Recently the project INnovations to generate estimates of children’s soil/dust inTakes (INGEST) received a $1.35 million research grant from the U.S. Environmental Protection Agency (EPA). A team of researchers from Hopkins and the University of California, San Francisco will quantitatively assess children’s dust and soil exposures using a set of novel research approaches.
Keeve Nachman, associate professor of environmental health and engineering at the School of Public Health, is the principal investigator of Project INGEST. In an interview with The News-Letter, he explained the significance of Project INGEST: EPA’s Superfund Program is responsible for cleaning up highly contaminated sites to redevelop them for future uses, and Project INGEST is able to provide quantitative insight into the extent to which people would be exposed to contaminants.
“We don’t have good data for soil and dust exposures because it is very hard to quantify just how much kids get in their mouths; however, soil and dust ingestion tends to be the most important pathway in many Superfund risk assessments,” he said. “Our goal is to tackle this barrier and to provide data that support our previous assumptions about soil and dust exposure.”
Sara Lupolt, postdoctoral fellow at the School of Public Health and a member of INGEST, expressed her excitement about the project’s potential impacts in an interview with The News-Letter.
“I’m an agricultural worker and am interested in soil and dust ingestion, which I studied for my dissertation, so it is really great to get this opportunity to pursue this research question in this highly exposed and vulnerable population,” she said.
According to Nachman, Project INGEST consists of three objectives. The first is to understand the natures and sites of the major activities children engage in. For eight days out of a year, the team will collect “time-activity diaries” that entail all the activities of children from waking up to going to bed. The diaries will be provided by the recruited caregivers of 150 children from ages six months to 6 years old in Baltimore and San Francisco.
Nachman elaborated on the reasoning behind this.
“The details about each activity would allow us to construct a temporal history of how children spend their time, which then gives us an insight about which activities involve a lot of soil and dust contact,” Nachman said.
The second objective is to identify which activities are the causes of high soil and dust exposure. Through macro-activity assessment, the team will quantify the time children engaged in hand-to-mouth or object-to-mouth activities.
Quantifying macro-activity is extremely labor intensive, involving recording videos behind children throughout the day, so previous studies involve videotaping only a few children. The team partnered with Alan Yuille, Bloomberg distinguished professor and computer vision expert, and Kristin Voegtline, assistant professor at the School of Medicine and a behavioral psychobiologist, to approach macro-activity quantification.
Nachman explained how the data will be collected.
“We will collect videos of children in their normal activities. Then, the computer will adopt an artificial intelligence approach to do the macro-activity counting. To make sure this method works properly, we will conduct the traditional behavioral coding of the same videos and compare the computational approach to the traditional one,” he said.
He pointed out that the comparison will serve to validate computer vision’s viability in macro-activity quantification, and that it would be a major step toward understanding those macro-activity behaviors if the computational approach is proven to be effective.
The third objective is to pinpoint the chemicals children are exposed to. Past studies looked at biological samples, such as stool samples, to identify the metal element tracers that are also present in dust and soil. This method is used to estimate the extent to which children are exposed to dust but is insufficient to give quantitative proof.
Nachman described INGEST’s approach.
“Our approach relies on organic chemical concentration instead of elemental concentration,” he said. “We will use a technique called ‘non-targeted analysis,’ developed by our colleague Carsten Prasse, to analyze an array of chemicals, many of which we didn’t set up to look for.”
In the past, the targeted chemicals had to be known prior to discovery, but this novel approach in INGEST eliminates this stage and points toward a diagnosis of what is present. In addition, non-targeted analysis is capable of returning results on thousands of chemicals in each sample.
The team will use this technique to identify signature chemicals in biological and environmental samples, including soil, dust, formula, stool and urine to determine which chemicals derive from dust and soil. Both Nachman and Lupolt are hopeful that the team will identify either individual chemicals or groups of chemicals that will help estimate the extent to which a child is exposed to soil or dust.
Nachman hoped the results from Project INGEST will improve the EPA’s current risk assessment methods for children’s exposure to chemicals, soil and dust.
Lupolt added on that she was excited to see the change in research approaches in the field of public health.
“I hope Project INGEST will inspire other researchers to continue taking multidisciplinary approaches, thinking outside the box and looking for techniques that scientists hadn’t used before. I also hope to see more uses of computer vision — there are many other applications for solving problems beside estimating total soil and dust ingestion,” she said.
An integral team member in the designing of Project INGEST, Lupolt shared her challenges during the process.
“In the initial stage of developing this project, the biggest challenge was to find other researchers who were able to see our vision and willing to be a part of it, but we ended with a great team,” she said. “I think persistence is an important element in finding people doing similar works and with the skillsets we needed. We were then able to overcome many challenges thanks to the awesome team we have.”
Looking forward, Nachman and Lupolt hope to keep exploring the various applications of computer vision.
“One thing we will do is to think about other applications in environmental and behavioral health sciences. Human behaviors are sometimes very difficult to quantify, and the idea of using computers to help us track those behaviors is really exciting and appealing,” Nachman said. “The use of non-targeted analysis to identify chemicals we never looked at before is a great opportunity to take a closer look at those and make better sense of what they might mean for health.”
Nachman noted that Project INGEST is currently looking for bilingual research assistants over the summer to assist in the process of collecting time-activity diaries and asked that those interested contact knachman@jhu.edu.
