Opening the Black Box: Explainable Machine Learning for Heart Disease Patients

Abstract

Unfortunately, the “black box” stigma that has long plagued machine learning has caused many doctors to remain leery of its applications. A “black box” is a model that is so complicated that it is difficult, if not impossible, for a person to understand. Especially in the field of medicine, where many choices have profound consequences for patients’ lives, the inability to understand the reasoning behind prediction models might erode their credibility. New studies in the area of explainable machine learning try to allay these fears. While the benefits of explainable machine learning are vast, they are especially pertinent when deciding whether or not to admit a patient to an intensive care unit. In this article, we take a look at the fundamentals of explainable machine learning and how they might be applied to the healthcare industry. In the first step, we use four well-known boosting methods for heart disease detection and prognosis. Finally, we provide an explanation of the aspects that contribute to the best model judgments, improving clinical reasoning by enhancing our capacity for better predictions and our comprehension of the reasons behind them

Ähnliche Arbeiten