Heart failure affects 1-2% of the adult population in developed countries, increasing to ≥10% in people aged 70 and over. There are numerous underlying causes of heart failure, including vascular heart disease, high output states and disease of the heart muscle (myocardial disease).
Our research team includes medical doctors, electrical and computer engineering professors, and researchers focused on developing technologies to aid in medical diagnosis. The collaboration between multi-disciplinary team-mates from the cardiology department of the American University of Beirut Medical Center (AUBMC), the electrical and computer engineering department at the American University of Beirut (AUB), and researchers from BRIC aims to improve the machine learning outcome for the diagnosis of cardiovascular diseases.
Our group has been working extensively over the last period to develop machine learning based models that incorporate clinical history, electrocardiography (ECG) and echocardiography features to differentiate ischemic from non-ischemic heart failure.
We test different classification algorithms to investigate the value of cardiac strain parameters in predicting the type of myocardial disease; and the value of strain data with or without ECG data to predict type of myocardial disease. If the strain data and other features are used to predict the type of myocardial disease (e.g. ischemic or non-ischemic) using a classification algorithm, this would lead to more rapid elucidation of specific underlying diagnosis by guiding subsequent investigations, e.g. angiography or genetic testing. This would mean:
1- fewer investigations, which would be beneficial for the patient’s well-being and the efficient use of hospital resources and funds
2- a faster diagnosis, meaning appropriate management can be commenced sooner. Furthermore, if the used classification models along with the input data predict outcome and/or specific diagnosis more accurately than previous works, a relevant tool could be developed in which strain and other relevant data could be input to predict diagnosis, guiding investigation and management, thus, improving treatment for patients and their families.