Ideas & Insights / All Perspectives / Ideas & Insights

Unleashing the Power of Machine Learning to Secure Safe Drinking Water

Ian Robinson — President and Chief Operating Officer, BlueConduit
Terrell Seabrooks — Former Program Associate, Innovation, The Rockefeller Foundation

In the United States, over 10 million homes receive drinking water from toxic lead pipes. Low-income and minority households are at increased risk for harms of lead contamination, perpetuating decades of social, health and environmental injustices. Progress towards addressing these lingering problems has been slow, but new technologies offer hope for speeding up lead pipe removal and replacement.

Lead contamination causes up to 400,000 deaths per year. According to the EPA, about 20% of lead exposure comes from lead in drinking water. Health effects of lead exposure, include neurological damage, developmental delays, and the interference of function of major organs. These primarily affect children.

Albeit slowly, the federal government has recently taken action to replace lead pipes throughout the United States. In 2022, the Environmental Protection Agency allocated $3 billion to states, Tribes and Territories for lead service line replacement. Close to half (49%) of these federal funds will go to disadvantaged communities as grants or forgivable loans.

But while additional funding is desperately needed, money alone can’t replace lead pipes at the speed their toxic affects require. Real, rapid progress requires working more effectively and efficiently. That’s where technology can help.

The cost of correctly identifying, excavating, and replacing contaminated lead pipes in a community can cost as much as $3 million per 1,000 excavations. For smaller and under-resourced communities, this is often a cost they are unable to accommodate, even with increased funding from the federal government.

For the past two years, The Rockefeller Foundation and BlueConduit, a water analytics software company, have worked together to apply opensource data analytic tools and interactive mapping to identify homes with lead pipes, providing precise details to city contractors on the right place to begin digging, and expand community education on how to take action, all at a fraction of the previous cost.

Let’s look at how technology helped Trenton Water Works to replace its lead water service lines. In Trenton 50% of service lines are comprised of lead. This is where BlueConduit’s tech can be most valuable. The tech can help officials locate areas with a high concentration of lead service lines so replacement teams can focus efforts in those areas. This tech and predictive modeling are important cost-cutting measures for officials. With Trenton’s need to be more cost-effective due to budget constraints, this makes a difference.

BlueConduit’s tool isn’t limited to improving precision and efficiency to just water lines. It has been leveraged with other infrastructure projects in future planning and even for strategic marketing for homeowners.

“We hope to be using BlueConduit’s lead service line predictions to prioritize our road paving operations and to coordinate lead service line removal with other planned city infrastructure projects,” adds Ms. Epstein. “We could even leverage BlueConduit’s predictive data to launch outreach Marketing to neighborhoods with lead service prevalence so we can invite those homeowners to participate in our replacement program.”

Recognizing the urgency of lead pipe replacement, the burden it places on communities of color, and the positive impact that technology has on improving efficiency, The Rockefeller Foundation has committed nearly $1M to help further scale BlueConduit’s machine learning analysis initiative.

Using data-driven technology to efficiently allocate resources—while also addressing critical infrastructure needs—is possible and necessary is targeting funding and support to the areas that need it most. All sectors must work together to bring funding, technology and will power to the fight against injustice. Our lives depend on it.