No abstract

AKI is an important finding in ACS, with a significant impact on hard clinical endpoints such as in-hospital and one-year mortality. It is associated with easily identifiable clinical factors and an invasive strategy does not increase its incidence.

No abstract

Funding Acknowledgements Type of funding sources: None. Introduction 3D left ventricular ejection fraction (LVEF) quantification methods are more accurate and reproducible than 2D echocardiography, however, conventional 3D is time consuming and requires extensive user expertise, thus hindering its routine implementation in busy echocardiography laboratories and its use by inexperienced physicians. HeartModel A.I. (HM) is a simple, fast, recently validated 3D automated analysis software that detects LV endocardial surfaces and calculates LVEF. The aim of this work is to evaluate the performance of HM with experienced and inexperienced physicians, its time saving potential and to assess whether this software can be a better alternative to 2D measurements in routine echocardiography. Methods Prospective analysis of 30 nonconsecutive patients referred for transthoracic echocardiogram in a university hospital echocardiography lab, from 1st February 2021 to 31st March 2021. 2D biplane LVEF was measured by an experienced and inexperienced physician (less than 250 echocardiograms performed), then the same physicians used the automated analysis software to assess LVEF (blinded for each other results). The time to make the measurements was registered. Comparisons of agreement between LVEF measurements (experienced versus inexperienced physicians) included linear regression with Pearson correlation coefficients and Bland-Altman analyses to assess the bias and limits of agreement (defined as 2SD around the mean). Results A total of 30 patients were included, mean age of 68.6 ± 20.1 years and 60% male. HM showed significantly lower acquisition times in both inexperienced (72±17s versus 173± 44s, P<0.01) and experienced (56±12s versus 126±29s, P<0.01) physicians. The difference in time of acquisition between 2D and HM was approximately 101s for inexperienced users and around 70s for experienced users. Regarding LVEF assessment, HM acquisitions compared to 2D measurements showed stronger correlations between experienced and inexperienced physicians (r= 0,98, P<0,01 versus r= 0,92, P<0,01) with minimal bias (−0,5 versus −0,6) and stronger agreement (HM limits of agreement: ± 5,8% versus 2D limits of agreement: ± 12,5%) Conclusion 3D LVEF assessment by HM significantly reduced acquisition times and exhibited higher interobserver agreement than 2D Simpson’s biplane method. These results suggest that automated 3D algorithms, such as HM, may play a key role in implementing 3D measurements in routine practice in busy echocardiography laboratories and allow the use of 3D echocardiography at early stages of physicians training.

No abstract