OAC / HEALTH LAB

Operational model · Synthetic preview

Signals made
legible.

Explore how eight operational counters influence the OAC model’s operator-attention advisory. The Python service scores your input; this page does not simulate a result.

Eight inputs. A fixed model. An inspectable source ledger.

Checking service identity…

Operational inputs

No clinical data

Invented defaults. Change counters, then request an advisory.

Synthetic presets
Configuration & integrity
Fraction, 0–1
Count, 0–20
Seconds, 0–86,400

Sent only to this service.
No browser storage or automatic retries.

The examples are invented—not live device readings. This page does not discover devices, ingest messages, or retain submitted features in browser storage.

Operator advisory

Non-authoritative

A synthetic attention score, not a clinical probability.

Awaiting a request

—

Synthetic attention score · 0–1

Advisory threshold—

Request an advisory to see the server’s result. No score has been generated yet.

Feature contributions

Normalized logit terms
No result yet
—

Positive terms raise the model’s linear score; negative terms lower it. These are not causal explanations or percentages. The intercept is not shown.

Source & artifact identity

Canonical Python source in GitHub. Model and synthetic dataset in Hugging Face. Revisions below are returned by the running service.

Canonical GitHub source ↗Model & kernel ↗Synthetic dataset ↗Identity manifest ↗Readiness endpoint ↗
Service identity
Checking…

A reported revision and a matching local artifact hash are not independent deployment attestation, clinical validation, or proof of physical device compatibility. No signed operational receipt is minted by this preview. This page checks the response hashes’ format; the independent release verifier checks their equality against canonical inputs and outputs.