The Early Psychosis Screener for Internet (EPSI)-SR: Predicting 12 month psychotic conversion using machine learning

Area of Science:

• Psychiatry
• Clinical Psychology
• Neuroscience

Background:

• Early identification and intervention are crucial for improving outcomes in individuals at risk for psychosis.
• Existing screening methods can be time-consuming and resource-intensive.

Purpose of the Study:

• To develop and validate a faster, more accurate self-report screener for early psychosis.
• To identify individuals at high risk of developing a psychotic disorder within 12 months.

Main Methods:

• Self-report Likert-scale items were administered to individuals screened with the Structured Interview for Psychosis-risk Syndromes (SIPS).
• Spectral Clustering Analysis reduced the item pool, followed by Support Vector Machine (SVM) classifier development.

Main Results:

• The Early Psychosis Symptom Inventory (EPSI) demonstrated a higher positive predictive value (PPV) than the SIPS in identifying individuals who would not convert to psychosis.
• Combined use of EPSI and SIPS increased the PPV to 86.6%.
• SVM classifiers accurately distinguished between high-risk and low-risk populations, with low misclassification rates.

Conclusions:

• The EPSI is a validated, time-efficient, and resource-sparing tool for identifying individuals at high risk for psychosis or already experiencing psychosis.
• The EPSI is the first validated assessment to predict 12-month psychotic conversion.
• An online screening system is under development to enhance accessibility.