Li_2025_Respir.Med__108554

OBJECTIVES: This study evaluates the impact of long coronavirus disease (Long COVID) on disease exacerbation in patients with myasthenia gravis (MG) and leverages machine learning (ML) approaches to identify high-risk patients. METHODS: Patients with MG and COVID-19 co-infection were recruited from December 5, 2022, to January 20, 2023, at the Second Affiliated Hospital of Guizhou University of Traditional Chinese Medicine (ChiCTR2400087937). Participants underwent structured interviews at 1, 6, and 12 months post-COVID-19 to assess the incident Long COVID and MG exacerbation. The primary outcomes included the incidence and severity of MG exacerbation; the secondary outcome was 12-month mortality. Cox proportional hazards regression was employed to explore the correlation between Long COVID and MG exacerbation. ML algorithms were implemented for predictive modeling. RESULTS: Among 180 patients, 62 (34.4%) experienced MG exacerbation within 12 months, and severe exacerbations (MG-ADL <= 6 points) accounted for 43.5% of cases. Among those with initial MG exacerbations, 14.4% continued to exhibit persistent symptoms without full recovery by the 12-month follow-up. Six fatalities (3.3%) were recorded. Long COVID served as the strongest predictor of exacerbation (aHR = 2.55, 95% CI 1.47-4.42, P = 0.001), followed by acetylcholinesterase (AChE) inhibitor use (aHR = 2.38) and severe COVID-19 classification (aHR = 2.30). The Random Forest model (RF) outperformed other algorithms, achieving an AUC of 0.83, with Long COVID as the top predictive feature. CONCLUSIONS: Long COVID is associated with MG exacerbation over time. RF model shows potential as a scalable tool for MG exacerbation risk stratification.