A Wearable Eye-Tracking Approach for Early Autism Detection with Machine Learning: Unravelling Challenges and Opportunities

2022 International Joint Conference on Neural Networks (IJCNN), Journal Year: 2024, Volume and Issue: unknown, P. 1 - 8

Published: June 30, 2024

Language: Английский

---

Ambiguous facial expression detection for Autism Screening using enhanced YOLOv7-tiny model

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Nov. 18, 2024

Autism spectrum disorder is a developmental condition that affects the social and behavioral abilities of growing children. Early detection autism can help children to improve their cognitive quality life. The research in area reports it be detected from tests physical activities present on facial attributes Children with show ambiguous expressions which are different normal To detect images, this work presents an improvised variant YOLOv7-tiny model. presented model developed by integrating pyramid dilated convolutional layers feature extraction network Further, its recognition enhanced incorporating additional YOLO head. faces presence features drawing bounding boxes confidence scores. entire has been carried out self-annotated face dataset. achieved mAP value 79.56% was better than baseline state-of-the-art YOLOv8 Small

Language: Английский

---

Real Time Eye-Tracking Mouse Control Using Recurrent Neural Network

Published: Nov. 24, 2023

Machine Learning (ML) techniques, specifically Support Vector (SVM) and Extreme Gradient Boosting (XGBoost), were employed to achieve precise intuitive real-time eye tracking mouse control through computer vision. However, XGBoost may suffer from overfitting when dealing with a large number of features compared the training data size, or noisy imbalanced data. To address this issue, paper introduces Recurrent Neural Network (RNN), for in Human Computer Interaction (HCI). Gaze Direction Estimation (GDE) is initially estimate gaze direction, utilizing pupil positions camera calibration parameters. The estimated direction then used as input RNN eye-tracking HCI. experimental results shows that GDE-RNN has 26.03% 8.24% superior accuracy, 24.93% 86.76% better precision, 27.36% 8.75% high recall comparison SVM control.

Language: Английский

---

Computer Vision and Eyetracking Approach for Assessing Visual Disorders in Premature Infants

Published: June 26, 2024

Language: Английский

---

Amelioration of social impairments in autism: Possible role of vagal afferent stimulation in modification of the prefrontal-amygdala connectivity

Medical Hypotheses, Journal Year: 2024, Volume and Issue: unknown, P. 111486 - 111486

Published: Sept. 1, 2024

Language: Английский

---

EyeTrAES: Fine-grained, Low-Latency Eye Tracking via Adaptive Event Slicing

Proceedings of the ACM on Interactive Mobile Wearable and Ubiquitous Technologies, Journal Year: 2024, Volume and Issue: 8(4), P. 1 - 32

Published: Nov. 21, 2024

Eye-tracking technology has gained significant attention in recent years due to its wide range of applications human-computer interaction, virtual and augmented reality, wearable health. Traditional RGB camera-based eye-tracking systems often struggle with poor temporal resolution computational constraints, limiting their effectiveness capturing rapid eye movements. To address these limitations, we propose EyeTrAES, a novel approach using neuromorphic event cameras for high-fidelity tracking natural pupillary movement that shows kinematic variance. One EyeTrAES's highlights is the use adaptive windowing/slicing algorithm ensures just right amount descriptive asynchronous data accumulation within an frame, across patterns. EyeTrAES then applies lightweight image processing functions over accumulated frames from single perform pupil segmentation (as opposed gaze-based techniques require simultaneous both eyes). We show two boost fidelity by 6+%, achieving IoU~=92%, while incurring at least 3x lower latency than competing pure event-based alternatives [38]. additionally demonstrate microscopic motion captured exhibits distinctive variations individuals can thus serve as biometric fingerprint. For robust user authentication, train per-user Random Forest classifier feature vector short-term kinematics, comprising sliding window (location, velocity, acceleration) triples. Experimental studies different datasets (capturing environmental contexts) EyeTrAES-based authentication technique simultaneously achieve high accuracy (~=0.82) low (~=12ms), significantly outperform multiple state-of-the-art competitive baselines.

Language: Английский

---

Editorial: Improving autism spectrum disorder diagnosis using machine learning techniques

Frontiers in Neuroinformatics, Journal Year: 2024, Volume and Issue: 18

Published: Dec. 6, 2024

Autism Spectrum Disorder (ASD) is a complex neurodevelopmental condition characterised by challenges in social communication, repetitive behaviours, and restricted interests [1]. Early accurate diagnosis critical for effective intervention, enabling individuals with ASD to achieve better developmental outcomes an improved quality of life. However, traditional diagnostic methods, often reliant on subjective behavioural observations, remain timeintensive inconsistently accessible. This underscores urgent need innovative, scalable, objective tools [2,3].Machine Learning (ML) has emerged as transformative approach diagnosis, offering the ability analyse large, datasets uncover patterns that surpass human capability. For instance, eye-tracking technologies have been extensively utilised quantify gaze behaviours such fixations saccades, well-established markers autism. Studies employing Deep achieved high accuracy classifying from typically developing based data [3,7]. These technological advancements provide foundation are not only efficient but also potentially generalisable across diverse populations.Furthermore, approaches transforming scanpaths into visual representations classification simplified pipeline, automation traditionally laborious processes [4]. Additionally, unsupervised learning techniques, including clustering data, demonstrated potential extracting unique insights variability presentations [5]. developments illustrate growing synergy between AI-driven clinical practices.Beyond eye tracking, other modalities structural MRI (sMRI), resting-state functional connectivity (rsFC), multimodal integrating genetic, behavioural, imaging shown promise identifying robust biomarkers ASD. methodologies underscore importance leveraging multidimensional improve precision reliability [2,6]. Despite these promising innovations, persist. Standardisation methodologies, reproducibility results, translation research applicability significant barriers. special issue seeks address presenting cuttingedge integrates ML neuroinformatics enhance accuracy, efficiency, accessibility diagnostics. By bridging gap technology practice, this collection studies aims drive field toward more equitable solutions diagnosis.The articles included explore various aspects through ML, innovative findings:Eslami et al. comprehensive review models applied sMRI fMRI datasets, examining their efficacy diagnosing related disorders. The study highlights key deep architectures identifies limitations heterogeneity challenges. https://www.frontiersin.org/journals/neuroinformatics/articles/10.3389/fninf.2020.575999Vector Machine (SVM) models. Their uncovers discriminative within Default Mode Network (DMN), achieving reinforcing rsFC https://www.frontiersin.org/journals/neuroinformatics/articles/10.3389/fninf.2022.761942Jia conduct bibliometric analysis, mapping global landscape AI applications findings highlight trends rise feature selection significance integration, providing roadmap future studies. https://www.frontiersin.org/journals/neuroinformatics/articles/10.3389/fninf.2023.1310400Ruan present exploratory using micro-expressions biomarkers. posed video quality, work emphasises combining neuroimaging data. https://www.frontiersin.org/journals/neuroinformatics/articles/10.3389/fninf.2024.1435091The contributions emphasise nature holistic framework. Challenges lack standardisation ethical considerations algorithm deployment, interpretability relevant. integration advanced computational methods expertise opens avenues personalised treatment strategies early intervention protocols.We envision should focus on:• Data Diversity Multimodal Integration: Combining imaging, model robustness.• Interpretable AI: Developing transparent algorithms clinicians can trust use effectively.

Language: Английский

---

New eye tracking metrics system: the value in early diagnosis of autism spectrum disorder

Frontiers in Psychiatry, Journal Year: 2024, Volume and Issue: 15

Published: Dec. 11, 2024

Background Eye tracking (ET) is emerging as a promising early and objective screening method for autism spectrum disorders (ASD), but it requires more reliable metrics with enhanced sensitivity specificity clinical use. Methods This study introduces suite of novel ET metrics: Area Interest (AOI) Switch Counts (ASC), Favorable AOI Shifts (FAS) along self-determined pathways, Vacancy (AVC), applied to toddlers preschoolers diagnosed ASD. The correlation between these new Autism Diagnostic Observation Schedule, Second Edition (ADOS-2) scores via linear regression the cut-off were assessed predict diagnosis. Results Our findings indicate significantly lower FAS ASC higher AVC (P<0.05) in children ASD compared their non-ASD counterparts within this high-risk cohort; significance was not seen total fixation time neither pupil size (p > 0.05). Furthermore, negatively correlated ADOS-2 social affect (SA) subscale < Among metrics, yielded best 88-100% 80-88% cut off score 0.305-0.306, followed by separate from Conclusions confirms utility innovative metrics—FAS, AVC, ASC—which exhibit markedly improved specificity, enhancing diagnostic processes.

Language: Английский

---

Utilizing Artificial Intelligence to Support Autism Research

Studies in neuroscience, psychology and behavioral economics, Journal Year: 2024, Volume and Issue: unknown, P. 87 - 108

Published: Dec. 18, 2024

Language: Английский

---

A systematic evaluation of autism spectrum disorder identification with Scanpath Trend Analysis (STA)

Biomedical Signal Processing and Control, Journal Year: 2024, Volume and Issue: 103, P. 107414 - 107414

Published: Dec. 27, 2024

Language: Английский

---