Data set on Employing Machine Learning Approaches to Detect Lung Cancer via Analysis of Clinical and Genetic Data

Lung cancer is most often caused by mutations in this gene, and several therapies are focused on targeting it specifically. Several feature selection methods, a change of linear and nonlinear, & ensemble projecting models for classification were utilised to categorize the binary result of wild-type and distorted EGFR alteration position. As shown by the findings, a comprehensive strategy that encompassed the lung with the nodule inside an ROI may better gather pertinent data and accurately forecast the status of the EGFR mutation than local nodule analysis.

The best-performing classifiers were determined using the Principal Component Evaluation feature selection method. These classifiers ranged from Linear Support Vector Machines, flexible the Internet, and Logistic Regression, and their Area Under the Curve (AUC) values ranged from 0.726 to 0.738. Applying a holistic approach, this technique demonstrates that further radio genomic studies should include data from broader lung regions, including the nodule, to provide a more complete characterisation of lung cancer.