Artificial Intelligence Predicts Optimal Glucocorticoid Dose for Noncritical COVID-19 Patients

Strategic Insights -

Researchers at Zhejiang University have used artificial intelligence (AI) to develop a predictive model that accurately predicts the optimal glucocorticoid dose for noncritical COVID-19 patients. According to the study, which was published in the journal Enfermedades Infecciosas Y Microbiologia Clinica, the AI-based model takes into account various factors, including pulmonary CT features, to determine the best dose of glucocorticoids for treatment. The study found that high-dose treatment was administered to 82.7% of patients in the training cohort, while low-dose treatment was administered to 17.3%. The predictive model, which was validated in a separate cohort, achieved an area under the curve (AUC) of 0.836, indicating a high level of accuracy.

Key Takeaways:

  • The study used AI to develop a predictive model that accurately predicts the optimal glucocorticoid dose for noncritical COVID-19 patients.
  • The model took into account pulmonary CT features, including normally inflated ratio, ground-glass opacity (GGO) ratio, and consolidation ratio.
  • High-dose treatment was administered to 82.7% of patients in the training cohort, while low-dose treatment was administered to 17.3%.
  • The predictive model achieved an AUC of 0.836 in the validation cohort.
  • The model was validated using a separate cohort of 30 patients, who received post-CT adjusted dosages.
  • The study concluded that the predictive model can predict glucocorticoid dosage, but should be validated by larger, prospective studies.

Statistics:

  • 273 noncritical COVID-19 pneumonia patients were analyzed in the study.
  • The patients were divided into two groups based on their glucocorticoid dose: high-dose (139 patients or 82.7%) and low-dose (29 patients or 17.3%).
  • The predictive model achieved an AUC of 0.803 in the training cohort and 0.836 in the validation cohort.
  • 30 patients with post-CT adjusted dosages were used to validate the model, with the predicted dosages highly matching the actual adjusted dosages.

Sources:

  • Factors associated with glucocorticoid dosing in treating patients with noncritical COVID-19 pneumonia: Insights from an artificial intelligence-based CT imaging analysis, Enfermedades Infecciosas Y Microbiologia Clinica, 2025;43(7):402-410.
  • NewsRx. New Artificial Intelligence Findings from Zhejiang University Described (Factors associated with glucocorticoid dosing in treating patients with noncritical COVID-19 pneumonia: Insights from an artificial intelligence-based CT imaging analysis). Robotics & Machine Learning. August 18, 2025; p 70.