An overview of cardiovascular disease infection: A comparative analysis of boosting algorithms and some single based classifiers

Abstract

Machine learning is a branch of artificial intelligence that helps machines learn from observational data without being explicitly programmed and its methods have been found to be very useful in the modern age for medical diagnosis and for early detection of diseases. According to the World Health Organization, 12 million deaths occur annually due to heart-related diseases. Thus, its early detection and treatment are of interest. This research introduces a better way of improving the timely prediction of cardiovascular diseases in suspected patients by comparing the efficiency of two boosting algorithms with four (4) other single based classifiers on cardiovascular official data. The best model was selected based on performances of 5 different evaluation metrics. From the results, Adaptive boosting is seen to outperform all other algorithms with a classification accuracy of 74.2%, closely followed by gradient boosting. However, gradient boosting was chosen as an acceptable technique because it trains faster than Adaboost with a better precision of 74.9% compared to 74.7% exhibited by Adaboost. Thus boosting algorithms are better predictors compared to single based classifiers with factors of age, systolic blood pressure, weight, cholesterol, height, and diastolic blood pressure as the major contributors to the model building.

Diagnosing Sexually Transmitted Disease from Some Symptoms Using Machine Learning Models

“…It is mostly utilized in logical and prescient issues when the dimensionality of the data sources is high. In spite of the effortlessness, this is many times utilized in more refined order techniques (Adeboye & Abimbola, 2020). Suppose 𝐴 represents the STDs status and 𝐵 represents any of the predictor variables (symptoms), then…”

“…The idea is that a large collection of uncorrelated models (trees) operating as an attribute will outperform any of randomly selected individual constituent models. According to Adeboye and Abimbola (2020), random forest has nearly the same hyperparameters as a decision tree or bagging classifier. The importance for each feature on the constructed decision tree is then calculated as:…”