Past Myocardial Infarctions and Gender Predict the LVEF Regardless of the Status of Coronary Collaterals: An AI-Informed Research

Abstract

BACKGROUND: The degree of the development of coronary collaterals is long considered an alternate – that is, a collateral – source of blood supply to an area of the myocardium threatened with vascular ischemia or insufficiency. Hence, the coronary collaterals are beneficial but can also promote harmful (adverse) effects. For instance, the coronary steal effect during the myocardial hyperemia phase and that of restenosis following coronary angioplasty. OBJECTIVES: Our study explores the contribution of coronary collaterals – if any exist – while considering other potential predictors, including demographics and medical history, toward the left ventricular (LV) dysfunction measured through the LV ejection fraction (LVEF). METHODS: Our cross-sectional design study used convenience sampling of 100 patients (n = 100; a male-to-female ratio of 4:1). We conducted frequentist inference statistics using IBM-SPSS version 24 and Microsoft Office Excel 2016 with the analysis ToolPak plugin; we ran parallel neural networks (supervised machine learning (ML)) and a two-step clustering (non-supervised ML) for robust conjoint inference with frequentist statistics. RESULTS: The past incidents of myocardial infarction (p = 0.036) and gender (p = 0.072) influenced the LVEF; both are significant predictors at a 90% confidence interval. We found that gender and past incidents of MI influenced the LVEF regardless of the status of coronary collaterals. Our study did not yield any positive or significant findings concerning the status of coronary collaterals or the coronary circulation dominance patterns. CONCLUSION: Regardless of the status of coronary collaterals, we verified that the female gender is protective of the LV function, contrary to the past infarction incidents that predispose to a deteriorated LV function. Our study’s innovation relates to its status as the first study from India to explore the coronary collaterals and the ejection fraction while incorporating frequentist statistics and narrow artificial intelligence to infer reliable results.