Development and validation of a nomogram for predicting ADL outcomes in patients undergoing subacute stroke rehabilitation based on machine learning and standard bedside clinical data: a retrospective cohort study - Report - MDSpire
Advertisement
Development and validation of a nomogram for predicting ADL outcomes in patients undergoing subacute stroke rehabilitation based on machine learning and standard bedside clinical data: a retrospective cohort study
Clinical Report: Predictive Nomogram for ADL Recovery in Subacute Stroke Patients
Overview
This study developed and validated a predictive nomogram for assessing activities of daily living (ADL) recovery in subacute stroke rehabilitation patients. The model utilizes machine learning and standard clinical data.
Background
Stroke is a leading cause of disability, significantly affecting patients' independence and quality of life. Accurate prediction of ADL recovery is crucial for rehabilitation planning and resource allocation. Existing models often lack applicability in the subacute phase.
Data Highlights
Cohort
AUC
95% CI
Training
0.832
0.779–0.885
Validation
0.866
0.806–0.926
Key Findings
The nomogram includes the Braden score, baseline Barthel Index (BI) score, and age as predictors.
SHAP analysis indicated the Braden score as the most significant predictor for ADL recovery.
ADL independence was defined as a BI score of ≥60.
The model was validated with a cohort of 165 patients, achieving an AUC of 0.866.
This predictive tool is designed for use with routine bedside clinical data collected within 72 hours of patient transfer to rehabilitation.
Clinical Implications
The nomogram provides a method for rehabilitation physicians to predict ADL recovery in subacute stroke patients.
Conclusion
The developed nomogram predicts ADL recovery at 3 months post-rehabilitation.
