Cooperative effects in feature importance of individual patterns: application to air pollutants and Alzheimer disease

AI can now identify "team effects" between multiple factors: Rather than treating each input (like air pollutants) as independent, this research shows how to measure when two or more factors work together synergistically to influence outcomes—in this case, Alzheimer's disease mortality. This reveals hidden relationships that standard AI analysis tools miss.

Location matters for understanding health risks: The study found that ozone (O₃) and nitrogen dioxide (NO₂) have particularly strong synergistic effects on Alzheimer's mortality in specific Italian provinces (Bergamo and Brescia), demonstrating that AI explanations need to account for local geographic and environmental contexts rather than applying one-size-fits-all conclusions.

Green spaces may buffer pollution effects: Urban green area density showed synergistic influence with air pollutants in predicting mortality rates, suggesting that environmental interventions like urban greening could be part of a multi-factor strategy to reduce health risks—a finding that wouldn't emerge from analyzing pollutants alone.

Better explainability for complex health decisions: By assigning three different scores to each data pattern (unique contribution, redundant effects, and synergistic effects), this framework provides more nuanced explanations of how AI models make predictions, which is crucial for building trust in AI systems used for public health policy.

Practical tool for One-Health research: The framework bridges environmental science, epidemiology, and AI, offering a replicable method for understanding how multiple environmental and health factors interact—applicable beyond this study to other complex health and environmental challenges.