Akshar Mitra: a multimodal integrated framework for early dyslexia detection.

Area of Science:

• Neurobiology
• Computational Linguistics
• Educational Technology

Background:

• Developmental dyslexia affects 10-15% of children globally, often undiagnosed in resource-limited settings.
• Current screening methods rely on unimodal data and require large labeled datasets, limiting accessibility.
• Existing assessments are often inaccessible in resource-limited settings.

Purpose of the Study:

• To introduce Akshar Mitra, a novel Multimodal Integrated Framework (MMF) for accessible and early dyslexia screening.
• To integrate low-cost digital biomarkers from eye-tracking, speech, and handwriting analysis.
• To bridge the global dyslexia diagnostic gap through a scalable and explainable system.

Main Methods:

• Developed a Multimodal Integrated Framework (MMF) integrating three modules: webcam-based eye-tracking, automated speech assessment, and optical character recognition for handwriting.
• Extracted interpretable features (e.g., fixation regressions, word-error rate, character reversals) from each modality.
• Standardized features via a shared data schema and augmented with a behavioral questionnaire.

Main Results:

• The MMF successfully integrates multimodal data for a holistic dyslexia risk profile.
• The system provides objective, interpretable digital biomarkers for early screening.
• Incorporated user-friendly support tools, including a dyslexia-friendly reading interface.

Conclusions:

• Akshar Mitra offers a viable pathway to address the inaccessibility of conventional dyslexia assessments.
• The framework's scalable, language-agnostic design facilitates early detection and intervention.
• This transformative tool has the potential to improve educational outcomes for children with dyslexia worldwide.