Eye Movement Patterns as Robust Biomarkers for Schizophrenia Identification Using a Novel Data Transformation...

Area of Science:

• Neuroscience
• Psychiatry
• Biomarkers

Background:

• Eye movement abnormalities are known in schizophrenia (SZ) but lack objective diagnostic biomarkers.
• Current diagnostic methods for SZ can be subjective and require further objective tools.

Purpose of the Study:

• To develop and validate a novel identification framework for schizophrenia using eye movement patterns.
• To assess the efficacy of the Sparsity-Scoring Kernel Entropy Component Analysis (SSKECA) algorithm in capturing SZ-specific eye movement characteristics.

Main Methods:

• Utilized a dataset of 40 patients with SZ and 50 healthy controls (HC) performing a free-viewing task with 100 semantic images.
• Developed a novel framework integrating the SSKECA algorithm with multidimensional eye movement features.
• Employed machine learning models (SSKECA-AdaBoost, SSKECA-XGBoost) for classification and feature analysis.

Main Results:

• The SSKECA-AdaBoost model achieved high accuracy (0.933) and AUC (0.960) for SZ identification.
• A reduced set of 25 images with the SSKECA-XGBoost model still yielded high accuracy (0.922).
• Feature ablation and misclassification analyses confirmed the distinctiveness of SZ eye movement patterns and highlighted deficits in misclassified patients.

Conclusions:

• The proposed framework effectively translates complex eye movement patterns into robust indicators for subject-level identification in SZ.
• This approach offers a practical and efficient tool to support objective assessment and potential diagnosis of schizophrenia.
• Further research can explore the clinical utility of this eye-tracking-based biomarker.