Multiviewunet: A Deep Learning Surrogate for Wall Shear Stress Prediction in Aortic Aneurysmal Diseases DOI
Md. Ahasan Atick Faisal, Onur Mutlu, Sakib Mahmud

и другие.

Опубликована: Янв. 1, 2023

Computational Fluid Dynamics (CFD) analysis is widely used to simulate hemodynamics and investigate the biofluid mechanics of different tissue, whole organs, tissue–medical device interactions. However, CFD simulations are time-consuming computationally expensive; hence not readily available practical for patient-specific time-sensitive clinical applications prohibiting quick responses from clinicians. Disturbed known influence progression many cardiac conditions. Aorta main blood artery in body diseases this vessel very common. One such condition Abdominal Aortic Aneurysm (AAA), where abdominal aorta widens has risk rupture. Precise determination Wall Shear Stress (WSS) on aneurysmal wall essential assess rupture tissue. In study, we have proposed a Deep Learning (DL) surrogate estimating aortic WSS distribution. The DL model was created trained receive input output distributions directly, bypassing procedure. A novel way analyzing geometry-to-geometry problems also using domain transformation, which compatible with existing state-of-the-art Neural Networks (NN). framework, MultiViewUnet, 23 real 230 synthetic geometries. algorithm predicted stress an average Normalized Mean Absolute Error (NMAE) 0.362%. We believe our will open up new dimensions precise levels important.

Язык: Английский

Improving operations through a lean AI paradigm: a view to an AI-aided lean manufacturing via versatile convolutional neural network DOI
Mohammad Shahin,

Mazdak Maghanaki,

Ali Hosseinzadeh

и другие.

The International Journal of Advanced Manufacturing Technology, Год журнала: 2024, Номер 133(11-12), С. 5343 - 5419

Опубликована: Июль 2, 2024

Язык: Английский

Процитировано

9
The Transformative Trajectory of Artificial Intelligence in Education: The Two Decades of Bibliometric Retrospect DOI

K. Kavitha,

V. P. Joshith

Journal of Educational Technology Systems, Год журнала: 2024, Номер 52(3), С. 376 - 405

Опубликована: Март 1, 2024

Integrating artificial intelligence (AI) stands out as the most dynamic and innovative breakthrough in introducing disruptive paths varied domains of education. This bibliometric analysis delved into trajectory AI’s evolving landscape within educational settings over two decades, encompassing 324 articles published from 2003 to 2023, sourced Scopus database. The study uncovers a substantial surge publications with steep increase 2020, peaking 2023. Notably, while established nations like China US lead publications, notable contributions other developing countries, including Saudi Arabia, India, Malaysia, underscored global shift. Key terms, students, machine learning, AI higher education, underpin central focus research areas emerging themes “generative AI” chatbots “chatgpt” mark trends. Further, prompts sustained partnerships, interdisciplinary collaborations, continued exploration technologies catalyze advancements.

Язык: Английский

Процитировано

7
Rapid wall shear stress prediction for aortic aneurysms using deep learning: a fast alternative to CFD DOI Creative Commons
Md. Ahasan Atick Faisal, Onur Mutlu, Sakib Mahmud

и другие.

Medical & Biological Engineering & Computing, Год журнала: 2025, Номер unknown

Опубликована: Фев. 17, 2025

Abstract Aortic aneurysms pose a significant risk of rupture. Previous research has shown that areas exposed to low wall shear stress (WSS) are more prone Therefore, precise WSS determination on the aneurysm is crucial for rupture assessment. Computational fluid dynamics (CFD) powerful approach calculations, but they computationally intensive, hindering time-sensitive clinical decision-making. In this study, we propose deep learning (DL) surrogate, MultiViewUNet, rapidly predict time-averaged (TAWSS) distributions abdominal aortic (AAA). Our novel employs domain transformation technique translate complex geometries into representations compatible with state-of-the-art neural networks. MultiViewUNet was trained $$\varvec{23}$$ 23 real and $$\varvec{230}$$ 230 synthetic AAA geometries, demonstrating an average normalized mean absolute error (NMAE) just $$\varvec{0.362\%}$$ 0.362 % in prediction. This framework potential streamline hemodynamic analysis other scenarios where fast accurate quantification essential. Graphical abstract

Язык: Английский

Процитировано

0
Interpretable and ethical learning assessment transformer (IELAT): an explainable transformer model for personalized student assessments DOI Creative Commons

S. Hariharasitaraman,

Amudhavel Jayavel,

S. Gnanasekaran

и другие.

Cogent Education, Год журнала: 2025, Номер 12(1)

Опубликована: Апрель 8, 2025

Язык: Английский

Процитировано

0
Assessing english Language teachers’ pedagogical effectiveness using convolutional neural networks optimized by modified virus colony search algorithm DOI Creative Commons
Zhang Li

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Май 1, 2025

Язык: Английский

Процитировано

0
Smart Classrooms: How Sensors and AI Are Shaping Educational Paradigms DOI Creative Commons
Xiaochen Zhang,

Yiran Ding,

Xiaoyu Huang

и другие.

Sensors, Год журнала: 2024, Номер 24(17), С. 5487 - 5487

Опубликована: Авг. 24, 2024

The integration of advanced technologies is revolutionizing classrooms, significantly enhancing their intelligence, interactivity, and personalization. Central to this transformation are sensor technologies, which play pivotal roles. While numerous surveys summarize research progress in few studies focus on the AI developing smart classrooms. This systematic review classifies sensors used classrooms explores current applications from both hardware software perspectives. It delineates how different enhance educational outcomes crucial role play. highlights technology improves physical classroom environment, monitors physiological behavioral data, widely boost student engagements, manage attendance, provide personalized learning experiences. Additionally, it shows that combining algorithms with not only enhances data processing analysis efficiency but also expands capabilities, enriching article addresses challenges such as privacy protection, cost, algorithm optimization associated emerging proposing future directions advance technologies.

Язык: Английский

Процитировано

3
A Plug-in for Cognitive Diagnosis Method based on Correlation Representation under Long-tailed Distribution DOI
Yuhong Zhang, Tiancheng He, Shengyu Xu

и другие.

Expert Systems with Applications, Год журнала: 2025, Номер 284, С. 127952 - 127952

Опубликована: Апрель 29, 2025

Язык: Английский

Процитировано

0
University Education of the Future: Students’ Perspective DOI
Tatiana Dyukina, Iuliia V. Diukina, Т. В. Терентьева

и другие.

Lecture notes in networks and systems, Год журнала: 2025, Номер unknown, С. 165 - 177

Опубликована: Янв. 1, 2025

Язык: Английский

Процитировано

0
Mathematics intelligent tutoring systems with handwritten input: a scoping review DOI
Luiz Rodrigues, Filipe Dwan Pereira, Marcelo Marinho

и другие.

Education and Information Technologies, Год журнала: 2023, Номер 29(9), С. 11183 - 11209

Опубликована: Окт. 18, 2023

Язык: Английский

Процитировано

7
ImageLM: Interpretable image-based learner modelling for classifying learners’ computational thinking DOI Creative Commons
Danial Hooshyar, Yeongwook Yang

Expert Systems with Applications, Год журнала: 2023, Номер 238, С. 122283 - 122283

Опубликована: Окт. 21, 2023

Predictive learner modelling is crucial for personalized education. While convolutional neural networks (CNNs) have shown great success in education, their potential via image data unexplored. This research introduces a novel and interpretable approach Image-based Learner Modelling (ImageLM) using CNNs transfer learning to model learners' performance accordingly classify computational thinking solutions. The integrates Grad-CAM, enabling it provide insights into its decision-making process. Findings show that our custom CNN outperforms other models (namely ResNet, VGG, Inception), with 83% accuracy predicting solution correctness. More importantly, the ImageLM identifies regions contribute most predictions, shedding light on knowledge advancing toward trustworthy AI These results underline of utilizing imagery from activities during process predict performance, especially challenging environments like programming where traditional feature extraction might struggle.

Язык: Английский

Процитировано

6