Μοντέλα διάγνωσης και πρόβλεψης ρίσκου επιβίωσης για τον καρκίνο του πνεύμονα
Date Issued
June 11, 2024
Type
Πτυχιακή Εργασία
Abstract
We continuously see innovative applications and technologies from artificial intelligence and machine learning, which provide new "capabilities" to computing machines. An example is decision-making by programs fed with appropriate data, applications of which are used from businesses to hospitals for better diagnoses. These decision support systems are valuable tools for specialized personnel, especially in diagnosing diseases such as lung cancer. Today, there are many implementations for diagnosis through image recognition from CT and MRI scans, but few use simpler data. All available implementations focus only on diagnosis and not on assessing the patient's survival risk. Therefore, we decided to create our own system that addresses both diagnosis and the assessment of survival risk for more than one year after surgery. We studied the existing literature, found and processed data, and fed it into machine learning algorithms to produce models capable of accurate decisions. This system was integrated into a web application we developed for easy access by medical and non-medical personnel. Using the Extreme Gradient Boosting algorithm in Python, we created two models with an accuracy of 87% for cancer diagnosis and 66% for survival risk assessment. Finally, the web application was created using tools such as HTML, CSS, Javascript, and Flask, providing an easy-to-use decision support system.
Subjects