CE Marking Algorithm for an IVD use only Medical Device

[F?lix_M]

Hello,

I am working as an Intern in a company currently developing an IVD kit. So I am working on the technical documentation for CE marking and I have few questions I hope someone can help me with.

The device is an IVD use only. The customer uses it and get quantitative results which are not directly interpretable. Then, these results are sent to experts that use a method (~algorithm) to translate these results into YES/NO diagnosis results.

Do the technical files / IFU need to include a verification validation for the algorithm also ? Even though it is used by experts within the company and not the customer.
Is an internal SOP describing how the algorithm works is enough ?

Thank you in advance .

Félix

 

[c.mitch]

Hi Félix
Your device falls into the scope of IVD directive. The associated softwate might fall also into the scope hence it can be seen as an accessory or a part of the IVD system as a whole.
Software validation is one of the essential requirements of IVD directive. Thus, yes, this software should be validated.

This is just my opinion based on your brief description of the device. To help you to assess whether your software is IVD or not, have a look at the guidance Meddev 2.1/6. It contains decision trees to help you answer this question.

 

[F?lix_M]

Thank you a lot Mitch for this reply. Very helpful.

I read the meddec guidance and they don't seem to make any differences either if the "software" is used by the customer or by the company to get interpretative results.

In our case, customers (labs) would get raw data using our kit, send them back to us, then we use our algorithm to get results and tell them YES/NO. So this is half IVD product / half service and the algorithm is used only in the "service" part and will never be used by the customer.

This is why we think an internal document describing the algorithm must be added in our QMS while this is not required in STED / IFU (but the SOP would be mentioned in it).

If anyone could confirm this is valid, would be great !

F?lix

 

[c.mitch]

Hi F?lix
The status of medical device doesn't depend on the location of its use.
I've already seen companies with a business model like yours. They develop and CE mark software only used inside the company.
Thus to get the CE mark of their internal software they build/maintain a STED for it.
Having a SOP to explain how to use/manage it is also preferable and could be added to the STED like the IFU.

 

[F?lix_M]

Actually, our software falls into this category from meddev 2.1/6 :

"In the case where software is necessary to render raw data, obtained from an IVD by means of in vitro examination of body samples,
readable for the user, this software is to be considered as an accessory to an IVD when it is specifically intended to be used together
with this IVD to enable it to be used in accordance with its intended purpose.
Example: software intended for the analysis and interpretation of ELISA reader optical density results, line patterns or spot patterns
of a blot."

Since our algorithm is very simple but contain some "keys" we wish to protect we can also take this into account :

"Note 3: Software intended to modify the representation of available IVD results is not considered as an IVD medical device as per
decision step 5 of Figure 1, e.g. basic operations of arithmetic (e.g. mean, conversion of units) and/or plotting of results in function
of time, and/or a comparison of the result to the limits of acceptance set by the user."

In fine, we will consider our algorithm as an accessory of the IVD and build a proper STED and internal SOP.

Thank you again for helping me. I wish i could return the favor... maybe next time !

F?lix.

 

[Ronen E]

Thank you again for helping me. I wish i could return the favor... maybe next time !

Actually you can - by donating to Elsmar (can be as little as $10). This is the platform that enabled the assistance you received, and naturally it has its running costs.

(I'm a volunteering moderator here and not involved in the forum financials in whatsoever way).


Ronen.

 

[F?lix_M]

That's true, I doubt my boss would agree tho but I would do it !

F.

 

[dgrainger]

Hi,
Unlike the MDD the IVDMDD does not have a requirement for validation of software (ER12.1a). It does have a performance requirement (ER 6.1)