Machine Learning-Based Lung Cancer Diagnosis

Abstract

Cancer is one of the leading health problems occurring in various organs and tissues of the body and its incidence is increasing in the world. Lung cancer is one of the deadliest types of cancer. Due to its worldwide prevalence, increasing number of cases and deadly consequences, early detection of lung cancer, as with all other cancers, greatly increases the chances of survival. As with all other diseases, the diagnosis of cancer becomes possible after the appearance of various symptoms through the examinations of specialists. The recognizable symptoms of lung cancer include shortness of breath, coughing, wheezing, jaundice in the fingers, chest pain and difficulty swallowing. The diagnosis is made by an expert on site based on these symptoms and additional tests. The aim of this study is to detect the disease at an earlier stage based on the symptoms present, to assess more cases with less time and cost, and to achieve results in new situations that are as successful or even faster than those of human experts by deriving them from existing data using various algorithms. The goal is to develop an automated model that can detect early-stage lung cancer based on machine learning methods. The developed model includes 9 different machine learning algorithms (NB, LR, DT, RF, GB, SVM). The success of the classification algorithms used was evaluated using the metrics of accuracy, sensitivity and precision calculated with the parameters of the confusion matrix. The results obtained show that the proposed model can detect cancer diagnosis with a maximum accuracy of 91%. The application of this model will help medical practitioners to develop an automated and reliable system that can detect lung cancer. The proposed interdisciplinary method can also be applied to other types of cancer.