Mahal Systems AgentClinic
Fits diagnostic or triage assistants that must gather information by asking questions rather than reading a full vignette. Borrow the four-agent setup (patient, doctor, measurement, moderator), the bias injection options, and the structured case JSON.
What it asks the agent to do
- interview a simulated patient to collect history and symptoms
- order exams and imaging from a measurement agent
- reach a diagnosis under incomplete information within a turn budget
- use tools such as a persistent notebook or retrieval across cases
What to borrow from it
user simulatorcase template
Caveats
Cases are MedQA exam vignettes (215) and NEJM case challenges (120) rewritten by GPT-4 into patient scripts, so the base sets are synthetic; the optional AgentClinic-MIMIC-IV suite uses real records and needs PhysioNet credentialed access; the moderator is an LLM judge of diagnosis only; the repo has not been updated since September 2024.
Details
| Workloads | healthcare, multi agent, knowledge qa |
|---|---|
| Interaction | multi turn simulated user |
| Environment | static |
| End state checked | No |
| Tools | LLM patient agent, LLM measurement agent, notebook, web and textbook retrieval |
| Grading | llm judge |
| Size | Not published |
| Data | mixed |
| Personal data | none |
| License | MIT |
| Harness | own repo |
| Status | active |