Construction of depression risk prediction model using genetic markers and machine learning

Area of Science:

• Genetics
• Computational Biology
• Psychiatry

Background:

• Depression is a major global health issue with subjective diagnostic criteria.
• Genetic factors play a significant role in depression development.
• Machine learning offers potential for objective depression risk prediction.

Purpose of the Study:

• To develop a machine learning model for depression risk prediction using genetic data.
• To differentiate between individuals with depression and healthy controls based on genetic markers.
• To establish an objective tool for assessing depression risk.

Main Methods:

• Utilized genetic data from 791 depression patients and 413 healthy controls.
• Selected 1309 candidate genes based on Kyoto Encyclopedia of Genes and Genomes pathways.
• Employed Boruta and LASSO for feature selection, identifying 12 single nucleotide polymorphisms (SNPs).

Main Results:

• A final predictive model incorporating 12 SNPs was developed.
• The model achieved an Area Under the Curve (AUC) of 0.66 in the testing set.
• Demonstrated good discrimination and stability in predicting depression risk.

Conclusions:

• The developed genetic-based risk prediction model shows potential for objective depression assessment.
• This tool could significantly aid in the early detection of depression.
• Highlights the utility of machine learning in psychiatric genetics.