Eye Movement Characteristics for Predicting a Transition to Psychosis: Longitudinal Changes and Implications

Area of Science:

• Neuroscience
• Psychiatry
• Computational Psychiatry

Background:

• Limited research exists on the predictive value of eye movement (EM) features for psychosis conversion in clinical high-risk (CHR) individuals.
• Understanding longitudinal EM trajectories is crucial for early psychosis detection.

Purpose of the Study:

• To investigate the predictive accuracy of machine learning models using EM indices for CHR conversion.
• To examine longitudinal changes in EM indices in CHR individuals over one year.

Main Methods:

• Eye movement assessments (fixation stability, free-viewing, smooth pursuit) were conducted on 140 CHR and 98 healthy controls.
• Cox regression and random forest models were used for prediction; linear mixed-effects models (LMMs) analyzed longitudinal changes.

Main Results:

• CHR individuals who converted to psychosis showed distinct EM patterns (prolonged fixation, decreased saccade amplitude, reduced smooth pursuit velocity gain).
• A random forest model achieved 0.776 accuracy in classifying converters versus non-converters based on baseline EM measures.
• No significant longitudinal changes in these EM indices were observed in converters after one year.

Conclusions:

• Aberrant eye movements may precede psychosis onset and remain stable over a year.
• Eye-tracking technology combined with predictive modeling shows promise for identifying CHR individuals at high risk of developing psychosis.