PP90 Artificial Intelligence To Detect Ischemic Heart Disease In Non-traumatic Chest Pain At The Emergency Department – SmartHeart Study

Abstract

Introduction

An estimated 17.9 million people died from cardiovascular diseases (CVDs) in 2019, which is 32 percent of all global deaths and 85 percent were due to heart attack and stroke. Chest pain is one of the most common reasons for presenting to the emergency department (ED). It is increasingly recognized that artificial intelligence (AI) will have a significant impact on the practice of medicine in the near future and may help with diagnosis and risk stratification. We aim to estimate a diagnostic prediction of acute myocardial infarction by the development and validation of an AI model.

Methods

Data on 134 variables of 3,986 consecutive patients who presented to the ED with non-traumatic chest pain were included in the analysis. Using AI tools, a neural network model was developed to establish the risk of acute myocardial infarction (AMI) to achieve n=150 patients over 18 years of age attending the ED.

Results

The mean age was 65.5 (±13.7) years and 63.6 percent were male. Most (60.1%) patients were admitted to hospital, with only 20.3 percent diagnosed at hospital discharge with ischemic heart disease (IHD). All patients were followed up for two months, and 6.3 percent were readmitted to the ED, but none presented with an episode of IHD. In the data analysis of the entire sample we obtained a probability of diagnosing IHD by the SmartHeart model (S=93.1%, E=47.3%, PPV=31.0%, and NPV=96.4%). When we analyzed the sample of patients with no history of IHD (n=104), the diagnosis accuracy was as follows (S=100%, E=77.5%, PPV=42.8%, and NPV=100%).

Conclusions

Our AI model provides information to predict patients who are suffering from acute IHD. AI has been reported to outperform emergency physicians and current risk stratification tools to diagnose IHD, but has rarely been integrated into practice. This study highlights the diagnostic applicability and accuracy of this type of tool and that is why studies should be implemented to see its effectiveness in routine practice in EDs.