Sopiyan Apandi, Nanang Nanang, Perani Rosyani, Nuraina Nuraina, Saiyah Awaliyah, Santi Ayu Purnamawati, Vera Oktaviani

Classification of Lung Diseases Using the Desicison Tree Method

Introduction

Classification of lung diseases using the desicison tree method. Detect lung diseases early with machine learning. This study classifies lung conditions using Decision Tree (98.39% accuracy) and GKNN, enhancing medical diagnosis & patient life expectancy.

Abstract

This research develops a machine learning-based classification method to detect lung cancer early, with the aim of increasing the life expectancy of patients. Lung cancer diagnosis generally requires manual interpretation of CT (Computed Tomography) images, which is prone to human error and time consuming. The proposed method is the Genetic K-Nearest Neighbor (GKNN) algorithm, which combines the advantages of genetic algorithms in parameter optimization with the K-Nearest Neighbor (KNN) approach for classification. The dataset used comes from Kaggle, includes 309 entries with 16 features, and has gone through pre-processing to minimize noise and improve data quality. The GKNN model was compared with other algorithms such as Support Vector Machine (SVM), Random Forest, Artificial Neural Network (ANN), and Decision Tree. Results show that Decision Tree achieves the highest accuracy of 98.39%, while GKNN offers a reliable solution with 90% accuracy and a low false positive rate. The results of this study are expected to be a reference in the development of artificial intelligence-based systems for medical applications, supporting faster and more accurate clinical decision-making.

Review

This research addresses a critical and highly relevant challenge: the early detection of lung cancer to improve patient prognosis. The authors aim to overcome the inherent limitations of manual CT image interpretation, such as susceptibility to human error and time consumption, by proposing a machine learning-based classification method. The core contribution outlined is the Genetic K-Nearest Neighbor (GKNN) algorithm, which intelligently combines genetic algorithms for parameter optimization with the KNN approach for classification. This objective is commendable, as advancements in this area hold significant promise for transforming clinical diagnostic workflows and enhancing patient care. The methodology involves training and evaluating the proposed GKNN model against several established machine learning algorithms, including SVM, Random Forest, Artificial Neural Network, and Decision Tree. A Kaggle dataset comprising 309 entries with 16 features, subject to pre-processing, was utilized. Interestingly, while GKNN is presented as the proposed method, the abstract highlights that the Decision Tree algorithm achieved the highest accuracy of 98.39%, outperforming GKNN's 90% accuracy. The GKNN is noted for its "reliable solution" and "low false positive rate," although specific metrics beyond the 90% accuracy are not provided for it. This discrepancy between the title, which explicitly names "Decision Tree Method," and the abstract's emphasis on GKNN as the *proposed* method, warrants clarification. While the study's ambition to improve lung cancer diagnosis is laudable, and the reported high accuracy for Decision Tree is impressive, several aspects warrant further attention. The most significant point of confusion is the title, which suggests Decision Tree is the primary method, yet the abstract positions GKNN as the proposed innovation, with Decision Tree emerging as the best performer among *comparative* methods. This inconsistency requires clarification to align the paper's focus. Additionally, a dataset of 309 entries, while pre-processed, might be considered relatively small for robust and generalizable deep learning models in complex medical imaging scenarios. For clinical application, external validation on larger, more diverse datasets, and detailed performance metrics beyond accuracy (e.g., sensitivity, specificity, F1-score for GKNN's "low false positive rate") would significantly strengthen the findings and better support the claim of faster and more accurate clinical decision-making.

Full Text

Comments

You need to be logged in to post a comment.