Using machine learning to define the association between cardiorespiratory fitness and all-cause mortality: The fit (henry ford exercise testing) project
Al-Mallah MH, Ahmed A, Qureshi W, Elshawi R, Brawner C, Blaha M, Ahmed H, Ehrman J, Keteyian S, Sakr S. Using machine learning to define the association between cardiorespiratory fitness and all-cause mortality: The fit (henry ford exercise testing) project. J Am Coll Cardiol. Mar 2017;69(11):1612-1612.
J Am Coll Cardiol
Background: Prior studies have demonstrated that cardiorespiratory fitness (CRF) is a strong marker of cardiovascular health. Machine learning (ML) can enhance the prediction of outcomes through classification technique that classifies the data into predetermined categories. Theaim of the analysis is to assess the relation between CRF and all-cause mortality (ACM) using ML approaches. Methods: We included 34,212 patients (55% males, mean age 54±13years) not known to have coronary artery disease or heart failure who underwent clinician-referred exercise treadmill stress testing at Henry Ford Health Systems Between 1991 and 2009 and had complete 10-year follow-up. The primary outcome of this analysis was ACM at 10 years. The probability of 10 year ACM was calculated usinglogistic regression (LR) and ML and the accuracy of these methods were calculated and compared. Results: A total of 3,921 patients experienced ACM at ten years. Using LR, thesensitivity to predict ACM was 44.9% (95%CI 43.3%- 46.5%) while the specificity was 93.4% (95%CI 93.1%-93.7%). Thesensitivity of ML to predict ACM was 87.40% (86.32%-88.42%) while the specificity was 97.21% (97.02%-97.39%). ML approach was associated with improved model discrimination, (area under the curve for ML (0.923 (95%CI 0.917-0.928)) compared to LR (0.836 (95%CI 0.829-0.846)), p<0.0001)(Figure 1). Conclusions: Our analysis demonstrates that ML provides better accuracy and discrimination of the prediction of ACM among patients undergoing stress testing.