Glossary of terms for health data and AI

[Alies Maybee]

Trust the Europeans to get on with it. The German group Data Saves Lives has been working on a glossary of terms around health data and AI - Fachgedöns. Here is the English version. ENG DSL DE Fachgedöns Update 25.pdf

Here is the blog post announcing it in German and in English if you scroll down. link text

Any thoughts? How can we use this? Does it work for us?

[Debra Turnbull]

@Alies-Maybee

After a quick cursory read - some terms apply; some do not.

The biggee for me was the difference between Anonymization and de-Identification. I know there are differences in the meaning and recently learned that Ontario references de-identification in their laws; whereby Quebec talks about anonymization.

Digital Health literacy made me laugh: umm... nope! I like how this paper: Digital Health Literacy as Super determinant of health... defines : digital, health, civic and digital health literacies.

And then, there is GDPR - which has no jurisdiction in Canada (except if stuff comes & goes to Europe). Our privacy laws are different.

On the other hand - this is a great template! in which we can adapt our own Glossary-CA. Awesome literacy tool...

Deb

[Chris Johnston]

@Alies-Maybee @Debra-Turnbull

After skimming, I’d mostly agree with Deb, some definitions are generic enough to work for us, some too specific to a European context, but some are oddly skewed.

• Patient Data for example is defined as data collected by the patient? Perhaps this is a translation error but it’s factually incorrect, since that term is used widely to reference any data collected about a patient by any means, by anyone and in any connection to their health or interaction with healthcare providers, systems, vendors or research.

• Digital twin is also inexplicably confined to virtual representation of a patient, organ or device. Yet in practice it is being used to represent facilities, organizations, systems, populations and clusters. Also not just in a traditionally ‘virtual’ sense but in training, testing and research settings for both people and AI, in algorithmic and interactive forms etc. And it’s important to understand that it may be based on actual or synthetic data, or any combination thereof.

As an example glossary, I think it’s useful in terms of knowing what’s out there in other jurisdictions, but as a base to build on it might be better to look to combining glossaries in use or under development by Canadian universities and institutions.

As a format it’s slick but lacks functionality, a wiki framework that facilitates discovery and comprehension through cross-referencing, internal and external linking, and multiple forms of both static and dynamic content, might prove more user-friendly.