Datasetpublic

Elastus/africa-mental-health-eritrea

Perinatal Depression & Maternal Mental Health Dataset Abstract This dataset provides 30,000 simulated perinatal women records (10,000 per scenario) from sub-Saharan Africa. Each record contains 45+ variables including depression screening, EPDS scores, risk factors, treatment, and maternal/infant outcomes. Three settings: urban integrated MH (24% detected), district hospital (5%), and rural health centre (2%). 1. Introduction Perinatal depression… See the full description on the dataset page: https://huggingface.co/datasets/Elastus/africa-mental-health-eritrea.

sourceHugging Facecc-by-4.0updated 4mo agoView on Hugging Face

0likes8downloads

Dataset Card

Perinatal Depression & Maternal Mental Health Dataset

Abstract

This dataset provides 30,000 simulated perinatal women records (10,000 per scenario) from sub-Saharan Africa. Each record contains 45+ variables including depression screening, EPDS scores, risk factors, treatment, and maternal/infant outcomes. Three settings: urban integrated MH (24% detected), district hospital (5%), and rural health centre (2%).

1. Introduction

Perinatal depression affects ~32% of women in SSA but remains massively underdiagnosed and untreated. The EPDS is the most validated screening tool but is rarely used routinely. Stigma is a major barrier to disclosure. IPV, unplanned pregnancy, poor social support, and food insecurity are key risk factors. Depression worsens maternal and infant outcomes including breastfeeding and bonding.

This dataset is entirely simulated. It must not be used for clinical decision-making.

2. Methodology

2.1 Parameterization

Parameter	Value	Source
Perinatal depression SSA	~25-33%	OJPHI 2025
EPDS cutoff	≥13	BMJ 2020
IPV prevalence	~25%	UNFPA
Stigma barrier	~60%	Frontiers 2024
Treatment access	<5% most settings	WHO

2.2 Scenario Design

Scenario	EPDS	Counselling	Psychiatry	Detected
Urban MH	Yes	Yes	Yes	24%
District hospital	No	No	No	5%
Rural HC	No	No	No	2%

3. Schema

Column	Type	Description
id	int	Unique identifier
age	int	Maternal age
perinatal_period	categorical	antenatal / postnatal
epds_score	int	Edinburgh Postnatal Depression Scale
depressed	binary	Meets depression criteria
depression_severity	categorical	none / mild / moderate / severe
screened_epds	binary	EPDS screening done
depression_detected	binary	Depression identified
treatmentreceivedany	binary	Any treatment received
suicidal_ideation	binary	Suicidal thoughts
stigma_barrier	binary	Stigma prevents disclosure
bonding_difficulty	binary	Mother-infant bonding issues

4. Validation

Key validation checks:

—Depression prevalence: ~32% across scenarios ✓
—Detection cascade: 24% → 5% → 2% ✓
—EPDS screening: 71% urban → 0% elsewhere ✓
—Stigma: ~60% barrier ✓
—Risk factors enriched in depressed group ✓
—Worse outcomes for depressed mothers ✓

5. Usage

python

from datasets import load_dataset
dataset = load_dataset("electricsheepafrica/perinatal-depression-maternal", "district_hospital")
df = dataset["train"].to_pandas()

6. Limitations

—Simulated: Not from real clinical data.
—No longitudinal: No trajectory tracking.
—Simplified: No detailed psychometric profiles.
—EPDS only: No PHQ-9 or other tools.

7. References

1.OJPHI (2025). E-Screening perinatal depression Kampala.
2.Nature (2024). EPDS validation Cameroon.
3.Frontiers (2024). Screening tools maternal MH SSA.
4.BMJ (2020). EPDS accuracy meta-analysis.

Citation

bibtex

@dataset{esa_perinatal_depression_2025,
  title={Perinatal Depression and Maternal Mental Health Dataset},
  author={Electric Sheep Africa},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/electricsheepafrica/perinatal-depression-maternal}
}

License

CC-BY-4.0