DNA Methylation Data Model for Early Lung Cancer Detection Using Decision Tree and Artificial Neural Network Methods Hendri Karisma 1,2, Astri Lestari 3
1 Informatics Engineering Departement, STMIK Tazkia, Bogor, West Java, Indonesia.
2 Tech Dept, Jejakin.com (PT. Jejak Enviro Teknologi), Jakarta, DKI Jakarta, Indonesia.
3 Graduate School of Master Program in Medical Sciences, Faculty of Medicine, Universitas Islam Bandung, Bandung, West Java, Indonesia
Abstract
Lung cancer has a relatively high incidence rate globally, with approximately 2.5 million cases according to the Global Cancer Observatory in 2022. In Indonesia, the mortality rate for lung cancer is 34,339 deaths out of 66,271 cases. The method commonly used for lung cancer screening is the Low-Dose CT-Scan, although it has low accuracy with a false positive rate 22%-93%. DNA methylation as a biomarker has become a promising alternative. Several studies have been conducted, and it has relatively high accuracy in detecting and diagnosing lung cancer non-invasively. Therefore, in this study, an experiment was conducted using DNA methylation data modeling for early detection of lung cancer, using more than 450,000 CpG sites from various genes on 23 human chromosomes. The data used to build the model came from the NCBI Geo GSE66836 from the Department of Cancer Genetics, Institute for Cancer Research, Oslo University Hospital. The sample consisted of 164 lung cancer tumor positives and 19 normal condition. The model was built using three different machine learning methods, and an analysis was conducted on the three resulting models. The methods used, along with their accuracy and the dominant gene in detecting lung cancer, were Decision Tree with 89% using RAB3B with 2 CpG sites, XGBoost with 91% using RAB3B with 1 CpG site and AC006156.5 and FAM197Y1 with 1 CpG site, and Artificial Neural Network with 94%. For the NeuralNetwork model, all CpG sites were utilized, and the top 5 genes are OTX1, HOOK2, MCIDAS, CDHR5, and SCT.
Keywords: Lung Cancer, DNA Methylation, Modeling, Machine Learning
