PREDICTIVE MODELING USING COMMUNITY SCIENCE: INSIGHTS INTO HOUSEHOLD LEAD

Community science is rapidly growing throughout the United States (U.S.), particularly in the realm of metal contamination. While community science has been shown to improve scientific communication and aid in remediation activities, little has been done to synthesize results at a national scale to aid in pollution prevention and remediation. Using hundreds of household dust samples collected via community scientists across the U.S. with matching questionnaire responses and bulk lead (Pb) data as part of the “DustSafe” initiative, we have developed a model that predicts whether house dust is elevated (≥ 80 mg/kg) with Pb 75% of the time. This is especially pertinent because Pb in household dust is one of the best predictors of elevated blood Pb levels (BLLs) in children, which is detrimental to their health. In order to promote household Pb screening, we have incorporated this model into an online app, to help people determine whether or not their home might be at risk for high dust Pb. We also tested our model using an equivalent Australian dataset (n = 1,254), which showed 76% predictive accuracy. Consequently, we plan to expand our app to other regions throughout the globe as part of our international “DustSafe” partnership, which will include updating our online app and predictive model to be more region specific, particularly because of differing Pb legacies in many countries relative to the U.S. and Australia.