AI for Detecting and Predicting Postpartum Depression: Scoping Review

Area of Science:

• Medical Informatics
• Computational Psychiatry
• Machine Learning in Healthcare

Background:

• Postpartum depression (PPD) affects a significant portion of mothers globally, necessitating improved early detection methods.
• Existing screening tools for PPD often lack scalability and predictive accuracy.
• Artificial intelligence (AI), encompassing machine learning, deep learning, and natural language processing, offers enhanced capabilities for accurately identifying mothers at risk of PPD.

Purpose of the Study:

• To systematically map and analyze the existing literature on artificial intelligence (AI)-based methods for detecting and predicting postpartum depression (PPD).

Main Methods:

• A comprehensive scoping review was conducted following PRISMA-ScR guidelines.
• Empirical studies utilizing AI for PPD detection or prediction were systematically searched across eight major databases.
• Data extraction focused on study characteristics, AI models, data sources, preprocessing, validation, and performance metrics, followed by narrative synthesis.

Main Results:

• 65 studies met inclusion criteria, with the majority from the United States and published in 2024.
• AI models were predominantly used for PPD prediction (80%) over detection (22%), with sociodemographic, psychological, and obstetric data being primary inputs.
• Machine learning, particularly ensemble-based boosting models, showed superior performance, though studies often lacked external validation and standardized reporting.

Conclusions:

• The review highlights the dominance of classical machine learning in AI-driven PPD research.
• There is limited adoption of deep learning and advanced preprocessing techniques, alongside inconsistent validation strategies.
• Future research should address limitations such as small sample sizes, geographic bias, and the need for standardized, multimodal data approaches for more reliable PPD prediction.