Data mining-based lung cancer diagnostic models and high-risk warning for occupational group
Yaru Chai1, Huijie Yuan1, Shuyin Duan1
1College of Public Health, Zhengzhou University, Zhengzhou, 450001, China.
Scientific Reports
|April 1, 2026
Summary
This study identified key protein markers for early lung cancer detection in coke oven workers. Machine learning models effectively screened high-risk individuals, improving early warning capabilities.
Area of Science:
- Oncology
- Biomarker Discovery
- Proteomics
Background:
- Lung cancer screening is crucial for improving patient survival rates.
- Coke oven workers represent a high-risk population for lung cancer.
- Identifying novel biomarkers and risk factors is essential for early detection.
Purpose of the Study:
- To investigate lung cancer risk factors in coke oven workers.
- To develop and validate screening models for early lung cancer detection.
- To identify high-risk individuals within the coke oven worker population.
Main Methods:
- Proteomic analysis of lung cancer and normal control groups.
- Establishment of mouse lung cancer and cellular malignant transformation models.
- Application of data mining and machine learning (Support Vector Machine, C5.0) for model construction using protein markers and epidemiological data.
Main Results:
- Dysregulated expression of CDH1, CLEC3B, CLU, sCD146, and VIM was associated with lung cancer.
- Support Vector Machine and C5.0 models demonstrated superior performance in screening.
- The developed models successfully identified 13 high-risk individuals among coke oven workers.
Conclusions:
- CDH1, CLEC3B, CLU, sCD146, and VIM are potential biomarkers for lung cancer screening.
- Machine learning models integrating proteomic and epidemiological data are effective for identifying high-risk individuals.
- This approach can enhance early warning systems for lung cancer in occupational settings.
