Copiloting Predictability with AI and ML to Beat the Second Hit in Colorectal, Breast and Lung Cancers

Overview

The most common lethal forms of cancers like colorectal, lung, and breast cancers will need early detection through precise molecular characterization, forsensitivity or specificity needed for optimal patient care. I used large-scale, real-world patient data to investigate epigenetic changes distinguishing normal from cancerous tissue and progression. I focused on three key tumor-suppressor genes: TP53 (many cancers), BRCA1 (breast cancer), and APC (colorectal cancer). Genomic and epigenomic datasets from The Cancer Genome Atlas (TCGA), were accessed through Genomic Data Commons (GDC), with colorectal, lung, and breast tissues. I studied ALU and LINE-1 (L1) elements, which are suppressed by DNA methylation levels, in and around TP53, BRCA1, and APC across samples and at different cancer stages. Lower DNA methylation values occurred much more frequently in cancerous samples, suggesting increased potential for genomic instability, fluctuating across progressing stages, as possible dynamic biomarkers. I developed three supervised machine learning models. First one classified normal versus cancerous tissue with high accuracy, with AUC values of 1.00 for colorectal (COAD), 1.00 for lung cancer (LUAD), and 0.93 for breast cancer (BRCA). Second one identified high-risk tumors, achieving AUC values between 0.81-0.86. Third one, identified possible cancer type (breast, colon, or lung) and if the sample was tumor or normal, with an overall accuracy of about 80%. Integrating cancer genomics with epigenetic analysis, machine learning can powerfully diagnose, identify cancer type, and reflect disease progression in colorectal, lung, and breast cancers. With further refinement, models could support community by earlier detection for better treatment planning.

Competition history

• CSEF 2026 Computational Science (Senior Division) · Entry S-07-12

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

• All Computational Sciences projects
• CSEF Computational Sciences projects
• Every CSEF 2026 project