Our work in health has mostly focused on developing preventative tools for adverse public health outcomes. Over the last few years, we’ve worked with several government agencies and non-profits on problems ranging from maternal mortality (Government of Mexico), adverse births (Illinois Department of Human Services), home visitation programs for pregnant women (Nurse Family Partnership), preventing lead poisoning (with Chicago Department of Public Health), and increasing retention for HIV patients (with UChicago and Chicago Department of Public Health).
A large focus of our work in health has been improving outcomes for communities that are underserved and vulnerable. We are determined to design data-driven strategies to identify individuals at risk of various adverse health outcomes and inform preventative interventions based on individual, demographic, behavioral, social and physical environmental factors.
Targeting Preventative Home Inspections to Reduce Childhood Lead Poisoning
Lead poisoning is a major public health problem that is irreversible and affects hundreds of thousands of children in the United States every year, leading to poor health and educational attainment outcomes. A common approach to identifying lead hazards is to test all children for elevated blood lead levels and then investigate and remediate the homes of children with elevated tests. This can prevent exposure to lead of future residents, but only after a child has been poisoned. DSaPP has been working with the Chicago Department of Public Health (CDPH) over the last three years to to develop a machine learning-based predictive system that can identify homes that are most at risk of having life-threatening lead hazards for young children. This system then informs preventative lead hazard inspections as opposed to the current approach of post-poisoning inspections.
Over a three-year period (2015-2017), we developed a predictive model that is accurate at identifying and prioritizing homes with lead hazards that lead to children getting elevated blood lead levels. The model went through a validation pilot in 2017 and we’re currently analyzing the results of the pilot. In parallel, we’re working with Chicago Department of Innovation and Technology to deploy the system inside the city so it can be used operationally by the health department to target proactive inspections. Another implementation project underway is working with hospital systems to provide them with an API that they can call when a pregnant women comes for her appointment. The predictive model will assess the risk of the future child getting lead poisoning, and this risk score will then be used to trigger a lead hazard inspection by CDPH.
The system we’ve created is open-source and with the goal of getting used nationwide to predictively inform the deployment of scarce public resources to address lead hazards accurately and proactively rather than the current approach of remediating after lead-exposed children have been identified.
More information about the project is available here.