Assisted artificial intelligence medical diagnosis system for heart disease.
Date
2023
Authors
Maambo, Mweemba
Journal Title
Journal ISSN
Volume Title
Publisher
The University of Zambia
Abstract
In recent years, the expanding array of innovative applications in the medical domain has been instrumental in propelling research forward. Among these advancements, Artificial Intelligence (AI) systems have emerged as influential tools, significantly contributing to the development of various medical applications and tools. However, heart disease remains a pressing health concern globally, highlighting the critical need for accurate diagnosis to enable effective regulation and intervention. This study focuses on harnessing the capabilities of an AI system to enhance the diagnostic process for heart disease. By leveraging input medical data sourced from a well-established dataset on Kaggle, the developed AI application is tailored specifically to cater to the demographic characteristics of Zambian patients. The primary objective is to evaluate the model's predictive accuracy when applied to medical data from the Zambian population. To facilitate this assessment, 80% of the collected dataset is allocated for training purposes, with the remaining 20% reserved for testing. Central to the prediction process is the utilization of a Bayesian data-mining algorithm, which plays a pivotal role in forecasting the risk level and likelihood of heart disease. An extensive array of medical parameters, including blood sugar levels, sex, heart rate, age, cholesterol levels, blood pressure, presence of exercise-induced angina, ST-slope, oldpeak, resting electrocardiogram results, and chest pain type, serves as the foundation for predicting heart disease in patients. Following a meticulous pre-processing phase, supervised learning techniques are employed to craft a robust prediction model. The outcomes of this process reveal a commendable prediction accuracy of 90.97%. Comparative analysis with established algorithms such as KNN, Random Forest, and Decision Tree algorithms further validates the efficacy of the proposed AI system in medical diagnosis. This study not only emphasizes the effectiveness of AI systems in medical diagnosis but also contributes valuable insights to the ongoing efforts to combat heart disease. The integration of data mining, artificial intelligence, and predictive modeling presents a promising avenue for advancing healthcare practices and outcomes, particularly in regions like Zambia grappling with cardiovascular health challenges. Achieving an 89% accuracy rate demonstrates the model's ability to adapt to Zambian patients' traits, while incorporating real-world data from the National Heart Hospital enhances its credibility. This validation underscores the potential of AI-driven diagnostic systems to improve healthcare and patient outcomes.
Keywords: Heart Disease, Artificial Intelligence, Bayesian Classification, Prediction Model, Supervised Learning Techniques, Data Mining Algorithms
Description
Thesis of Master of Science in Computer Science