Our Evaluative Artificial Speech Intelligence-Autism Screener project (EASI-AS) is an NSF/SBIR Phase II grant that expands on the work from our Phase I NSF/SBIR grant. EASI-AS will feature cutting-edge machine-learning technology that will reliably screen for autism by identifying non-linguistic acoustic features, extracted from recordings of children’s speech. The system will use familiar mobile devices to collect speech samples, which will employ a user-friendly interface that parents and professions can operate with little to no training.
What We Do
We are working to help autistic children and their families by creating a tool that anonymously screens for autism by using speech samples. Research has shown that early intensive intervention services can improve social and communication skills in autistic children. Other studies have shown that these gains can be maintained years later, after services have concluded. So increasing access to early intervention services is one of the key goals of our project.
But in order for our system to work, we will need your help! To ensure that the technology properly differentiates between autistic and non-autistic children, we will need lots of data from both populations. In the first stage of the project, we will be creating a website that will collect anonymous speech recordings from a variety of children. The recordings will help train the system to identify the unique acoustic traits that are often present in autistic children’s voices.
If you would like to contribute recordings to the project, please enter your recording(s) here. If you have any questions about EASI-AS and our Phase II SBIR work, please contact us at email@example.com
Who We Are
Lois Jean Brady has over 30 years of experience as a practicing Speech-Language Pathologist, she’s a Certified Autism Specialist (CAS), and she also holds certification in Assistive Technology and in computer-based interventions. Career accomplishments include: Winner of two Autism Speaks App Hack-a-thons, Benjamin Franklin Award for Apps for Autism and an Ursula Award for Autism TodayTV. In addition to Apps for Autism, she has co-authored Speech in Action and Speak, Move, Play and Learn with Children on the Autism Spectrum. Lois gives international presentations to both family members and fellow professionals at conventions and seminars on autism and technology. Currently she is researching and developing engaging, multi-sensory products to enhance communication, attention, cognition and quality of life for individuals with autism. She has co-developed VAST Autism, a series of three highly effective apps for increasing speech production. Lois is also registered in Animal Assisted Therapy. Buttercup, her potbelly pig, frequently accompanies Lois in therapy sessions and is a huge motivator for all. In 2015, Lois and Matthew Guggemos earned a grant through the National Science Foundation's Small Business Innovative Research program. Lois served as the principal investigator in 2018, 2019, and currently for SBIR/NSF and SBIR/NIH grants. As a team, they are working on expanding their platform using technology from Epic Unreal Engine and Snap Camera, Snapchat, and Spectacles.
Matthew is the chief technology officer for iTherapy, LLC’s EASI-AS project. In addition to the current Phase 2 EASI-AS project, Matthew has served as the CTO and/or co-investigator on two Phase one NSF/SBIR projects and one NIH/SBIR project — all of which developed educational technology specifically designed for autistic children and adults. Outside of government-funded projects, Matthew has collaborated with Microsoft, NewSchools Venture Fund, Epic Games, and Snap to develop multi-sensory digital learning products for neurodiverse individuals. Additionally, he received the 2013 Mensa Intellectual Benefits to Society Award for his design contributions to InnerVoice, which was co-developed with iTherapy CEO and speech-language pathologist Lois Jean Brady and MotionPortrait, Inc. In addition to being a speech-language pathologist, Matthew is a father, Brazilian jiu-jitsu student, and drummer. You can read about, watch, and listen to his work here: https://linktr.ee/drumlanguage
Garrett is a speech language pathologist, musician, and research coordinator based between California and Tokyo with a multidisciplinary background in music, language, and cognitive science. Graduating from Emerson College in 2014, he has given presentations in a wide variety of language topics at school districts along with the American Speech Language and Hearing Association Conference in 2018. He is also an internationally touring music professional and an avid surfer. Given his multidisciplinary background, he is passionate about how concepts and technologies from other fields can be re-purposed for the use of treatments for communication disorders. To this end, he was a co-creator of a beatboxing for speech therapy curriculum and part of ongoing research into how current machine learning technology can be applied to speech assessment and intervention.
Dr. Weiqing Gu, founder and chief scientist of Dasion (meaning data-to-decision where Geometric Unified Learning was born) and a McAlister Professor of Mathematics at Harvey Mudd College. Dr. Gu is deeply involved in research for predictive models especially using nonlinear data including voice data, cell phone data and UAV data and anomaly detection in Machine Learning. Her former research on the geometry of a manifold (e.g. sphere or Grassmann manifold) and computational geometry applies to fundamental prob-lems in dynamics and control theory. Dr. Gu went to college at age 15 and started teaching linear algebra, calculus and probability theory when she was just 18. When the Ameri-can Mathematical Society selected Dr. Gu and her students to present their NSF funded research proposal to the US Congress in 2007, she demonstrated how mathematics can help cure cancer. She was then appointed as a program manager at the NSF where she helped select research proposals and earned an outstanding evaluation. Since then she has been involved in Big Data Analytics for the US Defense Threat Reduction Agency. Her previous work on UAV data, radar data and hyperspectral image data at the Navy also earned her a recognition award for exceptional performance. She has published 36 papers in the ﬁelds of Machine Learning, AI, diﬀerential geometry, topology, geometric modeling and math-biology. She has created many courses to bridge the gap between industry and academia including nonlinear data analytics, mathematics of big data, and information geometry. When the global pandemic began and students had to head home, she quickly developed a special topic course “Covid-19: Data Analytics and Machine Learning” to keep students engaged in learning while they are at home. On top of her academic/research credentials, Dr. Gu is well-connected in the industry by working with over 50 companies through Harvey Mudd’s clinic program(Proof-point, Virgin orbit, etc.), teaching over a thousand students at both Harvey Mudd College and Claremont Graduate University, and doing industry collaborations during summers with over 30 companies (UnifyID, Intel, Laserﬁche, Unilever, etc.). She has been in touch with companies interested in working with Dasion such as State Farm, Seeq and Decisive Data. She has a broad view of math that allows her to understand how all the components of massively complicated systems can be beautifully integrated. She understands that many of the competing approaches in machine learning are actually the same but coming from diﬀer-ent perspectives. Her combination of industry connections, broad mathematical expertise,applied knowledge and dedication to advance the ML ﬁeld makes her the perfect person to execute on this proposal.