A Machine-Learning-Algorithm-Based Prediction Model for Psychotic Symptoms in Patients with Depressive Disorder.

Area of Science:

• Psychiatry
• Computational Psychiatry
• Machine Learning in Medicine

Background:

• Psychotic symptoms rarely co-occur with depression, and their classification as a distinct subtype remains debated.
• Understanding predictors of psychotic depression is crucial for accurate diagnosis and treatment.

Purpose of the Study:

• To develop and validate a machine learning model for predicting concurrent psychotic symptoms in patients with depressive disorders.
• To identify key clinical variables associated with psychotic depression.

Main Methods:

• Utilized data from 1171 patients with depressive disorders from the Research on Asian Psychotropic Prescription Patterns for Antidepressants study.
• Developed a machine learning algorithm-based prediction model to identify patterns and trends.
• Evaluated model performance using area under the curve (AUC) and overall accuracy.

Main Results:

• The machine learning model achieved an AUC of 0.823 and an overall accuracy of 0.931.
• Severe depression, diminished appetite, and suicidal ideation were significant predictors of psychotic symptoms.
• Other important variables included subthreshold depression, outpatient status, age, and psychomotor agitation/retardation.

Conclusions:

• A machine learning-based model effectively predicts concurrent psychotic symptoms in major depression.
• Findings support the "severity psychosis" hypothesis, linking symptom severity to psychosis occurrence.
• The model offers a valuable tool for identifying patients at higher risk for psychotic depression.