Oncologic quality
Case volume by month
Clavien–Dindo distribution
of complications gradedBreakdown by tumor type
within current filters| Tumor type | Cases | Complication rate | Major (≥ IIIa) | Median postop stay | Median OR time | 30d mortality |
|---|
Most recent cases
click a row to open| Date | Type | Procedure | Approach | ASA | Clavien–Dindo | LOS (d) | Status |
|---|
New case
De-identified entry — use a chart reference, not a name.
Follow-up & survival
add a visit each time you see or hear from this patientSaved when you click "Update case" above.
Saved when you click "Update case" above.
| Code | Date | Patient | ID / MRN | Hospital | Type | Procedure | Clavien–Dindo | LOS (d) | OR (min) | Status | Data |
|---|
Backup & capacity
the registry lives inside this page — keep a copy outside itFull backup
One file with every case, every follow-up visit and the private name/ID store. This is the copy that can rebuild the registry: the CSV is for analysis and does not carry the visits one by one.
Restore
Reads a backup file and writes back only the cases that aren't here. Anything already in the registry is left untouched, so restoring twice changes nothing.
Prototype · not live
The multi-surgeon registry
This tab is a walkthrough of the system designed to let colleagues register their own
patients, built and tested but not yet deployed. Everything below uses invented cases —
none of your data appears here. The registry you are using today, in the other tabs,
is unaffected.
How it is put together
Cloud Run · Firestore · Identity Platform
The browser never reads the database directly — the rules deny it outright. Every request
passes through one place on the server that decides what to return, so there is no second
copy of the permission logic that can drift out of step with the first.
The same case list, seen by three people
switch the roleWhere it stands
✓
Built and tested. Case entry, follow-up with organ-specific outcomes, dashboard, colleague accounts, and an append-only log of who changed what. 36 automated checks, including that a colleague cannot see another surgeon's patients or learn who entered a case.
1
A Google Cloud project. Billing enabled and a region chosen — São Paulo or Santiago; there is none in Colombia. Running cost at this volume is a few dollars a month. Deployment is then one script.
2
Ethics committee approval before colleagues are given access. Retrospective chart review counts as no-risk research, and for that category the committee may waive individual consent.
3
Who owns the consolidated database — you, the university, or a clinic. The design does not depend on this: the project can move to an institutional account later without touching the code.