FAQ
The questions that come up most often on a first call. If yours is missing, write to [email protected].
What does it cost?
The pilot is free and ends in no commitment. There is no company yet and nothing to sell, so there is no price list either.
How is this different from ordinary document search?
Ordinary search matches a string of characters. A record written up as "painful shoulder syndrome" does not come back for "shoulder pain", because those words are not in it, and finding it means remembering how it was described at the time.
Here a description is compared with a description. Every document is given a numerical representation of its own description, the query is given one of the same kind, and the result is a measure of how close the two are. That is why the same record comes back without sharing a single word with the query.
The second difference is in the answer: every result carries a number saying how close it is to the query, and a reference to the record in your own archive. When nothing clears the threshold you see no results rather than the nearest guess.
Is the tool a medical device?
No. The tool searches for documents in a facility's own archive and presents them with a reference to the source document. It does not assess a patient's condition, does not propose a course of action, and is not used for diagnosis or for making professional decisions.
Every professional activity concerning a patient, and every decision about how to proceed, remains with the facility's staff.
Do we have to hand over patients' personal data?
No, and that is the first choice. Under option A you anonymise the set yourselves, so formally we receive no personal data and no processing agreement is needed. Under option B we work inside your system and the data never leaves you.
Do you plug into our records program?
No. We work on an export of files, alongside your program. We install nothing on your equipment and change nothing about your infrastructure.
What if our program cannot export?
That is the first thing worth checking, which is why we ask for a sample file with two cases on the very first call. If the export is awkward, there is usually something to be done. If there is none at all, we will say so plainly.
How much of our time does it take?
About three calendar weeks, of which your work is a few hours: pointing out the range of records, anonymising them with our help, and twenty queries at the assessment stage.
What do we get at the end?
A written assessment, whatever the result, including when it goes badly. The report stays with you.
How large does the archive have to be?
At least several dozen documents with text descriptions. On a smaller set there is nothing for search by description to work with, and the assessment would not be honest.
Do you train a model on our data?
No. We train no model on your set, we do not combine it with another facility's, and we use it for nothing except the pilot with you.
What happens to the set after the pilot?
We delete it within fourteen days of the end, or immediately on your written request. We confirm the deletion in writing.