A Multilayer Perceptron model achieved 88.9% overall accuracy and 97.7% sensitivity in predicting high social dysfunction among economically disadvantaged Iranian women.
Cross-Sectional (n=780)
A Multilayer Perceptron neural network model accurately predicted high social dysfunction among economically disadvantaged Iranian women, highlighting cognitive functioning, anxiety, and depression as pivotal predictors.
In the year 2022, this research conducted an in-person study involving 780 single or widowed women, aged between 20 and 70, falling within the bottom three economic deciles and possessing varying levels of education. All participants held educational qualifications below a high school diploma and were beneficiaries of charitable financial support in Khorasan province, Iran. The study aimed to investigate the predictive factors of social dysfunction in this specific demographic. Data collection spanned a 12-month period throughout 2022, with participants completing the GHQ-28 questionnaire during their visits to the charity office. Clinical in-person interviews were also conducted to gather comprehensive data. Data analysis was carried out using IBM SPSS version 27. The research employed a Multilayer Perceptron (MLP) neural network model, considering an extensive set of input factors and covariates. These factors included cognitive functioning, anxiety, depression, age, and education levels. The MLP model exhibited robust performance, achieving high overall accuracy and sensitivity in identifying cases of high social dysfunction. The findings emphasized the significance of cognitive functioning, anxiety, and depression as pivotal predictors of social dysfunction within this specific demographic, while education and age displayed relatively lower importance. The normalized importance scores provided a relative measure of each covariate's impact on the model's predictions. These results furnish valuable insights for the development of targeted interventions and evidence-based policies aimed at addressing social dysfunction and promoting societal well-being among economically disadvantaged, single or widowed women. Notably, the research underscores the potential of MLP modeling in social science research and suggests avenues for further research and refinement to enhance the model's predictive accuracy, particularly for cases of low social dysfunction.
Bagheri et al. (Thu,) conducted a cross-sectional in Social dysfunction (n=780). Multilayer Perceptron (MLP) modeling was evaluated on Overall accuracy in predicting social dysfunction. A Multilayer Perceptron model achieved 88.9% overall accuracy and 97.7% sensitivity in predicting high social dysfunction among economically disadvantaged Iranian women.