Interpretable AI for Analysing Spatial Patterns of Lung and Bronchus Cancer Mortality in the Contiguous USA
Abstract
Research investigates how interpretable artificial intelligence (AI) models evaluate the geographical variation of death rates caused by lung and bronchus cancer (LBC) throughout the contiguous United States. Planned public health strategies require a thorough understanding of geographic mortality patterns to achieve proper policy development. This study combines AI methods to determine essential mortality components alongside methods which support transparent and interpretable model structures. We use spatial data analysis together with AI strategies involving decision trees and SHAP values along with clustering approaches to discover patterns in geographic areas. The analysis reveals major inequality in LBC death rates that indicates residential regions with elevated mortality patterns related to cigarette use and healthcare service quality and surrounding environmental issues. The study enables effective intervention planning through interpretable AI by revealing specific regions requiring targeted preventive measures. The study indicates that AI applications can boost both epidemiological analysis and evidence-based policy developments thereby boosting cancer prevention and clinical care practices throughout the United States.