Using machine learning to develop a five-item short form of the children's depression inventory

Area of Science:

• Psychiatry
• Clinical Psychology
• Developmental Psychology

Background:

• Adolescent depression is a significant concern, often undetected and untreated.
• The 27-item Children's Depression Inventory (CDI) is widely used in China but is time-consuming.
• A need exists for a localized, simplified, and efficient depression screening tool for adolescents.

Purpose of the Study:

• To develop a shortened version of the Children's Depression Inventory (CDI) for efficient prediction of high depression risk in adolescents.
• To enhance the speed and accessibility of depression screening in the adolescent population.
• To create a localized and validated tool for use in China.

Main Methods:

• Backward elimination was used to create short-form scales (e.g., three-item, five-item).
• Five machine learning algorithms evaluated short-form scale performance using AUC.
• The optimal scale and decision threshold were determined, followed by reliability and validity assessments.

Main Results:

• A five-item CDI scale with a decision threshold of 4 was identified as optimal.
• The scale demonstrated good predictive performance (AUC=0.81, Accuracy=0.83).
• The five-item scale showed good measurement properties in a sample of 315 middle school students (Cronbach's alpha=0.72, validity=0.77).

Conclusions:

• The developed five-item CDI is the first shortened and revised version for China.
• This scale offers an efficient method for predicting high depression risk in adolescents.
• The tool is based on a large, local data sample, ensuring relevance and applicability.