Mental health ML datasets remain English-heavy, non-social media sources scarce

Illustration accompanying: Mental Health Disorder Detection Beyond Social Media: A Systematic Review of Available Datasets

A systematic review identifies a critical gap in mental health ML research: most datasets come from social media, introducing sampling bias and privacy risks. This survey catalogs non-social media alternatives across multiple languages, revealing that available free-text datasets remain heavily skewed toward English and depression detection. The finding matters because practitioners building clinical NLP systems now have a structured resource to identify underexplored data sources, while researchers can see where dataset diversity is needed to improve model generalization across populations and conditions.

Modelwire context

Explainer

The survey doesn't just list datasets; it quantifies the skew. The finding that free-text mental health corpora remain heavily English-dominant and depression-focused reveals a concrete bottleneck for building models that generalize across populations and conditions, not merely a gap in coverage.

This connects directly to two prior findings on data collection bias. The disaster-reporting study from early July showed how sampling methodology itself introduces systematic bias that propagates downstream; this survey applies that lens to mental health datasets specifically. Separately, the MultiSynt/MT release from the same period demonstrated that synthetic translation can compress data efficiency gaps for underserved languages. Together, these stories frame a two-part problem: mental health datasets are structurally biased toward English and specific diagnoses, and practitioners now have both a catalog of what's missing and evidence that synthetic approaches can help fill gaps.

If researchers use this survey to train multilingual depression models on the non-social media datasets cataloged here and report performance parity across languages by end of 2026, that validates the resource's utility. If the survey gets cited in fewer than 50 clinical NLP papers within 18 months, it suggests practitioners still default to social media sources despite the documented risks.

Coverage we drew on

The Course of News Events: A Comparison of Bottom-Up and Top-Down Approaches for Collecting Text-Based Data about Disasters · arXiv cs.CL

This analysis is generated by Modelwire’s editorial layer from our archive and the summary above. It is not a substitute for the original reporting. How we write it.

MentionsPRISMA methodology

Read full story at arXiv cs.CL →(arxiv.org)

Modelwire Editorial

This synthesis and analysis was prepared by the Modelwire editorial team. We use advanced language models to read, ground, and connect the day’s most significant AI developments, providing original strategic context that helps practitioners and leaders stay ahead of the frontier.

Our mission How we write

Modelwire summarizes, we don’t republish. arXiv cs.CL originally reported this story as “Mental Health Disorder Detection Beyond Social Media: A Systematic Review of Available Datasets”. The full content lives on arxiv.org. If you’re a publisher and want a different summarization policy for your work, see our takedown page.