Key result
XGBoost machine learning model identifies post-PCI kinesiophobia in older CHD patients with ~0.98 AUC.
Why the study?
Kinesiophobia reduces participation in cardiac rehabilitation and impairs exercise adherence among older patients with CHD after PCI, necessitating tools for its identification.
Does an XGBoost machine learning model accurately identify concurrent kinesiophobia in older patients with coronary heart disease after percutaneous coronary intervention?
Cross-Sectional (n=652)
Yes
Does an XGBoost machine learning model accurately identify concurrent kinesiophobia in older patients with coronary heart disease after percutaneous coronary intervention?
An interpretable machine learning model utilizing 15 clinical and psychosocial predictors effectively identifies kinesiophobia in older adults following PCI, providing a tool to target personalized cardiac rehabilitation.
Answers are grounded in this paper and cite their sources.
May aid kinesiophobia identification post-PCI; leaves open prospective validation and rehabilitation impact.
Background: Kinesiophobia reduces participation in cardiac rehabilitation and impairs exercise adherence among older patients with coronary heart disease (CHD) after percutaneous coronary intervention (PCI). This study aimed to develop and validate an interpretable machine learning (ML) model for identifying prevalent kinesiophobia in this population at the time of assessment. Methods: = 150). A further 150 patients recruited from the same hospitals between September and October 2025 constituted a temporal validation cohort. Five candidate algorithms were evaluated: Random Forest (RF), Logistic Regression (LR), Decision Tree (DT), eXtreme Gradient Boosting (XGBoost), and Support Vector Machine (SVM). Model development was conducted using a rigorous nested cross-validation framework with a 10-fold outer loop and a 5-fold inner loop. Data preprocessing, feature selection, synthetic minority oversampling technique (SMOTE) resampling, and hyperparameter tuning were independently repeated within the corresponding training partitions. The final algorithm was selected based on predictions generated from inner cross-validation within the development set, jointly considering discrimination, overall prediction error, calibration, and clinical net benefit. Results: A total of 15 predictors were identified, including the Barthel Index, avoidance, residence, fear of disease progression, clinical classification, confrontation, resignation, pain severity, educational level, smoking status, gender, depression, drinking status, exercise self-efficacy, and self-perceived burden. The mean outer-loop area under the receiver operating characteristic curve (AUC) values for LR, DT, RF, SVM, and XGBoost were 0.884 ± 0.076, 0.795 ± 0.069, 0.876 ± 0.071, 0.879 ± 0.072, and 0.878 ± 0.072, respectively. XGBoost demonstrated the best overall performance. The XGBoost model achieved an AUC of 0.821 [95% confidence interval (CI): 0.755-0.884] in the internal validation set and 0.981 (95% CI: 0.963-0.994) in the external validation set. The Brier scores were 0.170 and 0.062 in the internal and external validation sets, respectively. The corresponding calibration intercepts were 0.269 and 0.150, while the calibration slopes were 0.744 and 2.072, respectively. Conclusion: This study developed an interpretable ML model for the concurrent identification of kinesiophobia in older patients with CHD after PCI. After further validation in broader populations, the model may support early identification and personalized cardiac rehabilitation management.
No takes yet. Share an insight, caveat, or question.
Xi et al. (2026) conducted a cross-sectional in Coronary heart disease (CHD) after percutaneous coronary intervention (PCI) (n=652). XGBoost machine learning model vs. Other machine learning models (Logistic Regression, Decision Tree, Random Forest, Support Vector Machine) was evaluated on Concurrent identification of kinesiophobia (measured by AUC) (AUC 0.821 (internal) and 0.981 (external), 95% CI 0.963-0.994). An XGBoost machine learning model accurately identified kinesiophobia in older patients with CHD after PCI, achieving an AUC of 0.821 in the internal validation set and 0.981 in the external validation set.
Synapse has enriched 10 related papers. Consider them for comparative context: