Evidence before marketing
We explain both the original validation findings and later real-world studies that found much weaker standalone diagnostic performance.
About this project
A more useful RAADS-R experience starts with honest boundaries
This independent website helps adults complete and understand the RAADS-R as a structured self-reflection tool. It is not a clinic, healthcare provider, diagnostic service, or substitute for a qualified professional.
Anonymous, resumable access
Progress stays in your browser while you answer. A completed result is saved under an unguessable permanent link so it can be reopened without an account.
AI only with consent
Answers are sent to the report service only after a clear opt-in. Model routing requests zero-data-retention endpoints and denies data-collecting providers.
Supportive, not diagnostic
Results focus on patterns, examples, and low-risk reflection prompts. We do not estimate a probability of autism or claim clinical certainty.
Model transparency
The score is calculated by rules, not guessed by AI
Total scores, domain scores, response timelines, and supporting question references are calculated by deterministic code. A language model cannot change those values.
Optional detailed-report wording currently uses OpenRouter to route requests to a Google Gemini Pro model. Each request asks for zero data retention and denies provider data collection. If the model is unavailable, the service returns a structured fallback report instead of silently using an unrestricted model.
Models and versions can change as providers update availability. We will update this page when the report model or data-handling boundary materially changes.
Research sources
Scoring and limitations are informed by the 2011 international validation study and later studies of adult referral populations. Source links are provided so readers can inspect the evidence.