recall depends on architecture dataset size. These results suggest possibility integrating into online learning tools identifying underlying memorization.

Язык: Английский

---

Joint intent detection and slot filling with syntactic and semantic features using multichannel CNN-BiLSTM

PeerJ Computer Science, Год журнала: 2024, Номер 10, С. e2346 - e2346

Опубликована: Окт. 8, 2024

Understanding spoken language is crucial for conversational agents, with intent detection and slot filling being the primary tasks in natural understanding (NLU). Enhancing NLU can lead to an accurate efficient virtual assistant thereby reducing need human intervention expanding their applicability other domains. Traditionally, these have been addressed individually, but recent studies highlighted interconnection, suggesting better results when solved together. Recent advances processing shown that pretrained word embeddings enhance text representation improve generalization capabilities of models. However, challenge poor joint learning models remains due limited annotated datasets. Additionally, traditional face difficulties capturing both semantic syntactic nuances language, which are vital filling. This study proposes a hybridized method using multichannel convolutional neural network three embedding channels: non-contextual information, part-of-speech (POS) tag features, contextual deeper understanding. Specifically, we utilized word2vec embeddings, one-hot vectors POS tags, bidirectional encoder representations from transformers (BERT) embeddings. These processed through layer shared long short-term memory (BiLSTM) network, followed by two softmax functions Experiments on air travel information system (ATIS) SNIPS datasets demonstrated our model significantly outperformed baseline models, achieving accuracy 97.90% F1-score 98.86% ATIS dataset, 98.88% 97.07% dataset. highlight effectiveness proposed approach advancing dialogue systems, paving way more real-world applications.

Язык: Английский

---

Detection of Elementary White Mucosal Lesions by an AI System: A Pilot Study

Oral, Год журнала: 2024, Номер 4(4), С. 557 - 566

Опубликована: Ноя. 14, 2024

Aim: Accurately identifying primary lesions in oral medicine, particularly elementary white lesions, is a significant challenge, especially for trainee dentists. This study aimed to develop and evaluate deep learning (DL) model the detection classification of mucosal (EWMLs) using clinical images. Materials Methods: A dataset was created by collecting photographs various including leukoplakia, OLP plaque-like reticular forms, OLL, candidiasis, hyperkeratotic from Unit Oral Medicine. The SentiSight.AI (Neurotechnology Co.®, Vilnius, Lithuania) AI platform used image labeling training. comprised 221 photos, divided into training (n = 179) validation 42) sets. Results: achieved an overall precision 77.2%, sensitivity 76.0%, F1 score 74.4%, mAP 82.3%. Specific classes, such as condyloma papilloma, demonstrated high performance, while others like leucoplakia showed room improvement. Conclusions: DL promising results detecting classifying EWMLs, with potential educational tools applications. Expanding incorporating diverse sources are essential improving accuracy generalizability.

Язык: Английский

---

DLT-GAN: Dual-Layer Transfer Generative Adversarial Network-Based Time Series Data Augmentation Method

Electronics, Год журнала: 2024, Номер 13(22), С. 4514 - 4514

Опубликована: Ноя. 18, 2024

In actual production processes, analysis and prediction tasks commonly rely on large amounts of time-series data. However, real-world scenarios often face issues such as insufficient or imbalanced data, severely impacting the accuracy predictions. To address this challenge, paper proposes a dual-layer transfer model based Generative Adversarial Networks (GANs) aiming to enhance training speed generation quality data augmentation under small-sample conditions while reducing reliance datasets. This method introduces module strategy traditional GAN framework which balances between discriminator generator, thereby improving model’s performance convergence speed. By employing network structure features signals, effectively reduces noise other irrelevant features, similarity generated signals’ characteristics. uses speech signals case study, addressing where are difficult collect limited number samples available for effective feature extraction analysis. Simulated timbre is conducted, experimental results CMU-ARCTIC database show that, compared methods, approach achieves significant improvements in enhancing consistency signal

Язык: Английский

---