Can Explainable AI Assess Personalized Health Risks from Indoor Air Pollution?

Indoor air pollution is a major health blind spot: A survey of 143 participants revealed that most people have limited awareness of indoor air pollution despite daily exposure to health risks, suggesting the public significantly underestimates this threat compared to outdoor air pollution.

Explainable AI can identify pollution sources with high accuracy: The research achieved 99.8% accuracy using Decision Trees combined with interpretability tools (LIME and SHAP) to pinpoint specific indoor pollution sources like cooking, smoking, incense, and AC usage—moving beyond simple monitoring to actionable source identification.

Personalized health risk assessment is now feasible: By collecting continuous 24-hour data on individual activities and pollution levels, the system achieved 91% accuracy in predicting personal pollution exposure, enabling tailored recommendations for reducing each person's specific indoor air quality risks.

Explainability matters for practical adoption: The use of interpretable AI models (LIME and SHAP) rather than "black box" algorithms allows people to understand why certain activities cause pollution spikes, making it more likely they'll trust and act on the system's recommendations for behavior change.

Common household activities are major pollution culprits: The research identified everyday behaviors—inadequately ventilated cooking, indoor smoking, excessive AC usage, and incense burning—as significant indoor pollution sources, suggesting simple behavioral changes could substantially improve home air quality.