How Is the Lung Cancer Incidence Rate Associated with Environmental Risks? Machine-Learning-Based Modeling and...

Area of Science:

• Public Health
• Environmental Science
• Data Science

Background:

• Lung cancer is a major public health concern, yet few studies adopt a holistic approach considering social, economic, and environmental factors.
• Existing research often focuses narrowly on clinical aspects, neglecting broader determinants of lung cancer incidence.

Purpose of the Study:

• To develop comprehensive predictive models for lung cancer incidence rates.
• To identify and analyze key social, economic, and environmental risk factors influencing lung cancer rates.
• To compare the efficacy of various machine learning algorithms in predicting lung cancer incidence.

Main Methods:

• Utilized stepwise regression and variance inflation factor for risk factor selection and multicollinearity elimination.
• Employed machine learning algorithms including linear regression, support vector regression, random forest, K-nearest neighbor, and cubist model tree with cross-validation.
• Conducted a case study in Taiwan to validate the predictive models.

Main Results:

• The cubist model tree, with feature selection, achieved the highest accuracy (RMSE: 3.310, R-squared: 0.960).
• Significant risk factors identified include average nitrogen dioxide (NO2) concentration, employment percentage, and number of factories, in addition to smoking.
• The random forest model effectively identified key contributing variables, with or without feature selection.

Conclusions:

• A holistic approach incorporating social, economic, and environmental factors improves lung cancer incidence prediction.
• The developed predictive models can aid in precise analysis and estimation of lung cancer rates for effective prevention.
• Identified risk factors provide a basis for future, comprehensive analyses of lung cancer incidence.