How to use HERA

Open the Command Center. Type trial criteria, ask a population question, or discuss a specific patient. Use the example dropdown for ready-made prompts.

In the Command Center sidebar, expand Patient roulette and click Draw patient. You will see their name, ID, age, sex, specialty, and encounter count. Hit Discuss this patient to bind that patient to your chat, then ask about meds, labs, trends, or eligibility. To double-check what HERA is citing, open Patient Atlas and use the collapsible Patient lookup panel on the left — enter the ID (e.g. PT-000541) to view encounters, labs, and investigations from the backend before returning to the chat.

Paste inclusion and exclusion rules (e.g. LVEF ≤ 35%, age 18–80). Not sure where to start? In the sidebar, expand Criteria roulette and click Draw criterion for a random example you can use as-is or adapt, then hit Use in conversation to drop it into the chat. HERA parses your rules, starts a background search, and gives you an audit link. Watch the progress bar in chat or open the audit dashboard from Recent searches in the left sidebar (below the roulettes). Every search you run is listed there, so you can return to any audit later, even after starting a new conversation. For cost and speed in this demo, retrieval is narrowed to a max of 3–5 candidates for full LLM chart review — you will not see a large unreviewed cohort in the audit dashboard.

The dashboard lists the capped shortlist (up to 3–5 patients). Pick a candidate, then read their SOAP timeline and criterion-by-criterion ledger. Toggle the copilot to ask follow-up questions. If you disagree with a verdict, use Overrule Verdict and add a short rationale.

Ask population questions (“average age?”, “how many mention heart failure?”) or patient-specific trends (“plot creatinine over time for PT-000121”). HERA can return tables and inline charts in the chat when a visual summary helps.

Visit Patients to open a full chart room outside of a matching task.