NoSysOracle: A Predictive Ensemble Learning Model for Cardio Disease Risk Assessment and Clinical Guidance

Heart disease is responsible for approximately 17.9 million deaths every year globally according to the World Health Organization [1]. Therefore, it’s extremely essential to identify individuals who are at risk of disease so we can prevent the onset of disease and to provide the appropriate medical intervention in a timely manner. The study presented focuses on a system called NoSysOracle which is designed to identify individuals who may be at risk of developing cardiovascular disease. NoSysOracle employs a mixture of machine learning models along with a specialized collection of software tools to assist us in interpreting the results of the analysis. The collection utilized to gather the training data consists of approximately 50,000 patient records, containing information such as demographics, lifestyle factors, & clinical presentation. The analytic process employed within NoSysOracle utilizes three different machine-learning algorithms—Random Forest, Extra Trees, and LightGBM—in conjunction with one another to help generate predictions regarding a patient's risk of developing heart disease. This synergistic approach improves prediction accuracy while ensuring stability in the resulting predictions. In addition, the way in which NoSysOracle presents its predictions is rather unique because it also converts each of the individual predictions into forms of actionable advice.