Article
Environmental Sciences
Clyvihk Renna Camacho, Augusto Getirana, Otto Correa Rotunno Filho, Maria Antonieta A. Mourao
Summary: In this study, the authors developed an artificial intelligence-based approach to monitor major Brazilian aquifers by combining ground-based hydrogeological measurements and Gravity Recovery and Climate Experiment (GRACE) data. Three different AI approaches were tested and the approach was further enhanced by wavelet and seasonal decomposition processes applied to GRACE data. The results showed that the proposed approach outperformed the Global Land Data Assimilation System (GLDAS) in reproducing groundwater storage (GWS) change in all studied Brazilian aquifers. By using the AI model outputs, the study estimated the GWS change of two major aquifers over the past two decades. Water loss in these aquifers was driven by a prolonged drought and increased groundwater pumping for irrigation. The study demonstrates the cost-effectiveness and potential global replicability of combining satellite data and AI for monitoring poorly equipped aquifers at a continental scale.
WATER RESOURCES RESEARCH
(2023)
Article
Microbiology
Caitlin M. A. Simopoulos, Zhibin Ning, Leyuan Li, Mona M. Khamis, Xu Zhang, Mathieu Lavallee-Adam, Daniel Figeys
Summary: Metaproteomics is a useful tool for studying microbial communities, but the data acquisition process can be time-consuming and resource-intensive. In this study, the researchers developed a computational framework called MetaProClust-MS1 to prioritize samples for follow-up analysis using tandem mass spectrometry (MS/MS). The framework successfully identified microbial responses and disease diagnostic features in gut microbiome data. The study also demonstrated the potential of MetaProClust-MS1 in clinical settings and large-scale metaproteomic screening.
Article
Geosciences, Multidisciplinary
Shaheryar Ahmed, Andres Abarca, Daniele Perrone, Ricardo Monteiro
Summary: This study presents an RVS procedure for large-scale seismic vulnerability assessment and applies it to a case study province in Northern Algeria. Using data collection and parameter study, the method estimates the seismic vulnerability of buildings and assists decision-makers in planning earthquake risk reduction strategies.
INTERNATIONAL JOURNAL OF DISASTER RISK REDUCTION
(2022)
Article
Optics
Carlo M. Valensise, Ivana Grecco, Davide Perangeli, Laudio Conti
Summary: This study presents an optical processor with a capacity exceeding any previous implementation, enabling large-scale text encoding and classification. It offers a solution to scale up light-driven computing and opens the route to photonic natural language processing.
PHOTONICS RESEARCH
(2022)
Article
Nanoscience & Nanotechnology
Sauradeep Majumdar, Seyed Mohamad Moosavi, Kevin Maik Jablonka, Daniele Ongari, Berend Smit
Summary: This study developed a database of around 20,000 hypothetical MOFs, visualizing and quantifying their diversity using machine learning techniques. The addition of these structures improved the overall diversity metrics of the databases, especially in terms of the chemistry of metal nodes. Evaluations using grand-canonical Monte Carlo simulations showed that many of these diverse structures outperformed benchmark materials in post-combustion carbon capture and hydrogen storage applications.
ACS APPLIED MATERIALS & INTERFACES
(2021)
Article
Computer Science, Artificial Intelligence
Leran Chen, Ping Ji, Yongsheng Ma, Yiming Rong, Jingzheng Ren
Summary: In this study, a novel algorithm for heart disease screening is introduced, aiming to improve model accuracy through customization and integration of patient data. The evaluation on two heart disease datasets showed significant results and outperformed traditional machine learning algorithms in terms of medical ethics and operability. This algorithm provides a powerful tool for large-scale disease screening and has the potential to save lives and reduce the economic burden of heart disease.
ARTIFICIAL INTELLIGENCE IN MEDICINE
(2023)
Article
Computer Science, Software Engineering
Alper Tufek, Mehmet S. Aktas
Summary: This study proposes a methodology for tracking and analyzing provenance in numerical weather prediction models, specifically the WRF model. By utilizing a machine learning-based parser, provenance information can be extracted and represented as provenance graphs, enabling easy management and understanding of the weather forecast workflows.
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
(2023)
Article
Computer Science, Information Systems
Yiyang Yang, Sucheng Deng, Juan Lu, Yuhong Li, Zhiguo Gong, Leong U. Hou, Zhifeng Hao
Summary: This paper proposes a framework called GraphLSHC to address the scalability issue faced by large-scale hypergraph spectral clustering. The framework expands the hypergraph into a general format, partitions vertices and hyperedges simultaneously to reduce computational complexity, and improves performance through hyperedge-based sampling techniques. Numerous experiments demonstrate the superiority of the proposed framework over existing algorithms.
INFORMATION SCIENCES
(2021)
Article
Biochemistry & Molecular Biology
Thomas Mortier, Anneleen D. Wieme, Peter Vandamme, Willem Waegeman
Summary: This study presents benchmarking results on an unprecedented scale for a wide range of machine learning methods using large datasets, showing acceptable identification rates in all three scenarios, but typically lower than reported in previous studies. The research also demonstrates that taxonomic information is generally not well preserved in MALDI-TOF mass spectrometry data. Additionally, for the novel species scenario, neural networks with Monte Carlo dropout are successfully applied for the first time.
COMPUTATIONAL AND STRUCTURAL BIOTECHNOLOGY JOURNAL
(2021)
Article
Computer Science, Artificial Intelligence
Meng Wang, Weijie Fu, Xiangnan He, Shijie Hao, Xindong Wu
Summary: This paper provides a systematic survey on existing Large-scale Machine Learning (LML) methods and offers a blueprint for future developments in this area. The methods are divided into three categories based on ways of improving scalability, and are further categorized according to targeted scenarios. Representative methods and their limitations are discussed, along with potential directions for future research and open issues to be addressed.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2022)
Article
Biochemistry & Molecular Biology
Yizhen Situ, Xueying Yuan, Xiangning Bai, Shuhua Li, Hong Liang, Xin Zhu, Bangfen Wang, Zhiwei Qiao
Summary: This study calculates and analyzes the permeability performance of metal-organic frameworks (MOFs) membranes for capturing CO2, and explores the relationship between structural descriptors and permeance performance. It provides explicit directions and guidelines for membrane separation research in flue gas purification.
Article
Computer Science, Information Systems
Lena Oden
Summary: The use of Deep Learning methods is seen as a key opportunity for processing large-scale scientific datasets, but efficient processing requires hierarchical storage architectures for faster access to frequently used data. Different staging techniques are evaluated for Deep Learning usecases, with DRAM staging or usecase specific staging techniques showing the best performance, while a technique called split staging provides improved performance compared to non-staged usecases and comparable performance to specialized solutions. Performance often depends more on data layout and transformations used than on storage layer bandwidth.
IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTING
(2021)
Article
Clinical Neurology
Sowmya M. Ramaswamy, Maud A. S. Weerink, Michel M. R. F. Struys, Sunil B. Nagaraj
Summary: The study aims to investigate whether dexmedetomidine-induced deep sedation mimics natural sleep patterns using large-scale EEG recordings and machine learning techniques. The random forest algorithm trained on non-rapid eye movement stage 3 (N3) EEG patterns predicted dexmedetomidine-induced deep sedation state with high accuracy, outperforming other machine learning models. Power in the delta band, theta band, and beta band were identified as important features for prediction.
Article
Engineering, Environmental
Min Cheng, Zhiyuan Zhang, Shihui Wang, Kexin Bi, Kong-qiu Hu, Zhongde Dai, Yiyang Dai, Chong Liu, Li Zhou, Xu Ji, Wei-qun Shi
Summary: We conducted large-scale molecular simulations to identify metal-organic framework materials for capturing gaseous iodine in the context of managing various radionuclides in spent nuclear fuel reprocessing. By utilizing the computation-ready experimental (CoRE) metal-organic frameworks (MOFs) database and grand canonical Monte Carlo simulation, we predicted the iodine uptake capabilities of MOFs, generated a ranking list based on their performance, and visualized the adsorption sites of the top 10 candidates. Furthermore, machine learning was used to establish structure-property relationships and design rules for effective MOF adsorbents. This research can contribute to the development of high-performing adsorbents for capturing and recovering radioactive iodine and other volatile hazardous species.
FRONTIERS OF ENVIRONMENTAL SCIENCE & ENGINEERING
(2023)
Article
Chemistry, Medicinal
Jin Zhang, Longbing Yang, Zhuqing Tian, Wenjing Zhao, Chaoqin Sun, Lijuan Zhu, Mingjiao Huang, Guo Guo, Guiyou Liang
Summary: An effective QSAR protocol for screening antifungal peptides was established in this study, which integrated a classification model and four activity prediction models, resulting in the identification of three outstanding peptides from over three million candidates. The screening process took only a few days, much faster than traditional experimental screenings, and proved to be useful in reducing repetitive laboratory efforts in antifungal peptide discovery.
ACS MEDICINAL CHEMISTRY LETTERS
(2022)
Article
Computer Science, Artificial Intelligence
Ioannis Tsamardinos, Giorgos Borboudakis, Pavlos Katsogridakis, Polyvios Pratikakis, Vassilis Christophides
Article
Biochemical Research Methods
Georgios Papoutsoglou, Vincenzo Lagani, Angelika Schmidt, Konstantinos Tsirlis, David-Gomez Cabrero, Jesper Tegner, Ioannis Tsamardinos
Article
Biochemical Research Methods
Kleio-Maria Verrou, Ioannis Tsamardinos, Georgios Papoutsoglou
Article
Medical Informatics
Karen-Inge Karstoft, Ioannis Tsamardinos, Kasper Eskelund, Soren Bo Andersen, Lars Ravnborg Nissen
JMIR MEDICAL INFORMATICS
(2020)
Article
Medicine, General & Internal
Makrina Karaglani, Krystallia Gourlia, Ioannis Tsamardinos, Ekaterini Chatzaki
JOURNAL OF CLINICAL MEDICINE
(2020)
Article
Computer Science, Artificial Intelligence
Giorgos Borboudakis, Ioannis Tsamardinos
Summary: This study introduces a strategy for identifying multiple solutions efficiently and proposes a taxonomy of features that considers the existence of multiple solutions. Experimental results show that the proposed algorithm is significantly more computationally efficient than alternative approaches while achieving similar predictive performance.
DATA MINING AND KNOWLEDGE DISCOVERY
(2021)
Review
Oncology
John L. Marshall, Beth N. Peshkin, Takayuki Yoshino, Jakob Vowinckel, Havard E. Danielsen, Gerry Melino, Ioannis Tsamardinos, Christian Haudenschild, David J. Kerr, Carlos Sampaio, Sun Young Rha, Kevin T. FitzGerald, Eric C. Holland, David Gallagher, Jesus Garcia-Foncillas, Hartmut Juhl
Summary: This paper discusses the development and importance of multiomics in clinical practice, highlighting the knowledge gap within the medical community and urging physicians to actively adapt and prepare for this new era.
Article
Medicine, General & Internal
Makrina Karaglani, Maria Panagopoulou, Christina Cheimonidi, Ioannis Tsamardinos, Efstratios Maltezos, Nikolaos Papanas, Dimitrios Papazoglou, George Mastorakos, Ekaterini Chatzaki
Summary: This study found that ccfDNA can be used as a minimally invasive biomaterial for the early diagnosis and monitoring of T2DM. The methylation levels of T2DM-related genes in ccfDNA were different, and the established biosignatures showed high discriminatory performance.
JOURNAL OF CLINICAL MEDICINE
(2022)
Article
Computer Science, Artificial Intelligence
Ioulia Karagiannaki, Krystallia Gourlia, Vincenzo Lagani, Yannis Pantazis, Ioannis Tsamardinos
Summary: This article introduces a novel dimensionality reduction algorithm called Pathway Activity Score Learning (PASL), which constructs features corresponding to molecular pathways and has a straightforward biological interpretation. Experimental results demonstrate that PASL outperforms other methods in predictive performance, and a universal feature dictionary across different tissues and pathologies is constructed.
Article
Computer Science, Artificial Intelligence
Vasilios Plakandaras, Periklis Gogas, Theophilos Papadimitriou, Ioannis Tsamardinos
Summary: This study applies the Just-Add-Data (JAD) system to automatically select machine learning algorithms and provide a fast prediction model for credit card fraud detection. The model selected by JAD performs well in the test sample and matches existing literature.
APPLIED ARTIFICIAL INTELLIGENCE
(2022)
Article
Oncology
Ioannis Tsamardinos, Paulos Charonyktakis, Georgios Papoutsoglou, Giorgos Borboudakis, Kleanthi Lakiotaki, Jean Claude Zenklusen, Hartmut Juhl, Ekaterini Chatzaki, Vincenzo Lagani
Summary: JADBio is an AutoML platform for low-sample, high-dimensional omics data in translational medicine and bioinformatics. It focuses on predictive modeling, feature selection, and identifying biosignatures for knowledge discovery. Compared to other machine learning libraries, JADBio achieves competitive predictive performance with a small number of features.
NPJ PRECISION ONCOLOGY
(2022)
Article
Multidisciplinary Sciences
Scott Bowler, Georgios Papoutsoglou, Aristides Karanikas, Ioannis Tsamardinos, Michael J. Corley, Lishomwa C. Ndhlovu
Summary: This study utilized machine learning to identify DNA methylation signatures associated with severe COVID-19 disease, providing a potential method for early identification of high-risk individuals. The findings contribute to our understanding of the utility of DNA methylation in COVID-19 disease pathology and serve as a platform for future COVID-19 related studies.
SCIENTIFIC REPORTS
(2022)
Review
Computer Science, Artificial Intelligence
Ioannis Tsamardinos
Summary: In a predictive modeling task, it is unacceptable to lose samples for estimation when data size is low. Instead, we need to estimate the performance of the pipeline that produces the model by training it on subsets of the samples. This avoids losing data to estimation and provides a more accurate estimate of the model's performance.
Article
Computer Science, Artificial Intelligence
Konstantina Biza, Ioannis Tsamardinos, Sofia Triantafillou
Summary: Causal discovery algorithms require a set of hyperparameters and selecting the optimal combination is a challenge for practitioners. This study proposes an out-of-sample causal tuning method that treats causal models as predictive models and uses out-of-sample protocols. The method can handle general settings and is evaluated against other tuning approaches, showing its effectiveness in causal discovery.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2022)
Article
Mathematical & Computational Biology
Kleanthi Lakiotaki, George Georgakopoulos, Elias Castanas, Oluf Dimitri Roe, Giorgos Borboudakis, Ioannis Tsamardinos
NPJ SYSTEMS BIOLOGY AND APPLICATIONS
(2019)
