Article
Environmental Sciences
Juliana Aparecida Anochi, Vinicius Albuquerque de Almeida, Haroldo Fraga de Campos Velho
Summary: Machine learning models show promising performance in predicting precipitation in South America, producing predictions with errors under 2 mm in most regions. They also outperform traditional atmospheric models for certain regions, showing potential for improving forecast quality and reducing computational costs in operational centers.
Article
Computer Science, Artificial Intelligence
Mykola Galushka, Chris Swain, Fiona Browne, Maurice D. Mulvenna, Raymond Bond, Darren Gray
Summary: This study presents a new deep learning model for conducting preliminary screening of chemical compounds in-silico and accurately predicting their properties. This approach has the potential to provide a more efficient pathway for pharmaceutical companies to discover new medications.
NEURAL COMPUTING & APPLICATIONS
(2021)
Article
Agronomy
Dilip Kumar Roy, Tapash Kumar Sarkar, Sheikh Shamshul Alam Kamar, Torsha Goswami, Md Abdul Muktadir, Hussein M. Al-Ghobari, Abed Alataway, Ahmed Z. Dewidar, Ahmed A. El-Shafei, Mohamed A. Mattar
Summary: This study utilized deep learning models (LSTM and Bi-LSTM) for daily and multi-step forward forecasting of ET0, with results showing that the Bi-LSTM model outperformed other models.
Article
Oncology
Catharina Silvia Lisson, Christoph Gerhard Lisson, Marc Fabian Mezger, Daniel Wolf, Stefan Andreas Schmidt, Wolfgang M. Thaiss, Eugen Tausch, Ambros J. Beer, Stephan Stilgenbauer, Meinrad Beer, Michael Goetz
Summary: The study compares the performance of deep learning algorithms and radiomics-based machine learning models in predicting relapse of mantle cell lymphoma (MCL) based on baseline CT scans. The optimized 3D CNN model shows the best accuracy of 70% in predicting MCL relapse, offering potential for precision imaging in clinical management.
Article
Computer Science, Interdisciplinary Applications
Qilin Li, Zitong Wang, Ling Li, Hong Hao, Wensu Chen, Yanda Shao
Summary: Prediction of structural responses is crucial for analyzing structural behavior, but existing approaches have limitations. In this study, a novel machine learning approach based on graph neural networks is proposed, which can efficiently and accurately predict structural dynamics.
COMPUTERS & STRUCTURES
(2023)
Article
Computer Science, Information Systems
Zhangmiaoge Liu, Ning Wang, Chen Zhang, Zhouxiao Liu
Summary: This article proposes an efficient bioactivity prediction model based on ADMET properties, using Pearson correlation analysis, Bayesian regularization algorithm, and neural network to train and analyze bioactivity data. The model has high accuracy and fast prediction speed, and is of great significance for the diagnosis and treatment upgrade of breast cancer.
Article
Computer Science, Interdisciplinary Applications
Vishnu Preetham Revelli, Gauri Sharma, S. Kiruthika Devi
Summary: This project aims to extract text from braille text images and provide translated English text and audio output using a customized CNN model. The CNN model demonstrates robustness in image recognition and classification tasks, making it valuable for addressing challenges faced by visually impaired individuals.
ADVANCES IN ENGINEERING SOFTWARE
(2022)
Article
Ecology
Marcio Carneiro Brito Pache, Diego Andre Sant Ana, Joao Victor Araujo Rozales, Vanessa Aparecida de Moraes Weber, Adair da Silva Oliveira Junior, Vanir Garcia, Hemerson Pistori, Marco Hiroshi Naka
Summary: This study developed a system for predicting the body biomass of live fingerlings using a computer vision system, which proved to be more convenient and accurate compared to traditional invasive estimation methods. Results showed that using automatic frame selection based on Euclidean distance and data augmentation through rotation optimized the prediction of fish biomass.
ECOLOGICAL INFORMATICS
(2022)
Article
Oncology
Ioannis A. Vezakis, George I. Lambrou, George K. Matsopoulos
Summary: Osteosarcoma is a rare bone cancer that primarily affects children and adolescents. Despite available treatment options, the risk of recurrence and metastasis remains high. This study examines the potential of machine learning and artificial intelligence in improving disease prognosis and diagnosis through the evaluation of deep learning networks.
Article
Health Care Sciences & Services
Max Schmitt, Roman Christoph Maron, Achim Hekler, Albrecht Stenzinger, Axel Hauschild, Michael Weichenthal, Markus Tiemann, Dieter Krahl, Heinz Kutzner, Jochen Sven Utikal, Sebastian Haferkamp, Jakob Nikolas Kather, Frederick Klauschen, Eva Krieghoff-Henning, Stefan Froehling, Christof von Kalle, Titus Josef Brinker
Summary: This study examines the learnability of common hidden variables in digital pathology, indicating their potential to introduce batch effects that can impact the accuracy of AI-based classification systems.
JOURNAL OF MEDICAL INTERNET RESEARCH
(2021)
Article
Oncology
Yanhong Chen, Lijun Wang, Ran Luo, Shuang Wang, Heng Wang, Fei Gao, Dengbin Wang
Summary: This study investigates the value of a convolutional neural network (CNN) based on dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) in predicting the malignancy of breast lesions. The CNN model demonstrated high accuracy in predicting malignancy among breast lesions. Further validation in a larger and independent cohort is needed.
FRONTIERS IN ONCOLOGY
(2022)
Article
Biochemistry & Molecular Biology
Oliver Wieder, Melaine Kuenemann, Marcus Wieder, Thomas Seidel, Christophe Meyer, Sharon D. Bryant, Thierry Langer
Summary: This work introduces a novel GNN architecture, called D-GIN, which improves the accuracy of predicting molecular properties in drug discovery. By combining different sub-architectures and strategies, the model shows increased predictive power and addresses limitations in evaluating deep-learning models.
Article
Computer Science, Information Systems
Gongbo Liang, Mohammad Salem Atoum, Xin Xing, Izzat Alsmadi
Summary: Deep neural networks have achieved remarkable performance in classification tasks and applications. However, the training process is often slow and limited by computing resources due to the large model size and training dataset. To address this, distributed training using multiple devices is proposed. This work evaluates the performance of Microsoft DeepSpeed, a distributed training library, on image classification tasks, and finds that it offers limited benefits for simpler learning tasks but can significantly improve training speed and performance for more complex tasks.
CLUSTER COMPUTING-THE JOURNAL OF NETWORKS SOFTWARE TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Rasheed El-Bouri, David W. Eyre, Peter Watkinson, Tingting Zhu, David A. Clifton
Summary: The paper presents a deep learning method for predicting the admission location of emergency patients in a hospital, allowing for better resource planning and faster care provision. The model achieved success in predicting initial hospital admission locations with significant AUROC values, demonstrating its potential value in improving patient care and optimizing hospital resource allocation.
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
(2021)
Article
Cardiac & Cardiovascular Systems
Matthias Unterhuber, Karl-Patrik Kresoja, Karl-Philipp Rommel, Christian Besler, Andrea Baragetti, Nora Kloeting, Uta Ceglarek, Matthias Blueher, Markus Scholz, Alberico L. Catapano, Holger Thiele, Philipp Lurz
Summary: In patients with cardiovascular risk factors, machine-learning models based on proteomics outperform traditional regression models and clinical scores for predicting all-cause mortality.
JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY
(2021)
Article
Engineering, Electrical & Electronic
I-Hsi Kao, Wei-Jen Wang, Yi-Horng Lai, Jau-Woei Perng
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2019)
Article
Horticulture
I-Hsi Kao, Ya-Wen Hsu, Ya-Zhu Yang, Ya-Li Chen, Yi-Horng Lai, Jau-Woei Perng
SCIENTIA HORTICULTURAE
(2019)
Article
Energy & Fuels
Ya-Wen Hsu, Yi-Horng Lai, Kai-Quan Zhong, Tang-Kai Yin, Jau-Woei Perng
Article
Chemistry, Analytical
Chia-Chi Yang, Po-Ching Yang, Jia-Jin J. Chen, Yi-Horng Lai, Chia-Han Hu, Yung Chang, Shihfan Jack Tu, Lan-Yuen Guo
Article
Engineering, Electrical & Electronic
I-Hsi Kao, Ya-Wen Hsu, Yi Horng Lai, Jau-Woei Perng
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2020)
Article
Environmental Sciences
Yi-Horng Lai, Ai-Yi Wang, Chia-Chi Yang, Lan-Yuen Guo
INTERNATIONAL JOURNAL OF ENVIRONMENTAL RESEARCH AND PUBLIC HEALTH
(2020)
Article
Emergency Medicine
Yuan-Heng Su, Kuan-Han Wu, Chih-Min Su, Chi-Yung Cheng, Cheng- Cheng, Chia-Te Kung, Fu-Cheng Chen
Summary: The outbreak of the new coronavirus disease 2019 has significantly impacted the global medical system and health-seeking behavior of people, affecting the management of patients with ST-elevation myocardial infarction. A decrease in the percentage of patients meeting the door-to-balloon time criteria was observed for patients who directly visited the emergency department during the pandemic, indicating differences in the clinical features of STEMI patients post-COVID-19.
EMERGENCY MEDICINE INTERNATIONAL
(2021)
Article
Chemistry, Analytical
Chiao-Sheng Wang, I-Hsi Kao, Jau-Woei Perng
Summary: This paper proposes a one-dimensional convolutional neural network model for the diagnosis of permanent magnet synchronous motors, which can extract features under various conditions and effectively diagnose motor states with an accuracy of 98.85%.
Article
Materials Science, Multidisciplinary
I-Hsi Kao, Jau-Woei Perng
Summary: This study utilized a convolutional autoencoder to predict the spread of COVID-19 in the contiguous United States, with results showing that the model with long short-term memory outperformed the one without it in terms of predictive performance.
RESULTS IN PHYSICS
(2021)
Article
Engineering, Electrical & Electronic
I-Hsi Kao, Ching-Yao Chan
Summary: This study conducted a comparative experiment on machine learning-based pedestrian trajectory prediction and found that incorporating social and posture features can improve prediction accuracy. Deep learning methods perform better in handling complex features, especially the 3D CNN model that combines social and posture features.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2022)
Article
Medicine, General & Internal
Chia-Te Kung, Chih-Min Su, Sheng-Yuan Hsiao, Fu-Cheng Chen, Yun-Ru Lai, Chih-Cheng Huang, Cheng-Hsien Lu
Summary: Increased levels of soluble triggering receptor expressed on myeloid cells 1 (sTREM-1) were found in early stages of sepsis and were associated with therapeutic outcomes. sTREM-1 levels correlate with biomarkers for endothelial dysfunction and clinical severity, serving as a valuable auxiliary diagnostic and prognostic marker for sepsis.
Article
Health Care Sciences & Services
Hung-Chen Wang, Pei-Ming Wang, Yu-Tsai Lin, Nai-Wen Tsai, Yun-Ru Lai, Chia-Te Kung, Chih-Min Su, Cheng-Hsien Lu
Summary: The study showed that HBOT can improve the serum oxidative stress in TBI patients, with markers potentially becoming part of clinical evaluation and treatment. Further large-scale studies may be needed to confirm these findings.
JOURNAL OF PERSONALIZED MEDICINE
(2021)
Article
Chemistry, Analytical
I-Hsi Kao, Ching-Yao Chan
Summary: This study aims to recognize drowsiness through human facial features and analyze the attention of neurons in the learning model. The results show that using eye images alone yields better results and the effect of Grad-CAM is more reasonable. Furthermore, the proposed feature analysis method KNN-Sigma is effective in estimating the homogeneous concentration and heterogeneous separation of extracted features. The fusion of face and eye signals gives the best results for recognition accuracy and KNN-Sigma.
Article
Engineering, Electrical & Electronic
Chiao-Sheng Wang, I-Hsi Kao, Ya-Wen Hsu, Tsung-Chun Lin, Der-Min Tsay, Jau-Woei Perng
Summary: This study analyzes the quality of a new tapping machine by measuring torque and speed. It finds that tapping torque magnitude is crucial for thread quality, and determines appropriate operational boundaries using regression trees and 1-D CNN methods. The study shows improved thread quality probability, high accuracy classification, and the ability to detect important signal areas effectively.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2021)