Identifying Lung Cancer Risk Factors in the Elderly Using Deep Neural Networks: Quantitative Analysis of Web-Based...

Area of Science:

• Gerontology
• Oncology
• Data Science

Background:

• Lung cancer poses a significant threat to the elderly population, with increasing morbidity and mortality.
• Understanding the complex pathogenesis and multifactorial risk factors is crucial for effective prevention and control.

Purpose of the Study:

• To identify key lung cancer risk factors in the elderly population.
• To quantitatively analyze the influence of these risk factors using deep learning.

Main Methods:

• Integrated multidisciplinary risk factors (behavioral, disease history, environmental, demographic) from web-based survey data.
• Trained deep neural network models on a large, stratified elderly population (n=235,673).
• Extracted and quantitatively analyzed lung cancer risk factors using the developed models.

Main Results:

• Deep learning models achieved high accuracy (0.927-0.962) and AUC (0.913-0.931) in identifying lung cancer risk factors.
• Smoking frequency was the primary risk factor for elderly men; time since quitting and lifetime smoking were key for elderly women.
• Elderly men exhibited the highest lung cancer incidence, particularly non-small cell lung cancer.

Conclusions:

• Developed a quantitative method for identifying elderly lung cancer risk factors.
• Models provide intervention indicators for lung cancer prevention, especially for older men.
• This approach can serve as a risk factor identification tool for other cancers and aid clinical decision-making.