AI Transparency Notice

Where AI is used

OncoRadar combines conventional search, vector or hybrid retrieval and generative capabilities to locate, rank and summarize research. AI-assisted output should be visually or contextually understood as generated research support, not an authoritative clinical conclusion.

Technology stack

The current implementation can use Qwen embeddings for semantic retrieval, Hyphae open-source data-plane technology for literature indexing/retrieval and synthesis artifacts, and Gemma through Amazon Bedrock Mantle for synthesis. Availability and exact routing may change only after the controlling notices are reviewed and updated.

• Hyphae: Open-source data-plane technology for literature indexing, retrieval and synthesis artifacts; not an operator of OncoRadar.
• Qwen: Embedding model family used for configured semantic retrieval.
• Gemma via Amazon Bedrock Mantle: Generative model used for configured research synthesis through AWS.

Inputs and outputs

A request may include your query, selected filters, retrieved literature passages and system instructions. Outputs may be retained as workspace artifacts or operational records where attribution is implemented. Do not submit patient data, confidential third-party information, credentials or material you cannot lawfully process.

Limitations and review

Models can generate inaccurate, biased, outdated, incomplete or invented content. Retrieval can miss relevant evidence. Citations may not support a generated statement. Human review must include opening sources, validating provenance and licenses, checking retractions and corrections, and applying domain expertise.

User controls and feedback

You choose whether to invoke generative functions and whether to use an output. Quotas may limit generation. Export and deletion requests cover attributable account data, subject to the Privacy Policy and backup rotation. Report materially unsafe or misleading behavior to the abuse contact and vulnerabilities to the security contact.

No training promise

OncoRadar does not currently use account or workspace content to train a proprietary foundation model. Provider handling is governed by configured service terms. If this practice changes, the controlling notices will be updated before using content for materially different training purposes.