# PERI_DEP: A dataset of mother's mental health in Pakistan

## TL;DR

The paper introduces PERI_DEP, a new dataset on maternal mental health in Pakistan, aiming to address class imbalance and improve understanding of perinatal depression.

## Contribution

The novel PERI_DEP dataset includes mental health and socio-demographic data from 14,008 women in Pakistan, using GAN to handle class imbalance.

## Key findings

- The dataset reveals socio-demographic differences between rural and urban women in Pakistan.
- Using GAN oversampling improves the balance of mental health data for better analysis.
- PERI_DEP supports more accurate identification of risk factors and targeted interventions.

## Abstract

Perinatal depression (PND) represents a multifaceted mental health issue that impacts women throughout the perinatal period. Existing datasets have a class imbalance issue, resulting in biased outcomes. In Pakistan, we developed a novel dataset called PERI_DEP. This dataset leverages the Patient Health Questionnaire (PHQ-9), Edinburgh Postnatal Depression Scale (EPDS), and socio-demographic questionnaires to gather information about mental health state and socioeconomic details of women participants in urban and rural areas. Our novel PERI_DEP dataset contains 14,008 samples, and women from Lahore and Gujranwala participated. To tackle the issue of class imbalance, we employed Generative Adversarial Network (GAN) oversampling technique on our data. A key insight derived from this dataset is the comparative socio-demographic divide of women in rural and urban areas of Pakistan. PERI_DEP dataset enhances hospital capabilities by streamlining screening processes, customizing interventions, and enabling researchers to identify risk factors and develop new treatments accurately.