Enhancing the Efficiency of Lung Disease Prediction using CatBoost and Expectation Maximization Algorithms

Due to rapid lifestyle changes, people are now more inclined to participate in behaviors that might seriously impair their health. The ongoing change hikes stress levels, for which people are settling into habits that may help them reduce stress. However, extraneous factors such as occupational hazards and allergies might also lead to health issues. These practices shall ultimately lead to one of the deadliest diseas es in the world i.e., lung cancer. Assessing the causes and symptoms of the same becomes a necessity. Researchers have been using machine learning approaches, as they are one of the fast-growing niches in this field. This article delves into predicting which symptoms are highly correlated to lung cancer, using supervised ML algorithms Logistic Regression and Decision Tree. Results show that it is vital to boost the performance of these models, for which two ML boosting algorithms, CatBoost and Expectation Maximization are employed. The results suggest that when using the aforementioned boosting frameworks, the accuracy of the model significantly increases from 93.4% to 99.8
^%
.