Εφαρμογή της τεχνητής νοημοσύνης στο ηλεκτροκαρδιογράφημα

Abstract

Technological advancements have resulted in an unprecedented volume of data, prompting the need for effective utilization and extraction of valuable information. The internet's evolution, widespread broadband access, and the advent of applications, especially in social networks, have amplified data availability across various sectors like healthcare, communications, and education. Processing this vast data efficiently has led to the emergence of new methods, particularly knowledge mining and machine learning. In this context, the thesis delves into the intersection of machine learning, healthcare, and diagnostic tools, particularly focusing on electrocardiograms (ECG or EKG). The three main chapters explore machine learning methodologies, their applications in healthcare, and a practical application of deep learning in diagnosing arrhythmias using ECGs. Notably, the discussion covers deep learning techniques, neural networks, and their application to ECG analysis. The key takeaway from the case study is the comparison of XGBoost and Neural Networks algorithms, revealing that the XGBoost algorithm proves more reliable, achieving accuracy rates exceeding 90%. The findings emphasize the algorithm's effectiveness in healthcare applications, showcasing its potential for automated pattern recognition and decision-making processes. The thesis provides valuable insights into the symbiotic relationship between machine learning and healthcare, with implications for diagnostic tools and algorithmic reliability.