Dataset: Dead mental health startups, 2000-2026, coded on 18 fields

A dataset of 542 digital mental health startups that failed between 2000-2026 reveals that who pays for the service is the strongest predictor of survival: consumer-funded companies had a 53% failure rate versus 21% for institution-funded ones, while having a medical co-founder made no difference in outcomes. The open-access dataset includes detailed coding across 18 fields including business model, funding, and reasons for failure, with transparent methodology including LLM-assisted classification and written reasoning available for verification.