Prediction of major adverse cardiovascular events in heart failure patients using face recognition: rationale and study design of the CARDIOMIRROR trial

Abstract

Background: Heart failure (HF) is a growing epidemic, currently affecting more than 26 million people around the globe. Although numerous prognostic markers of death and HF hospitalization have been identified, their clinical applicability is limited and precise risk stratification in HF remains challenging. Aim: The aim of this project is to develop a software (intelligent mirror or CARDIOMIRROR) with the ability to detect significant facial changes in patients with heart failure with reduced ejection fraction (HFrEF). Our main hypothesis is that facial changes will correlate with major adverse cardiovascular events (MACE) in these patients. Methods: For this purpose, we have designed an observational, prospective, multicenter pilot study, which will include 100 adult ambulatory patients with HFrEF, in NYHA class ≥ II and, at least, one hospitalization due to decompensated HF in the past 12 months or BNP ≥500 pg/ml. The primary end-point is to analyze the morphological and colorimetric facial changes that occur in patients with HFrEF, and to determine the association between these changes and the presence of MACE at 1 year, including episodes of acute decompensated HF, myocardial infarction, stroke or death. Secondary end-points include the correlation between facial changes and natriuretic peptides, LVEF, quality of life, NYHA class and daily physical activity. All patients will be given a Smartphone, and they will be required to take a daily 7-seconds video of their face during 12 months of follow-up. This daily video will be sent to an automatic server for its morphologic and colorimetric analysis by experienced engineers (Figure 1), employing Deep Learning, virtual and augmented Reality, Internet of Things and Data Analytics. Conclusion: The CARDIOMIRROR study represents a revolutionary strategy, as it involves shifting from a traditional reactive model to an innovative predictive model, in which medical care anticipates the event in question. Its results could potentially change the conventional approach to HF patients.