Artificial intelligence algorithms based approach in evaluating COVID-19 patients and management.

Area of Science:

• Radiology
• Pulmonology
• Artificial Intelligence in Medicine

Background:

• COVID-19 pneumonia presents with diverse symptoms, from mild to severe.
• Chest CT is vital for assessing pulmonary lesions and disease severity.
• AI tools aid in rapid diagnosis and quantitative lesion evaluation for COVID-19.

Purpose of the Study:

• To correlate lung lesions on CT scans with biological inflammatory markers.
• To establish the survival rate among patients with COVID-19 pneumonia.
• To evaluate the role of AI in assessing COVID-19 pneumonia severity.

Main Methods:

• Retrospective analysis of 120 moderate to severe COVID-19 pneumonia patients.
• Chest CT scans analyzed by radiologists and AI post-processing software.
• Assessment of inflammatory markers including neutrophil-lymphocyte ratio (NLR) and interleukin-6.

Main Results:

• Strong correlations found between inflammatory markers, interleukin-6, and AI-derived CT severity scores (e.g., NLR r=0.851).
• Elevated inflammatory markers and lung opacity scores correlated with reduced survival.
• Crazy paving pattern on CT associated with increased mortality risk (OR=2.89).

Conclusions:

• AI-based chest CT analysis is crucial for evaluating COVID-19 pneumonia.
• Combining AI CT analysis with inflammatory markers offers a reliable, objective assessment.
• This integrated approach enhances diagnostic accuracy and prognostic evaluation for COVID-19 patients.