KMAN Publication Inc. (KMANPUB) Applied Family Therapy Journal (AFTJ) 3041-8798 6 Serial Number 30 2025 11 01 Recurrent Neural Network Forecasting of Marital Satisfaction Under Economic Stress Conditions Recurrent Neural Network Forecasting of Marital Satisfaction Under Economic Stress Conditions 1 10 10.61838/kman.aftj.4956 EN Journal Article 2024 05 21 Objective: The objective of this study was to develop and evaluate a recurrent neural network model for forecasting longitudinal changes in marital satisfaction among couples experiencing economic stress. Methods and Materials: This longitudinal predictive study examined 540 married couples from major urban regions of Argentina over a twelve-month period. Participants completed four waves of assessment measuring marital satisfaction, economic stress, perceived stress, depressive and anxiety symptoms, social support, and objective financial indicators. Data were preprocessed and structured into temporal sequences for each couple. A long short-term memory recurrent neural network was trained to predict future marital satisfaction based on prior economic and psychological states. Model performance was evaluated using root mean squared error, mean absolute error, and explained variance, and compared with linear regression, random forest, and support vector regression baselines. Findings: The recurrent neural network significantly outperformed all baseline models, achieving an R² of 0.81 and reducing prediction error by more than 30%. Economic stress was the strongest predictor of marital satisfaction trajectories, followed by perceived stress and social support. The model revealed delayed and cumulative effects of financial strain on marital satisfaction, with peak relational impact occurring two to three months after major stress changes. High predictive accuracy was maintained across socioeconomic groups. Conclusion: The findings demonstrate that marital satisfaction under economic stress follows complex nonlinear temporal patterns that can be accurately forecasted using recurrent neural networks, providing a powerful framework for early identification of couples at risk and informing targeted interventions.

https://journals.kmanpub.com/index.php/aftj/article/download/4956/8922