TabNet Based Prediction Model for ICU admission in Covid-19 patients DOI

Djihane Houfani,

Sihem Slatnia,

Okba Kazar

et al.

Published: Dec. 7, 2022

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the responsible virus for disease 2019 (COVID-19). It was reported first time in Wuhan (China) by late December 2019. The COVID-19 pandemic has become a global health risk due to urgent need an Intensive Care Unit (ICU) that exceeded its capacity. To cope with this exponential spread fast adoption of Artificial Intelligence (AI) tools and advanced technology crucial. For reason, many research works AI are conducted. In current paper, we intend report applications solutions based on machine learning, deep data mining algorithms detecting, predicting, diagnosing COVID-19. Furthermore, study aims develop new learning-based method capable predicting whether patient requires admission intensive care unit using clinical tabular from Kaggle. This model will contribute optimization ICU resources. experimental results showed combining Synthetic Minority Oversampling Technique (SMOTE) TabNet classifier improved prediction performance surpassed state-of-the-art models: MLP, RF, LR, KNN.

Language: Английский

Forecasting COVID-19 new cases using deep learning methods DOI Creative Commons
Lu Xu, Rishikesh Magar, Amir Barati Farimani

et al.

Computers in Biology and Medicine, Journal Year: 2022, Volume and Issue: 144, P. 105342 - 105342

Published: Feb. 23, 2022

After nearly two years since the first identification of SARS-CoV-2 virus, surge in cases because virus mutations is a cause grave public health concern across globe. As result this crisis, predicting transmission pattern one most vital tasks for preparing and controlling pandemic. In addition to mathematical models, machine learning tools, especially deep models have been developed forecasting trend number patients affected by with great success. paper, three including CNN, LSTM, CNN-LSTM predict COVID-19 Brazil, India Russia. We also compare performance our previously notice significant improvements prediction performance. Although used only these countries, can be easily applied datasets other countries. Among work, LSTM model has highest when shows an improvement accuracy compared some existing models. The research will enable accurate support global fight against

Language: Английский

Citations

77
Investigating the impact of the COVID-19 pandemic on crime incidents number in different cities DOI Creative Commons
Miaomiao Hou,

Zhaolong Zeng,

Xiaofeng Hu

et al.

Journal of Safety Science and Resilience, Journal Year: 2022, Volume and Issue: 3(4), P. 340 - 352

Published: Feb. 17, 2022

The COVID-19 pandemic is strongly affecting many aspects of human life and society around the world. To investigate whether this also influences crime, differences in crime incidents numbers before during four large cities (namely Washington DC, Chicago, New York City Los Angeles) are investigated. Moreover, Granger causal relationships between incident new cases examined. Based on that, with significant correlations used to improve prediction performance. results show that generally impacted by pandemic, but it varies different types. Most types crimes have seen fewer than before. Several found these cities. More specifically, theft Chicago City, fraud DC Angeles, assault robbery Angeles significantly caused case COVID-19. These may be partially explained Routine Activity theory Opportunity people prefer stay at home avoid being infected giving chances for crimes. In addition, involving as a variable can slightly performance terms some specific crime. This study expected obtain deeper insights into cities, provide attempts pandemic.

Language: Английский

Citations

20
Geospatial Analysis and Mapping Strategies for Fine-Grained and Detailed COVID-19 Data with GIS DOI Creative Commons
Ángel Miramontes Carballada, José Balsa‐Barreiro

ISPRS International Journal of Geo-Information, Journal Year: 2021, Volume and Issue: 10(9), P. 602 - 602

Published: Sept. 12, 2021

The unprecedented COVID-19 pandemic is showing dramatic impact across the world. Public health authorities attempt to fight against virus while maintaining economic activity. In face of uncertainty derived from virus, all countries have adopted non-pharmaceutical interventions for limiting mobility and social distancing. order support these interventions, some governments opted sharing very fine-grained data related with in their territories. Geographical science playing a major role terms understanding how spreads regions. Location cases allows identifying spatial patterns traced by virus. Understanding makes controlling spread feasible, minimizes its vulnerable regions, anticipates potential outbreaks, or elaborates predictive risk maps. application geospatial analysis must be urgently optimal decision making real near-real time. However, aspects process map sensitive emergency not yet been sufficiently explored. Among them include concerns about datasets information shown depending on aggregation, scaling, privacy issues, need know advance particularities study area. this paper, we introduce our experience mapping incidence during first wave region Galicia (NW Spain), after that discuss mentioned aspects.

Language: Английский

Citations

23
Advancing Geographic Information Systems With Machine Learning DOI
E. Ivette Cota-Rivera

IGI Global eBooks, Journal Year: 2025, Volume and Issue: unknown, P. 253 - 270

Published: March 6, 2025

A Geographic Information System (GIS) is a technological tool that allows for the capture, storage, analysis, and visualization of geographically referenced data. These systems integrate various forms spatial non-spatial data, facilitating analysis geographic phenomena patterns.The integration Machine Learning (ML) into Systems has revolutionized way geospatial data analyzed used. Learning, with its ability to learn from large volumes make accurate predictions, complements analytical capabilities GIS, allowing extraction complex patterns performance advanced predictions were not previously possible. The purpose this chapter explore applications empowered by use machine learning, highlighting their impact on environmental management.

Language: Английский

Citations

0
Merging GIS and Machine Learning Techniques: A Paper Review DOI Open Access

Chikodinaka Vanessa Ekeanyanwu,

Inioluwa Obisakin,

Precious Aduwenye

et al.

Journal of Geoscience and Environment Protection, Journal Year: 2022, Volume and Issue: 10(09), P. 61 - 83

Published: Jan. 1, 2022

GIS (Geographic Information Systems) data showcase locations of earth observations or features, their associated attributes and spatial relationships that exist between such observations. Analysis varies widely may include some modeling predictions which are usually computing-intensive complicated, especially, when large datasets involved. With advancement in computing technologies, techniques as Machine learning (ML) being suggested a potential game changer the analysis because comparative speed, accuracy, automation, repeatability. Perhaps, greatest benefit using both ML is ability to transfer results from one database another. tools have been used extensively medicine, urban development, environmental landslide susceptibility prediction (LSP). There also problem loss during conversion systems while geotechnical areas erosion flood prediction, lack variability soil has limited use techniques. This paper gives an overview current methods incorporated into obtained for LSP, health, development. The Supervised Learning (SML) algorithms decision trees, SVM, KNN, perceptron including Unsupervised k-means, elbow algorithms, hierarchal algorithm discussed. Their benefits, well shortcomings studied by several researchers elucidated this review. Finally, review discusses future optimization

Language: Английский

Citations

7
MGLEP: Multimodal Graph Learning for Modeling Emerging Pandemics with Big Data DOI Creative Commons

Khanh-Tung Tran,

Truong Son Hy, Lili Jiang

et al.

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

Published: July 16, 2024

Abstract Accurate forecasting and analysis of emerging pandemics play a crucial role in effective public health management decision-making. Traditional approaches primarily rely on epidemiological data, overlooking other valuable sources information that could act as sensors or indicators pandemic patterns. In this paper, we propose novel framework, MGLEP, integrates temporal graph neural networks multi-modal data for learning forecasting. We incorporate big sources, including social media content, by utilizing specific pre-trained language models discovering the underlying structure among users. This integration provides rich dynamics through with networks. Extensive experiments demonstrate effectiveness our framework analysis, outperforming baseline methods across different areas, situations, prediction horizons. The fusion enables comprehensive understanding landscape less time lag, cheap cost, more potential indicators.

Language: Английский

Citations

1
Grasping Emergency Dynamics: A Review of Group Evacuation Techniques and Strategies in Major Emergencies DOI Creative Commons
Hai Sun, Guorui Han, Xiaowei Zhang

et al.

Journal of Safety Science and Resilience, Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 1, 2024

Language: Английский

Citations

1
A Hybrid LSTM and MLP Scheme for COVID-19 Prediction: A Case Study in Thailand DOI Creative Commons

Peerapol Kompunt,

Suparat Yongjoh,

Phet Aimtongkham

et al.

Trends in Sciences, Journal Year: 2023, Volume and Issue: 20(10), P. 6884 - 6884

Published: Aug. 1, 2023

After the COVID-19 epidemic, Thailand was affected in a variety of ways, with most obvious being economic downturn and huge impact on health, including loss medical human resources to combat epidemic. However, still lacks analysis prediction tools required prepare for future epidemic situations. Therefore, we present development models predicting spread In particular, application long short-term memory (LSTM) multilayer perceptron (MLP) model investigated predict new cases, total deaths, deaths. There are 77 provinces Thailand. The data used this trial were obtained from Department Disease Control (DDC) Thai government. modeling employed 2 types data: dynamic (time series) static. phases: 1) LSTM manipulate time series 2) MLP static data. Then, merged further analysis. We evaluated performance combined model, yielding an accuracy 99.72 % based R2 values, higher than values state-of-the-art methods. addition, results can be GIS each province displayed via easy-to-use web mapping. HIGHLIGHTS A architecture that as tool situation is proposed Deep learning applied create predictive MLPs displaying predictions using form map developed show detail GRAPHICAL ABSTRACT

Language: Английский

Citations

3
Underground storage tank blowout analysis: Stability prediction using an artificial neural network DOI Creative Commons
Nhat Tan Duong, Van Qui Lai, Jim Shiau

et al.

Journal of Safety Science and Resilience, Journal Year: 2023, Volume and Issue: 4(4), P. 366 - 379

Published: Oct. 18, 2023

Most geotechnical stability research is linked to “active” failures, in which soil instability occurs due self-weight and external surcharge applications. In contrast, on passive failure not common, as it predominately caused by loads that act against the self-weight. An earlier active trapdoor investigation using Terzaghi's three factor approach was shown be a feasible method for evaluating cohesive-frictional stability. Therefore, this technical note aims expand assess drained circular (blowout condition) under axisymmetric conditions. Using numerical finite element limit analysis (FELA) simulations, cohesion, surcharge, unit weight effects are considered factors (Fc, Fs, Fγ), all associated with cover-depth ratio internal friction angle. Both upper- bound (UB) lower-bound (LB) results presented design charts tables, large dataset further studied an artificial neural network (ANN) predictive model produce accurate equations. The proposed problem conditions significant when considering blowout owing faulty underground storage tanks or pipelines high pressures.

Language: Английский

Citations

1
COVID-19 and regional resilience: Trends, priorities, and the geography of research DOI Creative Commons
Irina Turgel, Olga Chernova

Journal of Safety Science and Resilience, Journal Year: 2024, Volume and Issue: 5(3), P. 295 - 305

Published: June 13, 2024

The global economic crisis of 2008–2013 led to the emergence concept resilience, which focuses on ability socio-economic system store cover socially, economically, and environmentally after external impacts. COVID-19 pandemic spurred scholarly interest in regional resilience as a new conceptual framework for sustainability theory. This paper aims examine influence trends geography studies. We analyzed data derived from Science Direct used VOSviewer perform clustering bibliometric network analysis. countries that suffered most showed largest socioeconomic disparities have become centers knowledge resilience. Moreover, has visible shift research focus. Thus, 2020, more attention been paid structural topological characteristics regions enable them reorganize their resources effectively times crisis. study investigates potential resilient development gaining insights into factors supporting adaptability.

Language: Английский

Citations

0