Invention:
This technology is an algorithm that takes natural language text as an input, and uses rule-based and machine learning techniques to identify behaviors that are consistent with the diagnostic criteria for autism spectrum disorders (ASD). It suggests a diagnosis and highlights and summarizes the relevant information to aid a clinician’s decision making process.
Background:
The prevalence of ASD is on the rise. A 2018 study by the CDC estimated that about 1 in 44 children in the US have ASD. Early detection is the best way to minimize developmental challenges among children with ASD. But many do not get diagnosed in a timely manner due to a lack of access to qualified diagnosticians as well as challenges with early diagnosis. Additionally, members of minority groups tend to be correctly diagnosed later and less often.
This technology serves to improve the state of ASD diagnosis, by speeding up the assessment process and empowering clinicians to make more informed decisions. Unlike many other diagnostic aids, it looks for behavior consistent with the diagnostic criteria in the Diagnostic Statistical Manual of Mental Disorders (DSM), the accepted authority on criteria to diagnose ASD conditions.
Applications:
- Diagnostics
- Automated screening
- Public health tracking
- Clinical care
Advantages:
- Appropriate for use by clinicians with varying degrees of expertise
- Speeds diagnostic process
- Highlights and summarizes relevant information
