The AI Behind MyTrial
We are transparent about the AI technology we use, how it works, and what its limitations are. This information is provided in compliance with California AB 2013.
Foundation Model
The AI that powers MyTrial
MyTrial uses Claude, an AI assistant created by Anthropic. Claude helps us understand your health information, match you to trials in our database, and explain trial requirements in plain language.
About Claude's Training
Claude was trained by Anthropic on a diverse dataset of text from the internet, books, and other sources. For detailed information about Claude's training data and methodology, please visit Anthropic's documentation.
Learn more about ClaudeHow We Use AI
AI involvement and human oversight for each use case
| Use Case | AI Involved | Human Oversight |
|---|---|---|
| Understanding your profile | Yes | Quality reviews |
| Matching to trials | Yes | Algorithm validation |
| Explaining trials | Yes | Content review |
| Site referrals | Yes | You confirm all actions |
| Eligibility determination | No | Sites determine this |
Training Data
How our AI is trained and what data is used
Important: We do NOT use your personal health data to train AI models.
MyTrial's AI is powered by:
- 1.Pre-trained foundation models - Claude by Anthropic, trained on publicly available data
- 2.Public clinical trial data - Information from ClinicalTrials.gov, a public registry
- 3.Our matching algorithms - Proprietary logic that connects profiles to trials
Your conversations and health profile are used only to serve you, not to improve AI models.
Accuracy and Limitations
What AI does well and where it has limitations
- Pattern matching across many trials simultaneously
- Processing large amounts of eligibility criteria
- Translating medical jargon to plain language
- Providing consistent, available assistance 24/7
- Scaling to serve many users at once
- Cannot guarantee matches are perfect
- May miss relevant trials (manual search recommended)
- Cannot replace clinical judgment
- Does not know everything about your health
- May occasionally produce inaccurate information
Always discuss clinical trial participation with your healthcare provider.
Human Oversight
How we ensure responsible AI use
Regular Audits
Our team regularly reviews AI outputs to ensure quality and accuracy. We monitor for errors and continuously improve our systems.
User Confirmation
Any action that shares your information or connects you with a trial site requires your explicit confirmation. The AI cannot act on your behalf without your approval.
Feedback Loop
We actively collect and respond to user feedback about AI responses. If you notice an error or have concerns, please let us know at ai-feedback@mytrial.ai.