By Gillian Hayes, University of California, Irvine, FitBaby Principal Investigator.The Project HealthDesign blog is supposed to offer a “behind the scenes” view of our research. Well, right now, in the FitBaby project we are working on trying to figure out whether—and how—we can sense anything and everything about babies pooping. Fifteen years ago when I first entered the field of computer science I never thought I’d be here.
Let’s back up a little bit. We have been interested in measuring the ambient gases in the home to try to make some connections between infant comfort, health, and wellbeing and any potential contaminants in the air. One of our team members, Sen Hirano, a graduate student at UCI has a passion for tinkering. Sen can often be found in the lab late at night soldering together something or another.
A few weeks ago Sen came in to my office and informed me that he was pretty sure he could sense “flatus” using the air sensors we had bought. I nearly fell out of my chair. Once the surprise wore off, I asked him to show me his sensor streams, visualize the various gases he was sensing on his computer, and get, well, a bit more “sciencey” about the whole thing. It seemed that our methane sensor was spiking a bit at the expected times, which made sense given our limited knowledge of the human excrement system. So, we began to talk about the idea that we could sense baby poops and baby gas in the crib using just our tiny gas sensors.
A few days later, on Sunday morning, I got an instant message from Sen. I think the conversation speaks best for itself:
Sen: The major components of the flatus, which are odorless, by percentage are:
Carbon dioxide: 10–30%
Sen: Maybe I chose the wrong gas
Gillian: Nitrogen! Really? Can we sense that?
Sen: I'm about to start looking
Sen: For example, in one study of the feces of nine adults, only five of the samples contained archaea capable of producing methane. Similar results are found in samples of gas obtained from within the rectum.
Sen: The thing is that it's still working?
Sen: But the lack of methane in some places might be troublesome
Sen: Though since these sensors are also sensitive to other gases it might be okay?
Gillian: Well that’s the real question I suppose. What else are the sensors picking up?
Gillian: See if you can find out what gases comprise baby poop when breastfeeding... i suspect it might be different from those in bottle fed babies
Sen: This is turning into a biology experiment
Sen: Found it
Gillian: That’s totally the right thing
Gillian: And funny
Gillian: All at once
Sen: "I can tell the difference between breast fed and formula fed babies"
Gillian: Do we need new sensors for this super power of yours?
Sen: ha ha
Since that Sunday, we have gotten really serious about this idea. We are starting to think that the potential to monitor this kind of detail about the baby digestive system could be incredibly interesting both clinically and in terms of basic science and research. Parents are often concerned about their infant’s constipation or gassiness, but when asked, they cannot always remember the last “poopy diaper” nor how many feedings were bottles as opposed to breast fed. The color of poop might be a warning sign of trouble to come if the infant is bottle fed but totally normal if breast fed. The very gas contents themselves might even tell us something clinically relevant. At this point, we just don’t know, and that is very exciting.
At the same time, we think there may be some interesting computer science problems in here. We are in the process of doing controlled experiments to test our sensors and algorithms. Sen and I are lucky enough to have Monica Tentori, a post-doc in my lab, and Don Patterson, a fellow professor down the hall, both around to advise on the machine learning aspects of this work. We keep ordering new sensors, soldering together interesting components, and yes, we are prepared to collect “poopy diapers” in sealed Tupperware to test out their gas production capabilities. It promises to be an exciting time in our lab… and hopefully not too smelly of one.