← Research programmes
Agent · clinical decision supportAwaiting approval
An agent that prepares the evidence for a tumour board
Status. Research use. Not in use in care. Treated as within the scope of EU MDR Annex VIII Rule 11, not CE marked, and no conformity assessment has been carried out.
The setting
A tumour board has forty minutes and eleven cases. For each one, somebody has to have gone back to the guideline. Usually that somebody is whoever had the evening free.
What we did
- Built a working prototype with an Italian local health authority. A clinician enters a case: age, sex, diagnosis, stage, ECOG, comorbidities, histology, what has been done and what is missing. The system searches published AIOM and ESMO guidance and returns a formatted document for the Gruppo Interdisciplinare Cure to read in the room.
- The partner is not named here. There is no written agreement yet, and naming a public authority without one asserts something they would have to stand behind.
The output is organised into sections including a proposal and a recommendation, with investigations graded by necessity, over a signature line for a Dirigente Medico. That is patient-specific information used in a therapeutic decision. Rule 11 classifies software providing information used to take decisions with diagnostic or therapeutic purposes as at least class IIa, and a clinician reviewing the output afterwards does not remove it from scope. So we treat it as in scope, and it does not go into care while that assessment is open.
What exists now
- A prototype. The pilot is waiting on regional approval, and the agent is not in use in care.
- The public interface and its backend relay have been taken offline while that approval is outstanding. They were removed rather than left reachable behind a disclaimer, because a disclaimer does not change a stated intended purpose, and MDR Article 2(12) makes a website part of that statement.
What it does not do
- It is not in use and it has never informed a clinical decision.
- It is not CE marked, it has no conformity assessment, and it is not a medical device on the market.
What has not been established
- Whether its guideline retrieval was complete or correct.
- Whether it changed what a panel discussed.
- Whether it saved preparation time.
- Concordance with an unaided panel.
- None of it was measured. The honest reading is that the regulatory question arrived before the evaluation question did.
Where it goes next
- The open research question was never answered, because the regulatory question arrived first: does a structured evidence brief change what a multidisciplinary panel discusses, or how long it takes to get there?
- Nobody has measured that, and answering it does not require deploying anything. Briefs can be prepared offline and sessions compared, which keeps the study entirely outside the question of what a product would need.
- If the answer turns out to be yes, then what to build becomes a live question, and that is a regulatory programme with a budget rather than a prototype. That is a later problem.
What we are looking for
Co-investigators to design that study, and a second tumour board willing to take part.
Get in touch →