Biomarker Discovery in Lung Adenocarcinoma Using a Machine Learning-Based Approach through Metabolomic Data

Lung adenocarcinoma (LUAD) is the most common form of primary lung cancer, representing approximately 40% of cases. In Chile, the high mortality rate associated with lung cancer is largely due to delays in diagnosis, highlighting the urgent need for earlier detection methods. This study aims to address this challenge by identifying biomarkers for early detection of LUAD through metabolomic analysis. Metabolomics, the large-scale study of metabolites within biological samples, offers a powerful approach for uncovering molecular signatures specific to LUAD. By examining the metabolome, which includes allsmall molecules involved in metabolic processes, we aim to gain insights into the pathogenesis of LUAD and identify potential biomarkers indicative of early-stage disease. For this study, we utilized the Lung Cancer Metabolome Database (LCMD), which is a comprehensive repository of metabolomic data, particularly relevant for cancer research. From this database, we selected 65 sample projects, each representing individual cases of LUAD, to ensure a robust and representative analysis. The data were analyzed usingadvanced machine learning algorithms, including Support Vector Machines (SVM) and Random Forest, which were specifically chosen for their ability to handle complex, high-dimensional data. Our in-house scripts were developed to preprocess the data, perform multivariate analyses, and identify patterns within the metabolomic profiles.The identification of these biomarkers could lead to the development of novel diagnostic strategies, which may be implemented in clinical settings to facilitate earlier detection of LUAD. By enabling diagnosis at earlier stages, these strategies have the potential to significantly improvepatient prognosis and increase survival rates. Acknowledgement: Departamento de Informática y Computación, UTEM; Escuela de Informática, UTEM: Laboratorio de Investigación Aplicada, Departamento de Informática y Computación, UTEM. This work was supported in part by Project supported by the “Competition for Research Regular Projects, year 2023, codeLPR23-09 and “Competition for Research Assistant Funding UTEM”, year 2023, code AI23-06, Universidad Tecnológica Metropolitana (AM-B)