Machine Learning Model Construction and Testing: Anticipating Cancer Incidence and Mortality

Area of Science:

• Environmental health
• Oncology
• Data science

Background:

• Escalating environmental challenges correlate with rising cancer incidence.
• Accurate prediction of cancer incidence and mortality is crucial for public health policy.
• Machine learning offers a novel approach to understanding cancer dynamics.

Purpose of the Study:

• To develop and evaluate a machine learning framework for predicting cancer incidence and mortality rates.
• To assess the performance of various machine learning algorithms on a comprehensive cancer dataset.
• To inform public health policy and sustainable healthcare planning through accurate cancer projections.

Main Methods:

• Utilized a dataset of 72,591 records with variables including age, case count, population size, race, gender, site, and year of diagnosis.
• Employed diverse machine learning algorithms: decision trees, random forests, logistic regression, support vector machines, and neural networks.
• Analyzed model performance based on testing accuracies.

Main Results:

• Testing accuracies achieved were 62.17% (decision trees), 61.92% (random forests), 54.53% (logistic regression), 55.72% (support vector machines), and 62.30% (neural networks).
• Neural networks and decision trees demonstrated the highest predictive accuracy among the tested models.
• The framework provides a robust capability for projecting future cancer incidence and mortality.

Conclusions:

• The developed machine learning framework enhances understanding of cancer dynamics and enables meticulous projections.
• This approach supports researchers and policymakers in making informed decisions for public health.
• Application of this framework promotes sustainable healthcare planning by considering long-term ecological and societal impacts.