Predicting Hospitalization Due to COPD Exacerbations in Swedish Primary Care Patients Using Machine Learning - Based...

Area of Science:

• Pulmonary Medicine
• Health Informatics
• Machine Learning in Healthcare

Background:

• Chronic obstructive pulmonary disease (COPD) exacerbations significantly increase disease severity, progression, mortality, and hospitalization rates.
• Predicting severe COPD exacerbations is crucial for timely intervention and improved patient outcomes.

Purpose of the Study:

• To develop and validate a predictive model for COPD-related hospitalizations using Swedish patient data.
• To identify key clinical factors associated with severe COPD exacerbations requiring hospitalization.

Main Methods:

• Utilized machine learning on a dataset of 7823 Swedish COPD patients (2000-2013) from EMRs and national registries.
• Developed models to predict the risk of COPD-related hospitalization within 10 days of an event.
• Assessed model performance using Area Under the Precision-Recall Curve (AUPRC) and Area Under Receiver Operating Curve (AUROC).

Main Results:

• The strongest predictors for severe exacerbations were prior exacerbations (past 6 months and overall history), number of COPD-related healthcare contacts, and comorbidity burden.
• The final model achieved an AUROC of 0.86 and an AUPRC of 0.08 on test data.
• These performance metrics were notably high compared to previous studies predicting COPD exacerbations.

Conclusions:

• Clinically available patient history data, retrievable from EMRs and national registries, can be leveraged for predicting severe COPD exacerbations.
• The developed model shows potential for integration into future clinical tools to proactively manage COPD exacerbation risk.