Machine learning predicts pulmonary Long Covid sequelae using clinical data

Area of Science:

• Medical Informatics
• Pulmonology
• Machine Learning

Background:

• Long COVID is a multi-systemic condition impacting quality of life, frequently involving pulmonary complications.
• Early prediction of Long COVID sequelae is essential for timely intervention and preventing severe outcomes.

Purpose of the Study:

• To investigate machine learning approaches for predicting pulmonary Long COVID sequelae using clinical hospitalization data.
• To develop predictive models for early identification of patients at risk for Long COVID pulmonary complications.

Main Methods:

• Utilized three distinct machine learning approaches: a traditional shallow learner, an ensemble of classifiers, and a multimodality-driven model.
• Trained and evaluated models on clinical data from 152 patients hospitalized with COVID-19.

Main Results:

• Achieved predictive accuracy of up to $$ for pulmonary Long COVID sequelae.
• Demonstrated the feasibility of using machine learning on clinical data for early prediction.

Conclusions:

• Machine learning models show significant potential in predicting Long COVID pulmonary complications.
• The study contributes a publicly available dataset to advance research in Long COVID prediction.