A method for predicting postpartum depression via an ensemble neural network model

Frontiers in Public Health, Год журнала: 2025, Номер 13

Опубликована: Апрель 14, 2025

Introduction Postpartum depression (PPD) has numerous adverse impacts on the families of new mothers and society at large. Early identification intervention are great significance. Although there many existing machine learning classifiers for PPD prediction, requirements high accuracy interpretability models present challenges. Methods This paper designs an ensemble neural network model predicting PPD, which combines a Fully Connected Neural Network (FCNN) with Dropout mechanism (DNN). The weights FCNN DNN in proposed determined by their accuracies training set respective values. structure is simple straightforward. connection pattern among neurons makes it easy to understand relationship between features target feature, endowing interpretability. Moreover, does not directly rely prevent overfitting. Its more stable than that DNN, weakens negative impact model. At same time, reduces overfitting risk enhances its generalization ability, enabling better adapt different clinical data. Results achieved following performance metrics dataset: 0.933, precision 0.958, recall 0.939, F1-score 0.948, Matthews Correlation Coefficient (MCC) 0.855, specificity 0.923, Negative Predictive Value (NPV) 0.889, False Positive Rate (FPR) 0.077, (FNR) 0.061. Compared 10 classic classifiers, under dataset split ratios, outperforms terms indicators such as accuracy, precision, recall, F1-score, also stability. Discussion research results show effectively improves prediction can provide guiding suggestions relevant medical staff postpartum women decision-making. In future, plans include collecting disease datasets, using predict these diseases, constructing online platform embed model, will help real-time prediction.

Язык: Английский

Sex-Specific Inflammatory Profiles Affect Neuropsychiatric Issues in COVID-19 Survivors

Biomolecules, Год журнала: 2025, Номер 15(4), С. 600 - 600

Опубликована: Апрель 18, 2025

Post-COVID syndrome has unveiled intricate connections between inflammation, depressive psychopathology, and cognitive impairment. This study investigates these relationships in 101 COVID-19 survivors, focusing on sex-specific variations. Utilizing path modelling techniques, we analyzed the interplay of a one-month 48-biomarker inflammatory panel, with three-months symptoms performance. The findings indicate that impairment is influenced by both inflammation depression overall cohort. However, prominent differences emerged. In females, lingering imbalance pro- anti-inflammatory responses-likely reflecting long-lasting immune alterations triggered COVID-19-significantly affects functioning shows marginal, though not statistically significant, association symptoms. suggests mixed profile may contribute to outcomes. Conversely, males, was inversely associated severity, protective effects from regulatory mediators (IL-2, IL-4, IL-6, IL-15, LIF, TNF-α, β-NGF) against depression. appeared be driven mainly symptoms, minimal influence markers. These results highlight distinct pathways responses post-COVID-19, potentially shaped endocrine mechanisms. suggest persistent foster long-term neuropsychiatric sequelae, possibly through its brain, underscore need for sex-tailored therapeutic strategies address lasting impact COVID-19.

Язык: Английский