An Innovation-Driven Approach to Specific Language Impairment Diagnosis

Area of Science:

• Speech-language pathology
• Human-computer interaction
• Diagnostic technology

Background:

• Current specific language impairment (SLI) diagnosis methods are manual, costly, and inconvenient.
• Computer-aided diagnosis offers a potential solution to these limitations.
• This study evaluates the feasibility of computer systems for SLI diagnosis.

Purpose of the Study:

• To assess the accuracy of Webgazer.js for gaze tracking under varying lighting conditions.
• To compare the diagnostic accuracy of automated vs. manual time estimation in a prototype SLI diagnosis task.
• To validate a newly developed SLI diagnosis classifier against existing software.

Main Methods:

• Webgazer.js gaze tracking accuracy was tested under global, low, and face-focused lighting.
• A prototype diagnosis task automation script's timing precision was compared to manual estimations.
• An SLI diagnosis binary classifier was created and validated against Singlims_ES.exe.

Main Results:

• Webgazer.js accuracy ranged from 52.7% (face-focused lighting) to 88.7% (global lighting).
• Automated script timing deviation was below 0.04%, significantly better than manual estimation's 3.37%.
• The new classifier and Singlims_ES.exe showed similar probability values to three decimal places.

Conclusions:

• Computer tools show potential for assisting speech-language pathologists in SLI diagnosis.
• Further research is needed to optimize computer-aided diagnostic tools for clinical use.
• Findings provide a basis for developing more efficient and accessible SLI diagnostic technologies.