Purpose Ex vivo heart perfusion (EVHP) provides a means to assess cardiac function before heart transplantation . End-systolic elastance (E es ) has been proposed as a relative load-insensitive index of cardiac function. However, due to the difficulties to obtain multiple stable pressure-volume (PV) loops at different preload conditions in EVHP, accurate prediction of organ viability using E es is limited. In this study, Ees estimation on single-beat basis prior to heart transplantation was used to predict early stage graft survival in a porcine model. Methods Single-beat determination of Ees was developed by employing end systolic pressure (P es ), end systolic volume (V es) and volume axis intercept of Ees (V0): Ees=Pes/(Ves-V0). V0 was estimated using machine learning method (support vector regression). 20 EVHP experiments using pig hearts were performed to train the Ees estimation model. The trained Ees estimation model was then used to predict early post transplantation outcomes in 6 experiments where donor pig hearts were procured and perfused ex vivo for 4 hours before transplant. Hearts were transitioned into working mode for functional assessment during the 4th hour perfusion and Ees was derived using the proposed single-beat method. Linear regression was performed to study the correlation between E es and post-transplant cardiac index. Results In the first 20 EVHP experiments, 10 parameters (Table 1) were selected as inputs of the machine learning model. Ees determined by single-beat method aligned well with conventional multi-beat estimates obtained by occlusion process (R=0.959, p Conclusion Determination of Ees from a single, steady-state beat using machine learning technology could provide an effective approach to predict early post-transplant outcomes in porcine model.