⚡ Quick Summary

This study explored the use of explainable machine learning models to predict moderate to severe obstructive sleep apnea (OSA) in patients with heart failure with preserved ejection fraction (HFpEF). The random forest (RF) model demonstrated exceptional predictive performance, achieving an AUC of 0.974 in internal validation, highlighting its potential for clinical application.

🔍 Key Details

• 📊 Dataset: 1140 patients diagnosed with HFpEF
• 🧩 Features used: Various clinical and demographic variables
• ⚙️ Technology: Random Forest, K-nearest neighbors, Light Gradient Boosting Machine, Support Vector Machine, Extreme Gradient Boosting, Logistic Regression
• 🏆 Performance: RF model: AUC 0.974 (internal), AUC 0.910 (external)

🔑 Key Takeaways

• 🛌 OSA is prevalent in HFpEF patients, worsening their condition.
• 📈 Machine learning can effectively predict OSA severity in this patient group.
• 🏆 The RF model outperformed other models in predictive accuracy.
• 🔍 SHAP analysis identified key variables influencing OSA risk.
• 💡 Decision curve analysis showed high net clinical benefits from the RF model.
• 🌍 Study conducted across multiple centers from 2017 to 2025.
• 📅 External validation cohort included 717 additional participants.

📚 Background

Obstructive sleep apnea (OSA) is a common comorbidity in patients with heart failure, particularly those with preserved ejection fraction (HFpEF). The presence of OSA can exacerbate heart failure symptoms and complicate management strategies. Understanding the interplay between these conditions is crucial for improving patient outcomes and tailoring interventions effectively.

🗒️ Study

This observational multicenter study enrolled 1140 patients diagnosed with HFpEF during hospitalization between January 2017 and December 2019. Participants were randomly assigned to training and internal verification cohorts, with an additional external verification cohort established from January 2022 to August 2025. The study aimed to evaluate the predictive performance of various machine learning models in identifying moderate to severe OSA.

📈 Results

The random forest model emerged as the most effective predictor of moderate to severe OSA, achieving an impressive AUC of 0.974 in the internal verification cohort and 0.910 in the external cohort. These results underscore the model’s robustness and reliability in clinical settings. Furthermore, decision curve analysis indicated that the RF model provided the highest net clinical benefits across various threshold levels.

🌍 Impact and Implications

The findings from this study have significant implications for clinical practice. By leveraging machine learning, particularly the RF model, healthcare providers can enhance the early identification and management of OSA in HFpEF patients. This advancement could lead to improved patient outcomes and more personalized treatment strategies, ultimately benefiting a vulnerable patient population.

🔮 Conclusion

This research highlights the transformative potential of machine learning in predicting OSA among patients with HFpEF. The RF model’s high accuracy and clinical applicability suggest a promising future for integrating advanced analytics into routine healthcare practices. Continued exploration in this field is essential to refine these predictive tools and enhance patient care.

💬 Your comments

What are your thoughts on the use of machine learning in predicting obstructive sleep apnea? We would love to hear your insights! 💬 Share your comments below or connect with us on social media:

• 🐦 X
• 📘 Facebook
• 🔗 LinkedIn