Related Experiment Video
Updated: Jul 15, 2025
08:14
MicroRNA Based Liquid Biopsy: The Experience of the Plasma miRNA Signature Classifier MSC for Lung Cancer Screening
Published on: October 26, 2017
15.7K
Enhancing Lung Cancer Classification through Integration of Liquid Biopsy Multi-Omics Data with Machine Learning
Hyuk-Jung Kwon1,2, Ui-Hyun Park1, Chul Jun Goh1
1Eone-Diagnomics Genome Center, Inc., 143, Gaetbeol-ro, Yeonsu-gu, Incheon 21999, Republic of Korea.
Cancers
|September 28, 2023
Summary
Early lung cancer detection is improved using machine learning and multi-omics blood analysis. This approach analyzes cell-free DNA (cfDNA) and cancer markers for accurate diagnosis.
Area of Science:
- Oncology
- Bioinformatics
- Genomics
Background:
- Early lung cancer detection significantly impacts patient survival and treatment efficacy.
- Liquid biopsy using cell-free DNA (cfDNA) analysis, including next-generation sequencing (NGS), can detect cancer-specific alterations.
- Machine learning (ML) offers a promising avenue for identifying complex patterns in cancer data.
Purpose of the Study:
- To develop and evaluate an ML-based diagnostic model for lung cancer detection using multi-omics blood data.
- To assess the diagnostic performance of ML models utilizing cancer markers, cfDNA concentrations, and copy number variations (CNVs).
- To determine if combining multi-omics data enhances diagnostic accuracy compared to individual markers.
Main Methods:
- Blood samples were collected from 92 lung cancer patients and 80 healthy controls.
- Lung cancer markers (Cyfra21, CEA) were quantified, alongside cfDNA concentrations and CNV screening.
- ML algorithms including Adaptive Boosting (AdaBoost), Multi-Layer Perceptron (MLP), and Logistic Regression (LR) were employed for analysis.
- Area Under the Curve (AUC) values were calculated to evaluate model performance.
Main Results:
- Significant differences in Cyfra21 and CEA levels were observed between lung cancer patients and healthy individuals.
- ML models incorporating cfDNA concentration and CNV screening demonstrated diagnostic utility.
- Combining multi-omics data in ML analysis yielded higher AUC values than analyzing individual components, indicating improved accuracy.
Conclusions:
- ML analysis of multi-omics data from blood samples shows a strong capability to differentiate lung cancer patients from healthy individuals.
- The developed model holds potential for a highly accurate, non-invasive diagnostic tool for lung cancer.
- Integrating diverse data types (multi-omics) is crucial for enhancing the precision of ML-based cancer diagnostics.
Keywords:
cell-free DNAcopy number variationgenomicsliquid biopsylung cancermachine learningmulti-omicsnon-invasivetumor marker
