Prof. Xavier Jouven (Georges-Pompidou European Hospital, France) and co-authors aimed to develop a sudden cardiac death prediction with a particular focus on the low-risk general population since they comprise the highest absolute number of sudden cardiac death and are more difficult to capture than patients with known high-risk factors, such as previous myocardial infarctions [1]. The study group included 12,784 cases of sudden cardiac death from the Paris Sudden Death Expertise Center and 10,000 matched controls. Using AI, an individualised 1-year predicted risk score for sudden cardiac death was calculated.
Selected were 200 variables from 10,000 medical codes. The developed deep-learning model displayed an area under the curve (AUC) of 0.88, a sensitivity of 81%, and a positive predictive value of 88%. Notably, when non-cardiovascular variables were added, the AUC of the model was elevated from 0.81 to 0.88. Furthermore, the model was able to identify participants who had a risk of 90% or more of experiencing sudden cardiac death in the coming year with a positive predictive value of 95%, capturing 43% of the total number of sudden cardiac death cases. The study group will be working to improve the sudden cardiac death risk prediction model by identifying clinical subgroups of interest in a successive study.
- Jouven X, et al. Prediction of sudden cardiac death using artificial intelligence. Late-breaking science 1, EHRA 2022, 3–5 April, Copenhagen, Denmark.
Copyright ©2022 Medicom Medical Publishers
Posted on