It’s an exciting time for Matthew Goodwin, director of clinical research at the Massachusetts Institute of Technology’s Media Lab and associate director for research for the Providence-based Groden Center, who will be working on a recently awarded, $10 million, five-year National Science Foundation Expeditions grant for autism technology.
Six different institutions will collaborate on developing technology designed to quantify an autistic child’s behavior. Although still in the “very, very” theoretical stage, the device would tell doctors, parents and caregivers how an autistic child is feeling, even though the child cannot. Though the Expeditions grant is not linked to commercial development, the technology could eventually be used to develop a device that could be sold to institutions or even the public.
Beyond Cash Donations: How New Forms of Giving Are Transforming Not-for-Profit Accounting
Evolving Funding Landscape for Not-for-Profits Not-for-profit organizations are being asked to do more with less,…
Learn More
Goodwin is one of 13 principal investigators to work on the project and eager to get started.
PBN: First of all, congratulations for working on this exciting project. What kind of technology are you looking to develop and why is this grant so special?
GOODWIN: The Expedition awards are about tackling hard problems that are not easily solved by one discipline or one group of researchers. They are interested in technological developments that either create new paradigms or new fields of inquiry.
The over-arching theme for this Expedition [award] is a new field that we would like to consider that we’re calling computational behavior science. The best analogy that I can give you is medical imaging technology. What CAT scans and MRIs did to look inside the body without invasive procedures … is what we would like to do for social and communicative behavior science.
We are going to take three different fields in the computer sciences – computer vision, audio and speech analysis, wireless physiological recording – and use it to create a multimodal system.”
PBN: Do you think that $10 million is enough to see through such an ambitious project?
GOODWIN: I don’t know how much we can get done in five years but I think there is a tremendous amount of potential. It has me more excited than any other grant because it can help the people currently affected by autism; a lot of funding goes into neuroscience and genetic research – which is good – but none of it does anything for the people that have the disease here and now. It’s absolutely revolutionary.
PBN: What resources do care givers have right now for diagnosing autistic children and how do you hope to change that?
GOODWIN: We want to quantify earlier diagnosis and quantify behavior change in therapeutic interactions. Right now, the state-of-the-art for diagnosis and behavioral intervention is usually an experienced clinician or educator, who has had years of experience of seeing kids who are developing typically and kids who are behaviorally or mentally disordered. They are bringing these lifetime experiences to the diagnosis and a lot of the evaluations are based on intuition. We’re trying to see if we can quantify what it is that these educators are seeing that is leading them to these diagnoses.
We want to understand what these successful people are doing, take that data and see if we can identify those patterns in a multimodal system and identify the signals to augment the ability of non-experts to be precise [when working with autistic children].
PBN: What does that mean for the average parent?
GOODWIN: If you’ve got a playroom at home, and a parent is seated at a desk with the kid, you’d have a camera clipped on the wall with a microphone. The kid would be wearing a bracelet that is monitoring their physiological arousal; you’d use the video channel to quantify how much the kid is looking at the parent as to looking away. Another example would be, once the parent delivers a stimulus, how long it takes the kid to react? At the same time, if the task is that the kid has to verbally respond, how correct is that response?
If the kid is in an aroused state, the system could say, “Let’s take a deep breath,” or “Let’s do something to bring arousal down.” It could also tell the parent to stimulate the kid if he’s in a low arousal state.
PBN: It sounds like the device would need to be very personalized. Would the readings be based off of an individual’s behavior or use broad-spectrum behavior patterns?
GOOWIN: We would have to do both, the way we’ll start with a person-dependent training. As we see more and more kids, on and off the spectrum, we may be able to build up algorithms that work for general classes of people. But you would have to fine-tune the device for a given person. If we start to see hundreds of kids, we’d look for similarities in profiles between groups of kids.
