A Research Classification for Long COVID: Symptom Frequency and Severity Improve Accuracy of Machine Learning Models.

Long COVID definitions based solely on symptom occurrence may reduce diagnostic specificity.Machine-learning models incorporating symptom frequency and severity outperformed occurrence-only models.Composite scoring achieved 90.12% accuracy compared with 88.73% for occurrence-based scoring.Composite models required fewer predictive symptoms, indicating greater efficiency.Measuring symptom burden improves the precision of research classification for Long COVID.