Predicting Long-Term Depression Progression in Parkinson's Disease: A Machine-Learning Survival Analysis and Risk Score - PubMed
3 hours ago
- #risk-stratification
- #survival-analysis
- #machine-learning
- The study developed a machine learning survival model and an integer-based risk score to predict long-term depressive progression in Parkinson's disease (dPD).
- Using data from 496 de novo, drug-naïve Parkinson's patients from the PPMI, the model identified key predictors including age, baseline GDS-15, SCOPA-AUT subscores, cognition measures, impulse control disorder, and MDS-UPDRS I symptoms.
- The Random Survival Forest model achieved the best discrimination with a test-set C-index of 0.744 and stratified patients into low, moderate, and high-risk groups with clear differences in progression rates over a median 6-year follow-up.
- Time-dependent AUC values ranging from 0.721 to 0.812 at 2 to 10 years demonstrated the model's predictive accuracy, supporting its use for early, personalized risk stratification and management of depression in Parkinson's disease.