Applications of artificial intelligence and machine learning in healthcare: A cardiovascular case
Open Access
- Author:
- Alamoudi, Hassan
- Graduate Program:
- Industrial Engineering
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- November 29, 2019
- Committee Members:
- Soundar Rajan Tirupatikumara, Thesis Advisor/Co-Advisor
Robert Carl Voigt, Program Head/Chair
Hui Yang, Committee Member
Robert Carl Voigt, Special Signatory - Keywords:
- Artificial Intelligence
Machine Learning
Healthcare
Cardiovascular - Abstract:
- Artificial Intelligence (AI) is becoming a ubiquitous term that is used in many fields of research or the popular culture. Among these fields that was affected by this hype is the healthcare sector. Along with its subdomain, Machine Learning (ML), they established an environment of interest in the promises of machines versus humans’ capabilities. Though artificial intelligence applications in healthcare such as interpreting ECGs could date back to the mid of the twentieth century, the promises of AI still at its beginning when it comes to new breakthroughs. This is due to the transformation into a digital world and new advancements in the processing capabilities. Computer vision has contributed the most to the healthcare sector where it can leverage doctors and practitioners with automated classification and annotations as a preparing step. This kind of mechanism is the best suited for applications of AI in healthcare. However, the amount of data in other forms such as textual or lab results is exceeding the force power. While a solution could be to use machines to learn and propose solutions, the results could be catastrophic and human lives are on stake. So, explainable AI could be beneficial where it analyzes and makes predictions that can be trusted by the users. The study here is conducted on cardiovascular patients’ dataset to predict the presence or absence of the disease. Classifications techniques used include Naïve Bayes, Logistic Regression, Decision Trees, Support Vector Machines, and Artificial Neural Networks. The Logistic regression model achieved the best Area under the curve. Moreover, an extension of the previous studies discussed is conducted to explain the model and to show how models of AI can be trusted and not used as black-boxes.