Incremental prognostic value of fully-automatic LVEF by stress CMR using machine learning

Abstract Funding Acknowledgements Type of funding sources: None. Background Although some recent reports showed that fully automated artificial intelligence (AI)-based left ventricular ejection fraction (LVEF) measured at stress has good performance, its prognostic value during a stress CMR exam to predict outcomes is not well established. Aim To determine in patients undergoing stress CMR whether fully automated AI-based LVEF (LVEF-AI) measured at stress can provide incremental prognostic value to predict death. Methods Between 2016 and 2018, we conducted a longitudinal study including all consecutive patients referred for vasodilator stress CMR. LVEF-AI was assessed using AI-algorithm combines multiple deep learning networks for LV segmentation. The primary outcome was all-cause death defined by the French National Registry of Death. Cox regression was used to evaluate the association of stress LVEF-AI with death after adjustment for traditional risk factors. Results In 9,712 patients (66±15 years, 67% men), there was an excellent correlation between stress LVEF-AI measurement and LVEF measured by expert (LVEF-expert) (r=0.94, p<0.001). Using Bland–Altman analysis, we found that the difference between the mean LVEF-expert and the LVEF-AI group was −0.1% (−0.066–0.067), that was not statistically significant (p = 0.46). Stress LVEF-AI was associated with death (median [IQR] follow-up 4.5 [3.7–5.2] years) before and after adjustment for risk factors (adjusted hazard ratio [HR], 0.84 [95% CI, 0.82–0.87] per 5% increment, p<0.001). Stress LVEF-AI had similar significant association with death occurrence compared with LVEF-expert. After adjustment, a lower stress LVEF-AI showed the greatest improvement in model discrimination and reclassification over and above traditional risk factors and stress CMR findings (C-statistic improvement: 0.11; NRI=0.250; IDI=0.049, all p<0.001; LR-test p<0.001), with a superior additional prognostic value than the LVEF-AI measured at rest. Conclusion AI-based fully automated LVEF measured at stress is independently associated with the occurrence of death in patients undergoing stress CMR, with an additional prognostic value above traditional risk factors, inducible ischemia and LGE.