Early Detection of Postpartum Depression Using Explainable Deep Learning Models: A Comparative Study of DNN, GATE, and SAINT Architectures
DOI:
https://doi.org/10.70112/ajcst-2026.15.1.4429Keywords:
Postpartum Depression, Deep Learning, Explainable AI (XAI), SHAP Analysis, Clinical Decision Support SystemsAbstract
Postpartum depression (PPD) affects up to 52.3% of women in Nigeria and sub-Saharan Africa, yet remains critically underdiagnosed due to cultural stigma, resource constraints, and the limitations of traditional screening tools such as the Edinburgh Postnatal Depression Scale. This study developed and comparatively analyzed three deep learning architectures—Deep Neural Network (DNN), Gated Attention Network (GATE), and Self-Attention Network for Tabular Data (SAINT)—to provide an explainable and deployable screening solution. Using a dataset of 1,503 postpartum records comprising sociodemographic and emotional variables, the models were trained using the Adam optimizer with dropout regularization. The DNN emerged as the superior architecture, achieving high performance across all evaluation metrics: 96% accuracy, 96% F1-score, and an ROC–AUC of 0.980, outperforming both GATE (93%) and SAINT (89%). To address the “black-box” barrier in clinical AI adoption, SHAP (SHapley Additive exPlanations) analysis was integrated, revealing that age (51.3% predictive contribution) and emotional indicators (48.7%) were the primary risk determinants. This level of transparency enables clinicians to interpret and justify individual risk forecasts. The final DNN model was implemented as a Flask-based web application, providing a real-time, scalable screening tool suitable for low-resource environments. The findings demonstrate that explainable deep learning can effectively address the PPD screening gap in sub-Saharan Africa. Future work should prioritize external validation with local Nigerian datasets and the integration of multi-class severity grading to further refine clinical decision support.
References
[1] World Health Organization, Mental Health Action Plan 2013–2020. Geneva, Switzerland: WHO Press, 2013.
[2] J. Suwalska et al., “Perinatal mental health during COVID-19 pandemic: An integrative review and implications for clinical practice,” J. Clin. Med., vol. 10, no. 11, p. 2406, 2021, doi: 10.3390/jcm10112406.
[3] E. T. Adeyemo, J. A. Balogun, and O. A. Olawale, “Postpartum depression as a cause of maternal mortality: A systematic review,” Int. J. Womens Health, vol. 11, pp. 715–724, 2019, https://doi.org/10.18203/2320-1770.ijrcog20253545.
[4] M. Bello, “Postpartum depression in African women: Causes, symptoms, and treatment approaches,” Dateline Health Africa, 2025. [Online]. Available: https://www.datelinehealthafrica.org/postpartum-depression-in-african-women-causes-symptoms-and-treatment.
[5] M. I. Ojeahere, I. T. Adeleke, and M. O. Oche, “Mental health literacy on postpartum depression among Nigerian women: A community-based survey,” BMC Psychiatry, 2025. [Online]. Available: https://pmc.ncbi.nlm.nih.gov/articles/PMC12405231/.
[6] R. Buhagiar, K. Bettenzana, and K. A. Grant, “Validation of the Edinburgh Postnatal Depression Scale and its 3-item anxiety subscale, and the Generalised Anxiety Disorder-7 item for screening of postpartum depression and anxiety in women in Malta,” Midwifery, vol. 141, p. 104256, 2025, doi: 10.1016/j.midw.2024.104256.
[7] M. G. Hoberg, “Improving understanding and detection of postpartum anxiety,” Ph.D. dissertation, Univ. Pittsburgh, Pittsburgh, PA, USA, 2022.
[8] Z. Ahmed, K. Mohamed, S. Zeeshan, and X. Dong, “Artificial intelligence with multi-functional machine learning platform development for better healthcare and precision medicine,” Database, vol. 2020, p. baaa010, 2020, doi: 10.1093/database/baaa010.
[9] M. Z. Alom et al., “A state-of-the-art survey on deep learning theory and architectures,” Electronics, vol. 8, no. 3, p. 292, 2019, doi: 10.3390/electronics8030292.
[10] S. C. Nouis, V. Uren, and S. Jariwala, “Evaluating accountability, transparency, and bias in AI-assisted healthcare decision-making: A qualitative study of healthcare professionals' perspectives in the UK,” BMC Med. Ethics, vol. 26, no. 1, p. 89, 2025, doi: 10.1186/s12910-025-01089-3.
[11] A. Das and P. Rad, “Opportunities and challenges in explainable artificial intelligence (XAI): A survey,” arXiv, preprint arXiv:2006.11371, 2020. [Online]. Available: https://arxiv.org/abs/2006.11371.
[12] F. T. Zohora, “Machine learning for predicting postpartum depression and suicidal tendencies in Bangladesh,” unpublished manuscript, 2025.
[13] G. R. Han et al., “Machine learning in point-of-care testing: Innovations, challenges, and opportunities,” Nat. Commun., vol. 16, no. 1, p. 3165, 2025, doi: 10.1038/s41467-025-58176-9.
[14] X. Huang, L. Zhang, C. Zhang, J. Li, and C. Li, “Postpartum depression risk prediction using explainable machine learning algorithms,” Front. Med., vol. 12, p. 1565374, 2025, doi: 10.3389/fmed.2025.1565374.
[15] Y. Wang, X. Li, R. Zhang, and L. Wang, “Interpretable machine learning model for predicting postpartum depression,” JMIR Med. Inform., vol. 13, p. e58649, 2025, doi: 10.2196/58649.
[16] K. Saqib, N. A. Abdullah, J. Siddiqui, and K. Saleem, “Machine learning methods for predicting postpartum depression: A scoping review,” JMIR Ment. Health, vol. 8, no. 11, p. e29838, 2021, doi: 10.2196/29838.
[17] S. Tortajada and J. M. Garcia-Gomez, “Machine learning predictive models for postpartum depression,” in Machine Learning for Healthcare Informatics, N. Peek, R. Morales Martin, and M. Peleg, Eds. Cham, Switzerland: Springer, 2019, pp. 335–352.
[18] Y. Lin, H. Wang, X. Chen, and Q. Liu, “A method for predicting postpartum depression via an ensemble neural network model,” Front. Public Health, vol. 13, p. 1571522, 2025, doi: 10.3389/fpubh.2025.1571522.
[19] A. Zafar, M. A. Qureshi, S. Khalid, and S. Bashir, “Prediction of perinatal depression among women in Pakistan using hybrid RNN–LSTM model,” PeerJ Comput. Sci., vol. 11, p. cs2673, 2025, doi: 10.7717/peerj-cs.2673.
[20] D. Shin, K. J. Lee, T. Adeluwa, and J. Hur, “Machine learning-based predictive modeling of postpartum depression,” J. Clin. Med., vol. 9, no. 9, p. 2899, 2020, doi: 10.3390/jcm9092899.
[21] R. S. Baidwan et al., “Multilayered ethical agenda for machine learning in medical diagnostics: Autonomy, privacy, and non-maleficence in the era of complex distributed systems,” IEEE Access, vol. 12, pp. 1–15, 2024, DOI:10.70112/ajcst-2024.13.1.4254.
[22] E. E. Onuiri et al., “Multidimensional data integration in predictive healthcare models: A meta-analysis of environmental and genetic interaction effects,” J. Biomed. Inform., vol. 158, p. 104710, 2024, doi: 10.1016/j.jbi.2024.104710.
[23] E. E. Onuiri et al., “Predictive analytics frameworks for cancer recurrence: Benchmarking sensitivity, specificity, and prognostic precision in machine learning models,” Comput. Biol. Med., vol. 178, p. 108720, 2024, doi: 10.1016/j.compbiomed.2024.108720.
[24] A. S. et al., “Optimized classification using recurrent neural networks: Mitigating overfitting in imbalanced healthcare datasets through hybrid gated and bidirectional architectures,” Expert Syst. Appl., vol. 263, p. 125812, 2025, doi: 10.1016/j.eswa.2025.125812.
[25] J. Pujari et al., “High-accuracy binary classification via CNN-LNet with optimized feature mapping using ReLU/Softmax activation for decision boundary refinement,” Remote Sens., vol. 17, no. 3, p. 412, 2025, doi: 10.3390/rs17030412.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Centre for Research and Innovation
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
