MEMOyA SUPER MDT

A meeting that never lies. Every word has a source, every contradiction is plain to see, and the case speaks for itself when the result the team has been waiting for arrives.

The doctor decides, the AI provides the evidence.

Today’s MDT meeting: the coordinator spends the afternoon hunting for PDFs, the doctors read the case file in the car on their way to the meeting, and a third of the decisions are postponed because ‘the RAS is missing’. Nobody’s lying; it’s just that nobody can see the whole file at once. So we’ve set up SUPER MDT.

What you’ll see when you open it

• Meeting overview. 15 patients; for each one, at a glance: ready, supporting documentation missing, or no decision can be made. Green indicates documentation is provided by a reference; red indicates a discrepancy between two reports; an empty space means it’s missing, and next to it is noted who will provide it.
• Case. Instead of a ring binder, you get a question for the consultation, discrepancies with two citations side by side, what’s missing and who’s dealing with it, CEA with trend, and the patient’s full history on a single timeline. Drag and drop a new report and the case recalculates automatically.
• Bea. You ask a question verbally or in writing. She’ll reply solely from this patient’s file, citing a source for every sentence, and if she doesn’t have one, she’ll say so straight away. No making things up.
• Meetings. Patient by patient, a timer showing what’s holding up decisions, who assesses what, voting, tasks and minutes. Nobody has to retype anything.

A consultation that never sleeps. In partnership with InterSystems IRIS for Health. Hospital systems match results with orders. We match them with what the consultation is waiting for. The laboratory sends standard HL7 data; IRIS translates it into FHIR, validates it, stores it and, via Vector Search, compares it with the open items on the consultation agenda. Is there a match? An FHIR Task with Provenance is created, and a tab titled ‘The result the consultation was waiting for has arrived’ pops up on the doctor’s case page. A person closes it with a single click. On a live IRIS instance: 3.49 seconds from the HL7 message to the Task.

Live on IRIS: FHIR R4 repository, validation, Provenance, HL7 v2 Interoperability, HL7 → FHIR, Vector Search, Embedded Python, Task, log entry.

The only rule: if it’s not in the file, it doesn’t exist. If a question for the consultation is missing, it remains blank and is filled in by the doctor. The AI does not decide on treatment and does not make any assumptions. That is why it can be trusted.

The data is synthetic and test data from Ostrava University Hospital, clearly labelled throughout.