The new, so-called EU MDR outlines significant changes to regulations for medical device companies that want to sell their product in the EU. There are a number of other articles describing best-practices in the lead up to this change (hint: start now!). However, beyond that, we also view this an opportunity for AI to significantly ease the burden the regulation places on companies.
With the new regulation there are three broad challenges where we think AI can significantly help.
Data, data, data
One of the bigger challenges involves the efforts to perform “Clinical Evaluation” and “Post-Market Clinical Follow-Up.” In both cases, the main idea is to surface important results for safety and efficacy, identify any evidence gaps, etc. They are meant to follow a pre-defined plan, part of which involves surfacing data from the literature.
As the new regulations involve more emphasis on data quality and frequency (more on this below), it becomes necessary to have a repository of validated results to pull from, which can be automatically updated. We believe AI is ideal for this.
In the Evid Science scenario, the system surfaces results, and the people can verify and save them to their personal (or team) repository of evidence. In this manner, the AI is doing the heavier lift of finding the evidence, and the human validates it. Further, this evidence is automatically centralized, so the team can search, analyze and build content directly, significantly reducing redundancy.
Compounding the challenge of Clinical Evidence generation and Post-Market Clinical Follow-Up is that fact that these reports likely need to be updated year-over-year. Without tools to centralize the data and make it easier to find results, this can increasingly become too onerous.
Again, AI can help. In our case, our system will alert you when it finds new evidence that might be relevant. Because the AI is focused on pulling out results, and not just pointing you to papers, it’s much easier to then screen what’s relevant. When the machine takes proactive steps to help users update their repository of evidence, as it comes up, this eases the herculean effort required to do it all at once, before the yearly submission. In this way, the AI helps mitigate the challenge of yearly submissions by helping efficiently find and incorporate evidence as it comes up, making it a bite-sized task.
Many Indications means Many Reports
Finally, there is the reporting itself. Many have argued that the interpretation of the EU MDR is such that it will require annual submissions, for each device, for each indication!
That could mean hundreds of reports for a single device (SKU)… So, if we do the math that is a huge number of reports to generate. Thankfully, AI can help there too. More specifically, AI can help a user by providing the data to plug into templated reports. That is, the AI can surface and include new evidence, and human simply needs to provide a template around the evidence. Then programs can combine the chosen evidence and template to produce a document (which can then be polished and submitted).
Here is a very simple video showing how a template is defined and the data is plugged into it. Then, as the AI finds new results, the human can simply choose to include it, and the report would automatically update. In this way, the reporting itself is more about writing the color and context, rather than plugging results into a table and generating citations (the machine should and does, do this…).
It’s clear MDR will impose significant time and resource challenges on organizations (we’ve highlighted just three of them here), but we believe AI is a tool to reasonably manage these challenges.
If you’d like to learn more about how Evid Science’s AI can help, please drop us a line!