Predicting and Prioritizing Interventions to Prevent Childhood Lead Poisoning from Lead Paint
Andrew Goldberg, Attorney at the New York City Coalition to End Lead Poisoning (NYCCELP)
Andrew Faciano, Assistant Commissioner at the New York City Department of Health and Mental Hygiene (DOHMH)
MENTOR:
Daniel B. Neill, Professor of Computer Science, Public Service, and Urban Analytics at CUSP at NYU Tandon; NYC Courant Institute; and NYU Wagner
Authors
Anne Driscoll, Harrison Boyer, Michael Okoro
Research Question
Can a predictive model using machine learning and large-scale datasets effectively identify which residential buildings in NYC pose the highest risk of lead exposure to children?
Background
In 2023, 5,078 children under age six in NYC were diagnosed with lead poisoning, primarily due to exposure to residential lead paint. Lead exposure causes serious harm, including brain damage, developmental delays, and behavioral issues. Young children are especially vulnerable to lead’s effects. Citywide, an estimated 300,000 apartments with young children are in 150,000 buildings built before 1960, where lead paint likely remains. Tens of thousands of NYC children are at risk.
In 2004, a new law required landlords to regularly inspect apartments housing children under age six, and mandated stricter enforcement by the City's Housing and Health Departments. Following these measures, the number of lead poisoning cases declined steadily from 2004 to 2018. However, since 2018 the rate of decline has slowed, and the number of cases has fluctuated around 5,000 annually.
Methodology
This project entails the creation of a predictive model utilizing machine learning to identify buildings with the highest lead exposure risks. It leverages numerous large datasets containing information on residential building age, construction, and violation histories; resident demographics; and child health statistics. For instance, data is made available by the NYC Department of Health and Mental Hygiene (DOHMH) reflecting buildings where children are known to have been exposed to lead from lead paint. Other datasets include records of 14.7 million tenant complaints about housing conditions and records of over 777,000 violations related to deteriorating paint conditions across 56,000 buildings.
Deliverables
- Summary report
- Project website
- Documentation of Python code for data preprocessing and predictive modeling of childhood lead poisoning risk
- Interactive dashboard featuring a map that highlights high-risk buildings and hot spots for internal use by DOHMH
Data
- Housing Maintenance Code Violations provided by the Department of Housing Preservation and Development (HPD) on NYC Open Data
- Housing Maintenance Code Complaints and Problems provided by the Department of Housing Preservation and Development (HPD) on NYC Open Data
- NYC Housing and Vacancy Survey (NYCHVS) provided by the Department of Housing Preservation and Development (HPD)
- Buildings Selected for the Underlying Conditions Program provided by the Department of Housing Preservation and Development (HPD) on NYC Open Data
- Buildings Selected for the Alternative Enforcement Program (AEP) provided by the Department of Housing Preservation and Development (HPD) on NYC Open Data
- Buildings Selected for the Heat Sensor Program (HSP) provided by the Department of Housing Preservation and Development (HPD) on NYC Open Data
- Buildings Subject to HPD Jurisdiction (Open Data) provided by the Department of Housing Preservation and Development (HPD) on NYC Open Data
- Lead poisoning data provided by DOHMH on the NYC.GOV Environment & Health Data Portal
- Primary Land Use Tax Lot Output (PLUTO) and Property Address Directory (PAD) provided by the NYC Department of City Planning
- Lead Service Line Location Coordinates provided by the Department of Environmental Protection (DEP) on NYC Open Data