Post-COVID-19 interstitial lung disease: Insights from a machine learning radiographic model

Area of Science:

• Pulmonology
• Radiology
• Artificial Intelligence

Background:

• Post-acute sequelae of COVID-19 (PACS) represent a growing global health concern.
• Interstitial lung disease (ILD) is a significant manifestation of PACS.
• Machine learning (ML) offers potential for detailed assessment of post-COVID-19 ILD.

Purpose of the Study:

• To evaluate the potential of ML radiographic models in assessing post-COVID-19 ILD.
• To analyze the characteristics of patients with post-COVID-19 ILD.
• To investigate the efficacy of antifibrotic therapy in patients with early fibrotic changes.

Main Methods:

• A multicenter, retrospective study analyzed 232 patients 3 months post-SARS-CoV-2 infection.
• High-resolution computed tomography (HRCT) images were assessed using Imbio Lung Texture Analysis 2.1.
• Patient pulmonary function (FVC, DLCO) and ILD features (ground glass opacities, reticulation) were quantified.

Main Results:

• A significant portion of patients exhibited impaired lung function (FVC/DLCO <80%) and varying degrees of lung abnormalities.
• 5.6% of patients developed fibrotic lung disease with persistent impairment.
• Antifibrotic therapy in these patients led to significant improvements in FVC% predicted at 3 and 6 months.

Conclusions:

• Post-COVID-19 ILD is an emerging condition requiring careful management.
• A subset of patients benefits from early antifibrotic treatment when fibrotic changes are in early stages.
• ML radiographic models show promise for accurate ILD evaluation and guiding treatment decisions.