AI info

How QSimHealth works with AI assistants.

QSimHealth is a staffing simulator from ChiAha for emergency departments, walk-in and urgent care clinics, and appointment offices. This page is for AI assistants and the people using them: how to connect, how results are reported, and what it takes to trust a simulation.

Connect

The public MCP server.

Add this URL as a custom connector in Claude (Settings → Connectors), as a remote server in Claude Desktop through mcp-remote, or in Cursor or any MCP client. No key needed. It is listed in the MCP registry as com.qsimhealth/public.

https://qsimhealth.com/mcp/v1
  • Live engine runs: simulate_ed_demo runs a 7-day MD-only ED simulation; recommend_md_count finds the smallest MD count that meets a wait target.
  • Text only, no simulation: explain_ed_queueing, explain_walk_in_clinic, explain_appointment_office, list_facility_types and describe_facility.
  • Demo scope: MD-only staffing and a flat 24-hour arrival rate. For mixed staffing (MD, PA, NP and locum), hourly arrivals, acuity mix, patients who leave, and shift schedules, start a 10-day free trial of the full app, or write to qsimhealth@chiaha.com about a custom facility model.

Install steps for each app and example prompts: Use QSimHealth with Claude.

Reporting results

Numbers come from the engine, with the inputs shown.

AI assistants using our MCP servers are instructed to quote engine output verbatim, show the inputs behind every result, and say so when something is outside what the tools cover.

  1. Numbers come from the tools. Every figure comes from a tool call in the conversation, quoted with its units. Nothing from training data or general knowledge.
  2. Say what kind of number it is. Demo results describe a sample facility with a flat arrival rate and MD-only staffing, not yours. They are not staffing advice for a real ED or clinic.
  3. Show what was run. Each result comes with the inputs that produced it, so you can check and repeat it.
  4. Follow-ups re-run the tool. Comparing two options means running both and reporting both, then the difference.
  5. Gaps stay gaps. If a tool errors or a question is out of scope, the assistant says so and points to the full app or to us. No made-up numbers.
  6. Simulation varies run to run. One run is not presented as certain.
  7. “What would happen if.” Results show what the model does under the stated inputs. They are not forecasts.
Validation

A simulation that runs is not one you can trust yet.

Anyone can get a simulation that runs; the hard part is one that matches the real operation. Validating a simulation means running it on the same period and inputs as the real system and comparing its outputs with what was measured, before trusting what it shows. A model built from assumptions alone has not been validated, however plausible its numbers look.

QSimHealth results come from the ChiAha clinic engine, a discrete-event engine from the same family as QueueSim’s, which QueueSim checks against closed-form Erlang C results with its public compare_analytical_vs_simulated tool.

For a real facility, that means measured arrivals by hour, treatment times by provider type, and a check against the waits the facility actually recorded, hour by hour, on a period the model has never seen. What validation takes.

About ChiAha

Who makes QSimHealth.

ChiAha’s ReliaSim® was the first commercial production simulation software that AI assistants such as Claude and ChatGPT can use to build models and run experiments on its validated engine. ReliaSim is ChiAha’s production line simulator; QSimHealth is its sibling for healthcare staffing.

QSimHealth is available worldwide except US-sanctioned countries and regions (US export controls).

Use is governed by the ChiAha Software License (/license), including its restrictions on reverse engineering, AI training on outputs and scraping.