Predicting Fatal Cardiovascular Outcomes: The Role of Sleep Related Variables

Abstract Rationale: For more than three decades, cardiovascular diseases have consistently been the leading cause of death worldwide. Inconsistent or insufficient sleep duration is linked to a higher risk of developing or experiencing fatal cardiovascular outcome. However, specific factors that affect fatal cardiovascular outcomes are still unclear. Methods: We analyzed an open-access cohort (Sleep Heart Health Study) of participants with cardiovascular disease to identify various factors that could predict the number of days from their sleep study to death. To perform this, we implemented several models using sleep related features from overnight polysomnography (PSG), demographics, respiratory metrics, health conditions, and chronic disorders to predict the year of death. We categorized the time to death into the 4 following classes: 1) 0-3 years, 2) 3-6 years, 3) 6-9 years, and 4) 9-13 years. We then selected features based on their importance scores derived using the Random Forest technique. Multiple classification models were trained, including Random Forest, K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Logistic Regression, and Naïve Bayse. The dataset was first split into 80% training data and 20% testing data. Consequently, hyperparameter tuning was conducted using grid search methods to maximize the accuracy. Ultimately, the best-performing model was selected based on the highest performance scores. Results: We studied 355 subjects, including 193 males, with an average age of 75.8 ± 7.5 years and a BMI of 27.5 ± 4.8. All subjects were deceased from cardiovascular disease following their initial sleep study after an average of 2495.7 ± 1200.6 days. We extracted 320 features to predict cardiovascular-related mortality. The best predictive model was KNN classifier, achieving an accuracy of 82.6%, a precision of 81.4%, a recall of 82.1%, and a F1 score of 81.52% (Figure 1). In terms of feature importance, a lower percentage of time spent in stage 2 of non-rapid eye movement (NREM) sleep, lower average oxygen saturation during NREM sleep, reduced sleep efficiency (measured as the ratio of total sleep duration to in-bed period), higher BMI, and worsen mental health condition were linked to increased risk of earlier death by 5%. Conclusion: This study was a proof of concept to indicate the significant role of specific sleep variables from PSG can be used as a predictor of mortality due to cardiovascular disease. Future studies should explore the benefits of early interventions and their potential to reduce mortality for the at-risk populations.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.