Related Experiment Video
Updated: Jun 28, 2025
07:50
Genome-Wide Analysis of DNA Methylation in Gastrointestinal Cancer
Published on: September 18, 2020
5.5K
Deep learning and machine learning approaches to classify stomach distant metastatic tumors using DNA methylation
Jing Shi1, Ying Chen1, Ying Wang2
1Department of Medical Oncology, The First Hospital of China Medical University, Shenyang, China.
Computers in Biology and Medicine
|April 24, 2024
Summary
Machine learning models accurately predict stomach adenocarcinoma distant metastasis using DNA methylation. Deep neural networks achieved 99.9% AUC, identifying key methylation markers for improved cancer prognosis.
Area of Science:
- Oncology
- Genomics
- Bioinformatics
Background:
- Distant metastasis in stomach adenocarcinoma significantly worsens patient prognosis.
- DNA methylation changes are emerging as critical biomarkers for cancer progression.
- Early detection of metastasis is vital for effective stomach cancer treatment.
Purpose of the Study:
- To develop and compare machine learning and deep learning models for predicting distant metastasis in stomach adenocarcinoma.
- To identify key DNA methylation markers associated with metastasis in stomach cancer.
- To improve the accuracy of metastasis detection using genomic data.
Main Methods:
- Development of predictive models including deep neural networks (DNN), support vector machines (SVM), random forest (RF), Naive Bayes (NB), and decision tree (DT).
- Utilized DNA methylation profiles from stomach adenocarcinoma samples.
- Employed a weighted random sampling technique to handle imbalanced datasets.
Main Results:
- Deep neural networks (DNN) demonstrated superior performance compared to other models.
- Achieved Area Under the Curve (AUC) of 99.9% and Area Under the Precision-Recall Curve (AUPR) of 99.5% for metastasis prediction.
- Identified crucial methylation markers linked to significant genes in metastatic tumors.
Conclusions:
- Machine learning, particularly DNN, offers a highly accurate approach for predicting distant metastasis in stomach adenocarcinoma based on DNA methylation.
- The study highlights the potential of DNA methylation profiling for early cancer metastasis detection.
- Identified methylation markers can aid in understanding tumor biology and developing targeted therapies.
