The Duke Activity Status Index (DASI), a self-administered 12-item questionnaire has been used to estimate functional capacity and recently demonstrated prognostic information. We sought to develop an abbreviated version for clinical applications. Leveraging the Cleveland Clinic GeneBank Study, we developed an abbreviated DASI questionnaire (aDASI) with the machine learning XGBoost algorithm with the goal of maintaining the accuracy and reliability of the original DASI. We validated the prognostic value of aDASI in a subset of patients with heart failure from GeneBank, as well as an independent dataset from the GUIDE-IT trial. The results confirmed the congruence and accuracy between the original and the abbreviated scores while reducing the number of questions (R=0.97, p<0.001). Both the original DASI score and the aDASI exhibited a strong correlation in both GeneBank and predictive value for all-cause mortality at different time points in the GUIDE-IT cohort. In conclusion, the abbreviated DASI tracks with original DASI assessment and performs comparably to the original DASI questionnaire in predicting all-cause mortality.
Keywords: Duke activity status index; Functional capacity; Prognosis.
Copyright © 2024. Published by Elsevier Inc.