- Original question: How can specific language impairment (SLI) be detected and prevented before young children begin showing signs of the disorder through their speech?
- Summary of feedback: Most students approved of our idea as a moonshot pitch, and they were mainly interested in seeing what solutions we would use after early detection and whether they would be novel.
- Revised question: We are still interested in finding how SLI can be detected and prevented. Given the feedback we received, we are revising our question to ask, “How can SLI be detected and prevented using technology before young children begin showing signs of the disorder through their speech? Additionally, what is the best intervention strategy after early detection occurs?”
- Our 8 solutions:
1) Test parents’ genes before the child is born to determine if the child will have SLI/DLD.
This choice failed as it was too vague and we were unsure that it would be plausible to identify whether a child will have SLI based solely on their parents’ genes.
2) Utilize ultrasounds to test for possible physical indicators.
An ultrasound can detect plenty of useful information, but there are no known physical markers of SLI, so an ultrasound would likely not be helpful in this case.
3) Implement algorithms to identify at risk children.
The idea of creating an algorithm was the foundation for our ultimate solution. We decided to take this concept and create a more specific solution around it.
4) Use NIRS to inspect infants’ brain activity in response to indicative stimuli.
The use of brain imaging in this field is already commonly practiced and therefore does not fit the “radical” nature of a moonshot solution.
5) Develop a virtual reality program that completely immerses a child in an effort to detect SLI early on.
Access to virtual reality is limited due to its cost, and this approach can be carried out more efficiently with an app or device.
6) Making series of instructional videos that are designed to instruct children on language skills.
This is another method that led us to develop one of our rapid evaluation solutions (see #8). The problem with this particular idea is that we ultimately wanted to go beyond making videos and create an app that allows infants to respond to stimuli in real time.
7) Create an app that records babies’ babbling and can detect if the infant is on track to have normal speech and comprehension abilities or whether the infant will develop SLI.
8) Use this same app to allow infants for which SLI has been detected to get extra practice responding to different speech-related stimuli.
5. Our “Rapid Evaluation” solutions:
1) Create an app that records babies’ babbling and can detect if the infant is on track to have normal speech and comprehension abilities or whether the infant will develop SLI.
We would first need to gather audio of babies babbling and use machine learning to identify aspects of speech that differs between those with and without SLI. This would take years to complete, but it would enable us to detect SLI at a younger age than we can now. After we have gathered enough information, families will be able to have their infants babble while recording on the app, and the app will assess whether the baby has SLI.
2) Use this same app to allow infants for which SLI has been detected to get extra practice responding to different speech-related stimuli.
Simply determining whether an infant has SLI is not a strong enough solution to this problem. Once early detection occurs, we want our app to enable infants with SLI to practice receiving and producing speech. The app will simulate a "dialogue" with the infant and will track the baby's progress as they transition from babbling to speaking words.