Diagnostic Models for Differentiating COVID-19-Related Acute Ischemic Stroke Using Machine Learning Methods DOI Creative Commons
Eylem Gül Ateş, Gökçen Çoban, Jale Karakaya

et al.

Diagnostics, Journal Year: 2024, Volume and Issue: 14(24), P. 2802 - 2802

Published: Dec. 13, 2024

Backgrounds: Although COVID-19 is primarily known as a respiratory disease, there growing evidence of neurological complications, such ischemic stroke, in infected individuals. This study aims to evaluate the impact on acute stroke (AIS) using radiomic features extracted from brain MR images and machine learning methods. Methods: retrospective included MRI data 57 patients diagnosed with AIS who presented Department Radiology at Hacettepe University Hospital between March 2020 September 2021. Patients were stratified into COVID-19-positive (n = 30) COVID-19-negative 27) groups based PCR results. Radiomic following image processing steps. Various feature selection algorithms applied identify most relevant features, which then used train classification models. Model performance was evaluated range metrics, including measures predictive accuracy diagnostic reliability, 95% confidence intervals provided enhance reliability. Results: assessed dimensionality reduction distinguishing cases radiomics scans. Without selection, ANN achieved highest AUC 0.857 (95% CI: 0.806–0.900), demonstrating strong discriminative power. Using Boruta method for k-NN classifier attained best performance, an 0.863 0.816–0.904). LASSO-based showed comparable results across k-NN, RF, classifiers, while SVM exhibited excellent specificity high PPV. The RFE yielded overall achieving 0.882 0.838–0.924) 79.1% 73.6–83.8). Among methods, consistent results, identified effective classifiers detection. Conclusions: proposed radiomics-based model effectively distinguishes associated MRI. These findings demonstrate potential AI-driven tools high-risk patients, support optimized treatment strategies, ultimately improve clinical implications.

Language: Английский

Analysis of risk factors and construction of a prediction model for posttraumatic stress disorder among Chinese college students during the COVID-19 pandemic DOI

Guangjian Li,

Xugui Sun,

Tingye Gao

et al.

Journal of Affective Disorders, Journal Year: 2024, Volume and Issue: 362, P. 230 - 236

Published: July 3, 2024

Language: Английский

Citations

1
Diagnostic Models for Differentiating COVID-19-Related Acute Ischemic Stroke Using Machine Learning Methods DOI Creative Commons
Eylem Gül Ateş, Gökçen Çoban, Jale Karakaya

et al.

Diagnostics, Journal Year: 2024, Volume and Issue: 14(24), P. 2802 - 2802

Published: Dec. 13, 2024

Backgrounds: Although COVID-19 is primarily known as a respiratory disease, there growing evidence of neurological complications, such ischemic stroke, in infected individuals. This study aims to evaluate the impact on acute stroke (AIS) using radiomic features extracted from brain MR images and machine learning methods. Methods: retrospective included MRI data 57 patients diagnosed with AIS who presented Department Radiology at Hacettepe University Hospital between March 2020 September 2021. Patients were stratified into COVID-19-positive (n = 30) COVID-19-negative 27) groups based PCR results. Radiomic following image processing steps. Various feature selection algorithms applied identify most relevant features, which then used train classification models. Model performance was evaluated range metrics, including measures predictive accuracy diagnostic reliability, 95% confidence intervals provided enhance reliability. Results: assessed dimensionality reduction distinguishing cases radiomics scans. Without selection, ANN achieved highest AUC 0.857 (95% CI: 0.806–0.900), demonstrating strong discriminative power. Using Boruta method for k-NN classifier attained best performance, an 0.863 0.816–0.904). LASSO-based showed comparable results across k-NN, RF, classifiers, while SVM exhibited excellent specificity high PPV. The RFE yielded overall achieving 0.882 0.838–0.924) 79.1% 73.6–83.8). Among methods, consistent results, identified effective classifiers detection. Conclusions: proposed radiomics-based model effectively distinguishes associated MRI. These findings demonstrate potential AI-driven tools high-risk patients, support optimized treatment strategies, ultimately improve clinical implications.

Language: Английский

Citations

0