Software in Medical Devices, by MD101 Consulting

We had a preview of this in a previous article in 2021, TGA repeats it in the Regulatory changes for software based medical devices - Version 1.3, June 2024

Software for diagnosis can be in class I

If the software for diagnosis is used by lay persons or professionals not relevant to diagnose the disease, the software is in class IIa at least.

But if the software for diagnosis is used by relevant professionals, regulatory classes are shifted one level down and we can have software in class I:
A software-based medical device that is intended to provide information to a relevant health professional so they can diagnose a disease or condition will be classified as:

• Class IIb if the disease or condition may:
  • a. lead to the death of a person without urgent treatment
  • b. lead to a severe deterioration in the state of a person’s health without urgent treatment
  • c. pose a high risk to public health.
• Class IIa if 1. does not apply, but the disease or condition:
  • a. is a serious disease or serious condition
  • b. may pose a moderate risk to public health.
• Class I if 1. and 2. do not apply.

Thus, SaMD can be in class I if the condition isn't a serious disease / condition and doesn't pose a moderate risk to public health.

Interestingly, this classification doesn't fully align with the SaMD classification scheme of IMDRF.
The three levels of Significance of information provided by SaMD to healthcare decision are replaced by the relevance of end-users using the software. While the three levels of patient condition are quite similar in the TGA and IMDRF documents.
Maybe a possibility to modify the IMDRF document and explain how relevance of the professional user can align with Significance of information provided by the SaMD?

Software excluded from or exempted of Australian regulation

Did Australian regulators borrowed this idea from the FDA? I don't know. But we retrieve similar exclusion / exemption rules for SaMD:

• FDA has a guidance on Clinical decision support software, we discussed it here,
• TGA has also a guidance on Clinical decision support software.

Excluded Software

Here we retrieve the classical rules of exclusion, based on the medical device definition found in the regulation. A rapid comparison between EU MDR, FDA and FDA shows that we retrieve the same software used for storage, communication and display of information without an intended purpose for diagnosis or treatment.
We also find Software module that provides alerts for adverse drug interactions based on established rules and guidelines both excluded by the FDA and TGA ; but regulated by the MDR, thanks to the interpretation of the definition of medical device by the ECJ in their great wisdom.

Exempted software

Here we retrieve software, which meet the definition of medical device, but are exempted if they meet the following criteria: A clinical decision support system is exempt if it meets all 3 of the following criteria:

1. does NOT directly process or analyse a medical image or a signal from another medical device (including an in vitro diagnostic device); and
2. is solely used to provide or support a recommendation to a health professional about prevention, diagnosis, curing or alleviating a disease, ailment, defect or injury; and
3. does NOT replace the clinical judgement of a health professional in relation to making a clinical diagnosis or decision about the treatment of patients.

FDA has four criteria, TGA has three criteria. They are quite similar. And they exempt quite a bunch of clinical decision support software from full regulatory oversight.

EU MDR does not.

MDR is wrong, AI Act is on its way

And we continue the MDR-bashing on this blog. Obliged to do so, seeing the absurdities of the European regulation with regard to medical device software.
MDR will continue to dig the grave of SaMD manufacturers as long as:

• Exemptions aren't defined for CDSS,
• Rule 11 doesn't allow Software intended to provide information which is used to take decisions with diagnosis or therapeutic purposes to be in Class I, when the patient condition isn't serious.

While US FDA, Australian TGA, and UK MHRA (still under the MDD regime) allow software to be in class I or exempted, the EU MDR forces SaMD manufacturers to pass a long and costly CE certification scheme in class IIa. First punishment.
Such SaMD, if they contain AI algorithms, will also have to pass the CE certification scheme according to the AI Act. Second punishment.

SaMD manufacturers will continue to conscientiously avoid the EU regulatory hurdles. They will place their software on fields where grass is greener.

EU regulators, something must be done urgently for SaMD and CDSS.
Or you won't have much to regulate.