Downloads: 1
Original Research | Computer Science | Volume 15 Issue 8, August 2026 | Pages: 1793 - 1802 | India
Self-Sturdy Deep Neural Network for Improving Postpartum Depression Detection
Abstract: Postpartum depression (PPD) is a prevalent maternal health issue which adversely affects both mothers and their newborns. It negatively impacts mother-baby bonding, infant cognitive development, language, behaviors, sleep quality, and physical-mental health. In extreme cases, depressed mothers may resort to suicide or infanticide. PPD needs to be overlooked, requiring timely and early-stage treatment to prevent serious complications. For this reason, Osprey Parameter Optimized MLP (OPOMLP) is developed which utilizes the Multi-Layer Perception (MLP) for classification and Osprey optimization algorithm (OOA) for the feature selection and parameter optimization. But, model?s performance was lower due to its inefficiency to perform on larger datasets. In this paper, Self-Sturdy Deep PPD network (SSPPDnet) is developed to tackle the for-mentioned challenges for the effective PPD prediction. In this method, Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) are used to learn all feature types autonomously. The temporal aspects of data are captured using LSTM, which derives time-lapse attributes individually for all data format. CNN converts the extracted time-series features into the feature vector to capture the relationship between divergent. The retrieved features are fed into Self-Sturdy Deep Neural Network (SSDNN). SSDNN is developed for faster convergence with fixed global learning, resulting in lower initial learning rate training. Fully Connected (FC) layers are adopted to compute the postpartum accomplishment levels by combining the women's demographic information, associative characteristic and time-lapse attributes. The softmax layer is employed for PPD detection and classification. Finally, test results show that the SSPPDnet model obtains an accuracy of 95.03% and 97.63% on datasets respectively outperforming other models.
Keywords: Postpartum Depression, Stochastic Gradient Descent, Per-Layer Stabilizer, Deep Neural Network, Long Short-Term Memory
How to Cite?: D. Suganthi, A. Geetha, "Self-Sturdy Deep Neural Network for Improving Postpartum Depression Detection", Volume 15 Issue 8, August 2026, International Journal of Science and Research (IJSR), Pages: 1793-1802, https://www.ijsr.net/getabstract.php?paperid=SR26826165622, DOI: https://dx.doi.org/10.21275/SR26826165622