Information from Noise: Measuring Dyslexia Risk Using Rasch-like Matrix Factorization with a Procedure for Equating...

Area of Science:

• Psychometrics
• Cognitive Science
• Data Science

Background:

• Dyslexia is a complex condition with multiple cognitive factors.
• Existing screening tools may lack comprehensive psychometric validation.
• A need exists for reliable, adaptable dyslexia screening methods.

Purpose of the Study:

• To evaluate the psychometric properties of a novel dyslexia screening protocol.
• To introduce and validate the Nous matrix factorization method for cognitive assessment.
• To develop an age-generalized Dyslexia Risk Scale (DRS).

Main Methods:

• Employed a novel matrix factorization technique (Nous) based on Alternating Least Squares.
• Constructed a common 6-dimensional space to extract multidimensional signals from test data.
• Utilized Rasch model objectivity requirements and information-theoretic principles.
• Calibrated the Dyslexia Risk Scale (DRS) with a sample of 828 individuals aged 7-82.

Main Results:

• The Dyslexia Risk Scale (DRS) demonstrated high reliability (0.95).
• Achieved excellent diagnostic accuracy with an Area Under the Curve (AUC) of 0.96.
• The DRS provides linear, equal-interval measures comparable across different item subsets.
• Standard errors for measures and estimates were efficiently calculated.

Conclusions:

• The Nous-based Dyslexia Risk Scale (DRS) is a psychometrically sound screening tool for dyslexia.
• The method allows for age-generalized assessment and longitudinal tracking of dyslexia risk.
• The study highlights the utility of advanced matrix factorization in cognitive assessment.