Optimize risk prediction after myocardial infarction through artificial intelligence and multidimensional evaluation

Introduction

Myocardial infarction (MI) is a leading cause of death worldwide. After MI, long-term antithrombotic therapy is crucial to prevent recurrent events, but it increases bleeding, which also impacts morbidity and mortality.

Importance of Prediction Tools

Giving these competing risks prediction tools to forecast ischemic and bleeding events is of paramount importance to inform clinical decisions, but their current precision is limited. Improving event prediction by discovering novel and innovative markers of risk would have a tremendous impact on therapeutic decisions and patients’ outcomes.

Hypothesis

I hypothesize that using innovative multidimensional information from wearable devices, biomarkers, behavioral patterns, and non-invasive imaging, integrated through artificial intelligence computation, we may discover novel “computational biomarkers” of risk and improve current standards of risk prediction.

Project Overview

In this project, I will enroll a large cohort of MI patients, whereby prospective collection of consolidated and innovative potential risk predictors will take place in order to generate a comprehensive and multidimensional dataset.

Data Collection

I will collect data from:

1. State-of-the-art non-invasive imaging
2. Blood biomarkers
3. Wearable medical devices that monitor continuous heart electrical activity, sweat, mobility, and behavioral patterns

This will create a large physiological time series allowing for patients’ deep phenotyping.

Data Analysis

We will analyze data leveraging artificial intelligence computation to find relevant associations with clinical outcomes and compare new algorithms with current risk prediction tools.

Expected Outcomes

This research will increase our knowledge of bleeding and ischemic risk factors, enabling enhanced capability in prediction models. In the near future, we hypothesize that our clinically-guided artificial intelligence algorithm might be integrated into clinical practice, helping clinicians to inform treatment decisions, allowing patients to better understand their risk profile, and finally setting a common ground for shared patient/physician decisions.