Abstract 4117883: Long noncoding RNAs and machine learning to improve cardiovascular outcomes of COVID-19

Introduction/Background: Cardiovascular symptoms appear in a high proportion of patients in the few months following a severe SARS-CoV-2 infection. Non-invasive methods to predict disease severity could help personalizing healthcare and reducing the occurrence of these symptoms. Research Questions/Hypothesis: We hypothesized that blood long noncoding RNAs (lncRNAs) and machine learning (ML) could help predict COVID-19 severity. Goals/Aims: To develop a model based on lncRNAs and ML for predicting COVID-19 severity. Methods/Approach: Expression data of 2906 lncRNAs were obtained by targeted sequencing in plasma samples collected at baseline from four independent cohorts, totaling 564 COVID-19 patients. Patients were aged 18+ and were recruited from 2020 to 2023 in the PrediCOVID cohort (n=162; Luxembourg), the COVID19_OMICS-COVIRNA cohort (n=100, Italy), the TOCOVID cohort (n=233, Spain), and the MiRCOVID cohort (n=69, Germany). The study complied with the Declaration of Helsinki. Cohorts were approved by ethics committees and patients signed an informed consent. Results/Data: After data curation and pre-processing, 463 complete datasets were included in further analysis, representing 101 severe patients (in-hospital death or ICU admission) and 362 stable patients (no hospital admission or hospital admission but not ICU). Feature selection with Boruta, a random forest-based method, identified age and five lncRNAs (LINC01088-201, FGDP-AS1, LINC01088-209, AKAP13, and a novel lncRNA) associated with disease severity, which were used to build predictive models using six ML algorithms. A naïve Bayes model based on age and five lncRNAs predicted disease severity with an AUC of 0.875 [0.868-0.881] and an accuracy of 0.783 [0.775-0.791]. Conclusion: We developed a ML model including age and five lncRNAs predicting COVID-19 severity. This model could help improve patients’ management and cardiovascular outcomes.

Authors

Institutions

Publication Details

Journal
Circulation
Published
2024-11-12
DOI
https://doi.org/10.1161/circ.150.suppl_1.4117883
Primary Topic
Cardiovascular Health and Risk Factors
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Abstract 4117883: Long noncoding RNAs and machine learning to improve cardiovascular outcomes of COVID-19

Lina Badimón, Marko Jordan, Bence Ágg, Kanita Karađuzović-Hadžiabdić et al.
Circulation
Cardiovascular Health and Risk Factors
article

Abstract 4117883: Long noncoding RNAs and machine learning to improve cardiovascular outcomes of COVID-19

Lina Badimón, Marko Jordan, Bence Ágg, Kanita Karađuzović-Hadžiabdić, Costanza Emanueli, Teresa Padró, Bettina Benczik, Markus W. Ollert, Maciej Rosołowski, Simona Greco, Timo Brandenburger, Muhammad Shoaib, Péter Ferdinandy, Gabriel R. Sánchez, Muhamed Adilović, Hueseyin Firat, Yvan Devaux, Mitja Luštrek, Guy Fagherazzi, Prashant K. Srivastava, Markus Scholz, Andrew Howard Baker, Fabio Martelli, Andrew I. Lumley, Venkata Satagopam, Lu Zhang, Pedro Domingo, Joanna Michel, Pranay Shah
article en

Abstract

Introduction/Background: Cardiovascular symptoms appear in a high proportion of patients in the few months following a severe SARS-CoV-2 infection. Non-invasive methods to predict disease severity could help personalizing healthcare and reducing the occurrence of these symptoms. Research Questions/Hypothesis: We hypothesized that blood long noncoding RNAs (lncRNAs) and machine learning (ML) could help predict COVID-19 severity. Goals/Aims: To develop a model based on lncRNAs and ML for predicting COVID-19 severity. Methods/Approach: Expression data of 2906 lncRNAs were obtained by targeted sequencing in plasma samples collected at baseline from four independent cohorts, totaling 564 COVID-19 patients. Patients were aged 18+ and were recruited from 2020 to 2023 in the PrediCOVID cohort (n=162; Luxembourg), the COVID19_OMICS-COVIRNA cohort (n=100, Italy), the TOCOVID cohort (n=233, Spain), and the MiRCOVID cohort (n=69, Germany). The study complied with the Declaration of Helsinki. Cohorts were approved by ethics committees and patients signed an informed consent. Results/Data: After data curation and pre-processing, 463 complete datasets were included in further analysis, representing 101 severe patients (in-hospital death or ICU admission) and 362 stable patients (no hospital admission or hospital admission but not ICU). Feature selection with Boruta, a random forest-based method, identified age and five lncRNAs (LINC01088-201, FGDP-AS1, LINC01088-209, AKAP13, and a novel lncRNA) associated with disease severity, which were used to build predictive models using six ML algorithms. A naïve Bayes model based on age and five lncRNAs predicted disease severity with an AUC of 0.875 [0.868-0.881] and an accuracy of 0.783 [0.775-0.791]. Conclusion: We developed a ML model including age and five lncRNAs predicting COVID-19 severity. This model could help improve patients’ management and cardiovascular outcomes.

CirculationVol. 150(Suppl_1)
Semmelweis University (HU), International University of Sarajevo (BA), University of Luxembourg (LU), Hospital de Sant Pau (ES), Jožef Stefan Institute (SI), IRCCS Policlinico San Donato (IT), Luxembourg Institute of Health (LU), Firalis (France) (FR), Heinrich Heine University Düsseldorf (DE), Imperial College London (GB), Universitat de Barcelona (ES), Leipzig University (DE), University of Edinburgh (GB)
Good health and well-being
Openalex Percentile: Top 63%
Cardiovascular Health and Risk Factors
DOISourceOpenAlex
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.