Tackling the Opioid Crisis, One Sewage Sample at a Time
A Sidewalk Talk Q&A with Newsha Ghaeli of the wastewater epidemiology startup Biobot Analytics.
There is probably a good joke — or at least a sci-fi flick — that ends down a manhole in Boston with a robot named Luigi. For Newsha Ghaeli, that was just another day at work.
A Toronto native, Ghaeli studied architecture at University of Waterloo then McGill before connecting with Carlo Ratti at the MIT Senseable City Lab, where she explored ways to help coastal communities improve their climate resilience using environmental sensors. Her roundabout path eventually took her to the lab’s Underworlds project, which developed Luigi to help study local public health problems by sampling city sewers.
“I sort of stumbled upon technology as a possible solution to urban issues,” says Ghaeli. “It wasn’t like I went into the whole smart city space because I was enamored by tech.”
Ten months ago, Ghaeli and MIT collaborator Mariana Matus broke off to launch Biobot Analytics, a wastewater epidemiology startup that analyzes sewage data to help cities understand — and respond to — big public health challenges. Whereas conventional wastewater research focuses on treatment plants far removed from population centers, Biobot scoops poop data right from neighborhood sewage streams to give local leaders more precise health insights. The approach represents the latest in a long legacy of disease-mapping innovations that helped give rise to modern cities, going back to John Snow identifying cholera in 1854 London.
“Density affords cities with the best of the best, which is increased innovation and all of that, but it also presents us with very challenging situations like increased crime and disease,” says Ghaeli. “A lot of those challenges are also opportunities to think through better ways to approach urban design or urban management.”
Ghaeli spoke to Sidewalk Talk about Biobot’s recent efforts to track the opioid crisis — and how sewer data can help cities get ahead of major health trends.
Biobot spun out of MIT, where it started as a project called Underworlds. What was the goal back then?
Underworlds started as a conversation between Carlo Ratti and Eric Alm, who is a microbiologist focused on the human microbiome. And the question was, in the same way that we can tell a lot about a person by sampling their gut, can we understand more about our cities by sampling and sensing our sewer system?
So Eric took that back up to his lab, and there was one student that got really excited about it, and that was Mariana Matus, my co-founder. She was a first year PhD student then. So she and Eric and Carlo decided: Why don’t we try and measure the flu season through sewage in Boston? Essentially the experiment failed, but that initial work prompted her to write a large research proposal recognizing that this isn’t as easy as just having scientists collect buckets of sewage and test for the virus. We actually need to take a much more integrated approach. Involve other disciplines: architects, planners, engineers. Think about where we’re sampling in the city to understand what we’re looking at, and what it says.
Why haven’t public officials used sewage analysis for this purpose before?
So wastewater epidemiology has actually been happening for over a decade, primarily in Europe and in Australia. It’s traditionally been scientists who are collecting these samples from treatment plants and then analyzing them to understand the concentrations of various illicit drugs.
Now, when you look at a wastewater treatment plant, it’s so far removed from a community. It’s very hard to actually say something or get some actionable data from a group of people when you’re looking at millions of people. Our method collects samples from manholes within a community. In taking that approach, there was actually an unintended finding: we found that we were able to measure the metabolized version of chemicals or drugs that we’re consuming, and that was the first time that was done in wastewater. What does that mean? When you take an Advil, as you digest it is modified to the metabolized form, and you excrete that metabolized version.
And we’re able to measure that. When you’re measuring at a treatment plant, the samples have been traveling for far too many hours and these metabolites break down. But because we’re sampling in-stream much closer to the source, we’re able to see them.
It just gives you a better chance at having a local answer to the local question?
Exactly. We’re 100 percent certain that what we’re measuring has been consumed and excreted by a population rather than being pharmaceutical waste. It’s much more directive in what it’s telling us. We can leverage that data to evaluate policy. When former Mayor Bloomberg, for example, wanted to put a ban on the size of soft drinks that could be sold in Manhattan, could we theoretically, in a very short amount of time — four months or six months — understand whether less sugar is being consumed by a population? Those are the kinds of questions that we can begin to ask.
What level of precision can you get? Down to the building level? Not the individual, I imagine?
We don’t sample a catchment area that has less than 5,000 people. There is absolutely no way to trace this information back to an individual.
So we sample areas that represent about tens of thousands of people, but we do map out the wastewater network ahead of time. So cities have the GIS maps and we’ll give the limitations that we want for our catchment areas, and then get the manhole location and we know what the corresponding kind of tributary area is based on those GIS maps.
Then we actually have built and designed the sampling device, or the hardware component, in-house. And that’s a device that we deploy under the manhole cover. Then those devices are concentrating gallons and gallons of wastewater, and then the membranes come back to our lab every couple of weeks for analysis.
You’re currently focused on measuring the opioid epidemic. Can you explain why you picked that issue and what you hope to find?
At the very beginning, we wanted to give cities a choice of what application was most beneficial to them. So we listed all the different things we could measure — infectious diseases, antibiotic resistance, illicit drugs — and we asked cities what application they’d like to implement. And we found that menu of options was a little too much.
So we ended up changing that question and sort of turning the question on its head. And then tried to speak with as many city and public health officials as we could, asking them, “What is your biggest health concern? What’s your biggest priority, top of agenda?” And we essentially got the opioid epidemic across the board. That was the thing that cities cared most about, so that’s what we started working on.
How is population data on opioid use collected today? Why doesn’t that give us a complete picture?
In learning about the crisis and then also the data involved, we found that there was a need for better data. So the majority of the data that’s being collected and analyzed and used today is limited to overdoses — a very small number of extreme users. It’s not indicative of everybody. For example last year, the New York Times printed a graph showing exponential growth of overdose deaths. That was based on a study that had just been published by the CDC showing overdose deaths from 18 months prior only.
So we’re designing policies based on this very old reactive data. And so that’s where we felt the data that we’re generating can really make a difference. It’s near real-time; our data is available within two weeks of the samples coming out of the ground, but also it’s representative of the entire city. So we get consumption data representative of everyone. And it’s consumption data. It’s not overdose data.
So we find that this information is really useful for cities, first and foremost, to really understand what’s happening. Public health departments don’t have a lot of visibility into their residential neighborhoods. So understanding what is the scope of the epidemic in my city, and then leveraging this information to re-allocate and redirect resources — to be much more targeted in the types of programming, whether that’s messaging and educational campaigns.
And finally, because we are sampling every two weeks, so we’re generating these frequent data points, we’re creating trend data that can be leveraged to actually evaluate and monitor the efficacy of programming, which is essentially evaluating the efficacy of the millions of dollars that are being spent.
In terms of your own capabilities as a company, is there something in particular you hope to learn with this first series of deployments?
We really believe in the value and the potential of wastewater epidemiology. For us the biggest thing has also been trying to really bring this technology to cities and have them realize, whether it’s the health department or the mayor’s offices, that we can actually start to generate a lot of useful information on community health. And that wastewater is actually an extremely valuable source of data and information.
That really resonates with cities. Obviously they really care about the opioid data, and that’s what is most important. But they really understand that the potential here is so much greater, and that we can generate all sorts of other information. We can look at nutrition disparities within a city. How does that feed into policy? We can look at building out an early warning system for the next infectious disease outbreak, or an early warning system for the emergence of a new antibiotic resistant strain. Things like that.
How does your team approach sharing access to the data you collect?
We had a moment a few months ago where we had a decision to make: How do we want to position ourselves as an urban-tech company? We can align our values very directly with those of a city, or we can build out a bit more like a tech company that is really protective of its data. We really wanted to align ourselves with cities.
At the end of the day, we’re building this out because we want to promote healthier communities. The best interest of the city is the best interest of us as a company. As a result, the cities own the data that we’re generating. We provide the data to the city. They’re free to share that however they see fit.
Is your hope to spark action, or do you just want to give cities the information to make their own decision?
It’s really about empowering the city, empowering health departments, decision makers, policymakers. Given that the data that we’re generating has never been generated before, we do find ourselves in this push and pull. There is this tension between how much do we want to advise versus be on just the data analytic side.
At the end of the day, we’re researchers. We’re scientists. What we’re best at is the data analytics and the data integrity. We recognize that cities and health departments are much better equipped at making decisions on what to do. We want to make those decisions easier for them to make because they have the best data possible.
Having gone from architecture into data, do you see a direct connection between the built form and urban data analysis?
They’re going to be inseparable. In a way, people have already been taken over by digital technologies, in that we’re all carrying our phones around all the time, and it’s impossible for any of us to be without these technologies at our fingertips so to speak. Very soon, our cities are going to follow suit. It’s really interesting that it happened with us before it did with the built environment.
Last question: Hand soap or Purell?
In the field, we had everything. We had Purell. We had bleach wipes. Both Mariana and I have had sewage-induced illnesses.
Definitely hand soap.
Follow what Sidewalk Labs is thinking, doing, and reading with our weekly newsletter, or on Twitter and Facebook.
