A predictive model for post-COVID-19 pulmonary parenchymal abnormalities based on dual-center data.

Area of Science:

• Pulmonology
• Radiology
• Infectious Diseases

Background:

• Coronavirus disease 2019 (COVID-19) survivors frequently exhibit radiological and physiological anomalies.
• Prompt recognition of residual pulmonary parenchymal abnormalities is crucial for managing chronic pulmonary sequelae.

Purpose of the Study:

• To develop a predictive model for identifying patients at high risk of pulmonary abnormalities post-COVID-19.
• To create a tool for effective management of long-term pulmonary consequences.

Main Methods:

• Retrospective cohort study of 501 hospitalized COVID-19 patients (July 2022-March 2023).
• Development of a prognostic model using stepwise regression (Akaike Information Criterion) with clinical and laboratory parameters.
• Model trained and validated on a split dataset; 240 patients underwent Chest CT scans at 3 months post-infection.

Main Results:

• Incidence of pulmonary abnormalities was 33.3% among patients with follow-up CT scans.
• The predictive model demonstrated strong discriminatory power (AUC: 0.885) and excellent calibration.
• Decision curve analysis indicated substantial net benefits for clinical application.

Conclusions:

• A reliable prognostic tool, including a nomogram, has been developed to estimate the probability of residual pulmonary parenchymal abnormalities 3 months post-COVID-19.
• The high-performing model shows promise for guiding clinical interventions and improving management of COVID-19-related pulmonary sequelae.