Emotion recognition and artificial intelligence: A systematic review (2014–2023) and research recommendations

Information Fusion, Journal Year: 2023, Volume and Issue: 102, P. 102019 - 102019

Published: Sept. 16, 2023

Emotion recognition is the ability to precisely infer human emotions from numerous sources and modalities using questionnaires, physical signals, physiological signals. Recently, motion has gained attention because of its diverse application areas, like affective computing, healthcare, human–robot interactions, market research. This paper provides a comprehensive systematic review emotion techniques current decade. The includes Physical signals involve speech facial expression, while include electroencephalogram, electrocardiogram, galvanic skin response, eye tracking. an introduction various models, stimuli used for elicitation, background existing automated systems. covers searching scanning well-known datasets followed by design criteria review. After thorough analysis discussion, we selected 142 journal articles PRISMA guidelines. detailed studies available recognition. Our also presented potential challenges in literature directions future

Language: Английский

---

Multi-modality approaches for medical support systems: A systematic review of the last decade

Information Fusion, Journal Year: 2023, Volume and Issue: 103, P. 102134 - 102134

Published: Nov. 10, 2023

Healthcare traditionally relies on single-modality approaches, which limit the information available for medical decisions. However, advancements in technology and availability of diverse data sources have made it feasible to integrate multiple modalities gain a more comprehensive understanding patients' conditions. Multi-modality approaches involve fusing analyzing various types, including images, biosignals, clinical records, other relevant sources. This systematic review provides exploration multi-modality healthcare, with specific focus disease diagnosis prognosis. The adoption healthcare is crucial personalized medicine, as enables profile each patient, considering their genetic makeup, imaging characteristics, history, factors. also discusses technical challenges associated heterogeneous multimodal highlights emergence deep learning powerful paradigm integration.

Language: Английский

---

Adazd-Net: Automated adaptive and explainable Alzheimer’s disease detection system using EEG signals

Knowledge-Based Systems, Journal Year: 2023, Volume and Issue: 278, P. 110858 - 110858

Published: July 29, 2023

Alzheimer's disease (AZD) is a degenerative neurological condition that causes dementia and leads the brain to atrophy. Although AZD cannot be cured, early detection prompt treatment can slow down its progression. effectively identified via electroencephalogram (EEG) signals. But, it challenging analyze EEG signals since they change quickly spontaneously. Additionally, clinicians offer very little trust existing models due lack of explainability in predictions machine learning or deep models. The paper novel Adazd-Net which an adaptive explanatory framework for automated identification using We propose flexible analytic wavelet transform, automatically adjusts changes EEGs. optimum number features needed effective system performance also explored this work, along with discovery most discriminant channel. presents technique used explain both individual overall provided by classifier model. have obtained accuracy 99.85% detecting ten-fold cross-validation strategy. suggested precise explainable technique. Researchers investigate hidden information concerning during our proposed Our developed model employed hospital scenario detect AZD, as accurate robust.

Language: Английский

---

Zipper Pattern: An Investigation into Psychotic Criminal Detection Using EEG Signals

Diagnostics, Journal Year: 2025, Volume and Issue: 15(2), P. 154 - 154

Published: Jan. 11, 2025

Background: Electroencephalography (EEG) signal-based machine learning models are among the most cost-effective methods for information retrieval. In this context, we aimed to investigate cortical activities of psychotic criminal subjects by deploying an explainable feature engineering (XFE) model using EEG dataset. Methods: study, a new dataset was curated, containing signals from and control groups. To extract meaningful findings dataset, presented channel-based extraction function named Zipper Pattern (ZPat). The proposed ZPat extracts features analyzing relationships between channels. selection phase XFE model, iterative neighborhood component analysis (INCA) selector used choose distinctive features. classification phase, employed ensemble distance-based classifier achieve high performance. Therefore, t-algorithm-based k-nearest neighbors (tkNN) obtain results. Directed Lobish (DLob) symbolic language derive interpretable results identities selected vectors in final ZPat-based model. Results: leave-one-record-out (LORO) 10-fold cross-validation (CV) were used. achieved over 95% accuracy on curated Moreover, connectome diagram related detection created DLob-based artificial intelligence (XAI) method. Conclusions: regard, both performance interpretability. Thus, contributes engineering, psychiatry, neuroscience, forensic sciences. is one pioneering XAI investigating criminal/criminal individuals.

Language: Английский

---

EpilepsyNet: Novel automated detection of epilepsy using transformer model with EEG signals from 121 patient population

Computers in Biology and Medicine, Journal Year: 2023, Volume and Issue: 164, P. 107312 - 107312

Published: Aug. 5, 2023

Epilepsy is one of the most common neurological conditions globally, and fourth in United States. Recurrent non-provoked seizures characterize it have huge impacts on quality life financial for affected individuals. A rapid accurate diagnosis essential order to instigate monitor optimal treatments. There also a compelling need interpretation epilepsy due current scarcity neurologist diagnosticians global inequity access outcomes. Furthermore, existing clinical traditional machine learning diagnostic methods exhibit limitations, warranting create an automated system using deep model detection monitoring database.The EEG signals from 35 channels were used train learning-based transformer named (EpilepsyNet). For each training iteration, 1-min-long data randomly sampled participant. Thereafter, 5-s epoch was mapped matrix Pearson Correlation Coefficient (PCC), such that bottom part triangle discarded only upper vectorized as input data. PCC reliable method measure statistical relationship between two variables. Based 5 s data, single embedding performed thereafter generate 1-dimensional array signals. In final stage, positional encoding with learnable parameters added correlation coefficient's before being fed developed EpilepsyNet The ten-fold cross-validation technique model.Our transformer-based (EpilepsyNet) yielded high classification accuracy, sensitivity, specificity positive predictive values 85%, 82%, 87%, respectively.The proposed both robust since employed evaluate performance model. Compared models studies diagnosis, our simple less computationally intensive. This earliest study uniquely together model, database 121 participants detection. With validation larger dataset, same approach can be extended other conditions, transformative impact diagnostics worldwide.

Language: Английский

---

Deep neural network technique for automated detection of ADHD and CD using ECG signal

Computer Methods and Programs in Biomedicine, Journal Year: 2023, Volume and Issue: 241, P. 107775 - 107775

Published: Aug. 23, 2023

Attention Deficit Hyperactivity problem (ADHD) is a common neurodevelopment in children and adolescents that can lead to long-term challenges life outcomes if left untreated. Also, ADHD frequently associated with Conduct Disorder (CD), multiple research have found similarities clinical signs behavioral symptoms between both diseases, making differentiation ADHD, comorbid CD (ADHD+CD), subjective diagnosis. Therefore, the goal of this pilot study create first explainable deep learning (DL) model for objective ECG-based ADHD/CD diagnosis as having an biomarker may improve diagnostic accuracy.The dataset used consist ECG data collected from 45 62 ADHD+CD, 16 patients at Child Guidance Clinic Singapore. The were segmented into 2 s epochs directly train our 1-dimensional (1D) convolutional neural network (CNN) model.The proposed yielded 96.04% classification accuracy, 96.26% precision, 95.99% sensitivity, 96.11% F1-score. Gradient-weighted class activation mapping (Grad-CAM) function was also highlight important characteristics specific time points most impact score.In addition achieving performance results suggested DL method, Grad-CAM's implementation offers vital temporal clinicians other mental healthcare professionals use make wise medical judgments. We hope by conducting study, we will be able encourage larger-scale larger biosignal dataset. Hence allowing biosignal-based computer-aided (CAD) tools implemented ambulatory settings, easily obtained via wearable devices such smartwatches.

Language: Английский

---

Intuitive Human-Robot-Environment Interaction with EMG Signals: A Review

IEEE/CAA Journal of Automatica Sinica, Journal Year: 2024, Volume and Issue: 11(5), P. 1075 - 1091

Published: April 15, 2024

A long history has passed since electromyography (EMG) signals have been explored in human-centered robots for intuitive interaction. However, it still a gap between scientific research and real-life applications. Previous studies mainly focused on EMG decoding algorithms, leaving dynamic relationship the human, robot, uncertain environment scenarios seldomly concerned. To fill this gap, paper presents comprehensive review of EMG-based techniques human-robot-environment interaction (HREI) systems. The general processing framework is summarized, three paradigms, including direct control, sensory feedback, partial autonomous are introduced. intention treated as module proposed paradigms. Five key issues involving precision, stability, user attention, compliance, environmental awareness field discussed. Several important directions, decomposition, robust HREI dataset, proprioception reinforcement learning, embodied intelligence, to pave way future research. best what we know, first time that methods system summarized. It provides novel broader perspective improve practicability current myoelectric systems, which factors human-robot interaction, robot-environment state perception by human sensations considered, never done previous studies.

Language: Английский

---

Artificial intelligence in retinal screening using OCT images: A review of the last decade (2013–2023)

Computer Methods and Programs in Biomedicine, Journal Year: 2024, Volume and Issue: 254, P. 108253 - 108253

Published: May 28, 2024

Language: Английский

---

Evaluation of AI tools for healthcare networks at the cloud-edge interaction to diagnose autism in educational environments

Journal of Cloud Computing Advances Systems and Applications, Journal Year: 2024, Volume and Issue: 13(1)

Published: Feb. 9, 2024

Abstract Physical, social, and routine environments can be challenging for learners with autism spectrum disorder (ASD). ASD is a developmental caused by neurological problems. In schools educational environments, this may not only hinder child’s learning, but also lead to more crises mental convulsions. order teach students ASD, it essential understand the impact of their learning environment on interaction behavior. Different methods have been used diagnose in past, each own strengths weaknesses. Research into diagnostics has largely focused machine algorithms strategies rather than diagnostic methods. This article discusses many techniques literature, such as neuroimaging, speech recordings, facial features, EEG signals. led us conclude that settings, diagnosed cheaply, quickly, accurately through face analysis. To facilitate speed up processing information among children we applied AlexNet architecture designed edge computing. A fast method detecting disorders from settings using structure. While investigated variety methods, provide appropriate about disorder. addition, produce interpretive features. help who are suffering disease, key factors must considered: potential clinical therapeutic situations, efficiency, predictability, privacy protection, accuracy, cost-effectiveness, lack methodological intervention. The diseases troublesome, so they should identified treated.

Language: Английский

---

Responsible Artificial Intelligence for Mental Health Disorders: Current Applications and Future Challenges

Deleted Journal, Journal Year: 2025, Volume and Issue: 4(1)

Published: Jan. 1, 2025

Mental health disorders (MHDs) have significant medical and financial impacts on patients society. Despite the potential opportunities for artificial intelligence (AI) in mental field, there are no noticeable roles of these systems real environments. The main reason limitations is lack trust by domain experts decisions AI-based systems. Recently, trustworthy AI (TAI) guidelines been proposed to support building responsible (RAI) that robust, fair, transparent. This review aims investigate literature TAI machine learning (ML) deep (DL) architectures MHD domain. To best our knowledge, this first study analyzes trustworthiness ML DL models identifies advances RAI investigates how related current applicability We discover has severe compared other domains regarding standards implementations. discuss suggest possible future research directions could handle challenges.

Language: Английский

---