Intersectional Bias Assessment: Synthetic Mental Health Training Data

A synthetic dataset contains demographic and clinical data used to train a classifier for predicting schizophrenia or depression diagnoses. This data was created for a tutorial on intersectional model evaluation presented at the ACM FAccT 2022 conference. The dataset is authored by Marta Maslej and is licensed under Creative Commons Attribution 4.0 International.

Use Cases

• Training classifiers for mental health diagnosis prediction based on synthetic demographic and clinical features.
• Benchmarking fairness metrics and bias detection methods based on the described demographic attributes.
• Conducting intersectional bias analysis in model evaluation based on the combined demographic and clinical data.
• Developing educational tutorials on ethical AI in healthcare based on the dataset's documented use case.

Strengths

• Dataset is explicitly designed for a peer-reviewed tutorial presented at the ACM FAccT 2022 conference.
• License is clearly specified as Creative Commons Attribution 4.0 International (CC-4Y).
• Focus on intersectional bias assessment addresses a specific and documented need in AI ethics.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for large-scale training.
• Last update date is unknown; freshness unverified.

Provenance

Source

openml

Collection Method

Synthetically generated for a research tutorial.

Time Range

Created for ACM FAccT 2022.

Freshness

Last updated date is unknown.

Geography

Spatial coverage is unknown.