Statistical Measurement and Analysis on How the Late-Life Function Disability Instrument Enhances the Frailty Assessment Compared to the National Standards Used on Transcatheter Aortic Valve Patients (Application Paper)

摘要

Transcatheter aortic valve replacement (TAVR) has become a more utilized procedure to perform on patients deemed too frail to handle the demands of a standard open heart approach. How to determine frailty in cardiac patients, particularly TAVR candidates, has been difficult to objectively quantify. The purpose of this research was to statistically measure which outcome tool most accurately depicted frailty in patients who underwent TAVR. Our study was performed based on the comparison between two approaches: the Kansas City Cardiomyopathy Questionnaire (KCCQ), which is the current national assessment standard conducted, and the Fried scale, which tests five frailty domains: gait speed, grip strength, low physical activity, exhaustion, and weight loss. Each domain of the Fried scale was explored and compared alongside the KCCQ with those frail/not frail to the TAVR patients with complications and deaths. Low physical activity was the strongest single-frailty domain predictor. Three statistical models - Logistic Regression, Support Vector Machines (SVM), and Artificial Neural Network (ANN), were used to build classification systems to predict complication conditions. Comparing static numbers, such as Sensitivity, Specificity and Area under Curve (AUC), it is believed that composed models based on the five domains of the Fried scale were able to demonstrate more accurate results than the traditional KCCQ approach. Both SVM and ANN showed significant performance, but further research is necessary to confirm specificity for the Fried scale with the TAVR population.