NODA Telehealth system improves access to an autism diagnostic assessment by guiding families to share video clips of their child at home, so diagnostic clinicians can directly observe and ?tag? video of any atypical behavior, and if warranted, render a diagnosis. This system is evidence-based and has been commercialized, with several published studies to discuss the benefits. We now propose to improve this service by developing a Deep (machine) Learning capability in a software product called ?NODA DL Classifier? to help clinicians more quickly identify and better quantify typical and atypical behaviors on videos they receive from families. If successful, this NODA DL feature within the NODA system will have a profound impact in the time to reach a firm diagnosis, and then the capability could be used subsequently to effectively monitor treatment progress of individuals diagnosed with autism. In this project, we will determine how much DL improves the diagnostic process. In Phase I, we will test our use previously generated datasets to qualify and quantify potential benefits. In Phase II, we will conduct a clinical study to document time-savings and other clinical benefits. Our proposed NODA DL innovation represents a large step change in identification and then the care for ASD individuals, not an incremental one. It will lead to a significant improvement in both health outcomes and in reduced time required by clinicians or psychologists for office visits and for analyzing video data. This reduced time can be translated into reduced costs. We anticipate that significant commercial benefits will result from the use of our innovative computer methodologies.
The proposed computerized Deep Learning (DL) function within our current NODA Telehealth System will have a profound impact in saving time to reach a firm diagnosis of individuals with ASD, plus provide other important benefits.
