Comparison with Different Algorithms: MDD Diagnosis from rs-fMRI Data

Dan Long's dataset contains functional connectivity matrices derived from resting-state fMRI data for 1,160 participants, including 597 patients with major depressive disorder and 563 healthy controls. The data was used to evaluate a novel Graph Contrastive Learning framework (EC-GCL) against conventional machine learning and GNN models. The dataset was last updated on May 5, 2026.

Use Cases

• Benchmarking graph neural network models for MDD classification based on functional connectivity matrices.
• Identifying key brain regions linked to MDD pathophysiology based on the interpretable features mentioned in the study.
• Training adversarial graph contrastive learning models to preserve spatial specificity of brain regions.
• Comparing the performance of edge convolution encoders against conventional GNNs for neuroimaging data.

Strengths

• Includes data from 1,160 participants, providing a substantial sample size for analysis.
• Clear diagnostic labels for 597 MDD patients and 563 healthy controls.
• The associated study reports a specific model performance metric (AUC = 71.2%).
• Data is shared under a permissive CC-BY-4.0 license.

Limitations

• The dataset is very small (5.5 KB), suggesting it may contain only summary or model results, not the raw connectivity matrices.
• Row count and column-level documentation are absent; field semantics must be inferred after download.
• The description focuses on the methodological study, so the actual data content and structure require manual inspection.

Provenance

Source

Dan Long via figshare

Collection Method

Derived from resting-state functional magnetic resonance imaging (rs-fMRI) data.

Freshness

Last updated 2026-05-05 17:43:04; freshness should be verified.