System lifetime for SaMD

[Bhavta]

Good afternoon everyone. I just wanted to find out if anyone would assist me with an approximate system lifetime for a Samd. Please please can someone assist

 

[yodon]

Not an easy answer since bits don't wear out (too often). What's YOUR vision for the market viability life?

 

[Bhavta]

This is an AI software. Does it also apply?

 

[yodon]

Unfortunately, yes. And we're talking over the lifetime of the device, not a particular instance of learning.

 

[Bhavta]

So is there no definitive answer

 

[yodon]

There's certainly no blanket answer for all software applications. These health monitoring apps are probably going to be obsolete in a few years. Image analysis software may be reasonably used for 5 or more years. You do need to define one, though. And that has implications on record retention so choose wisely.

 

[dgrainger]

Could be related to expected lifetime of any software/hardware you're reliant on.

 

[psivakup]

The system lifetime for a SaMD depends on different factors, and some of them consider are:
1. Operating System
2. Technology
3. SOUP or OTS Software
4. HW
5. Product Security
6. Risk

 

[Tidge]

Disclosure: I try to avoid doing much thinking about the regulatory side of SaMD, as I have no current, relevant experience... plus I have some internal biases about "the miracle of software & apps" in the current era(*1). I do want to believe this is true: If the Software really is a medical device, the provider ought to be obligated to notify all users of the software when it reaches "end-of-life".

(*1) As regulatory eyes turn to cybersecurity concerns, I suspect that the window of "easy" marketing "low (medical) risk" software as SaMD will be closing. In the USA, the Consolidated Appropriations Act of 2023 directs the FDA to issue a pre-market guidance within 6 months, and requires the publication of some sort of report within 1 year.. and allocated them millions to do these. Once the FDA lands somewhere, I wouldn't be surprised if they start going after "low-hanging fruit" (like "low (medical) risk" software) to demonstrate the scope of the potential problem. It's obvious to me that some part of the Agency's broader enforcement strategy is to simply google tobacco/nicotine products... for "SaMD manufacturers" who previously filed paperwork with the Agency, I think it is going to be trivial for the Agency to pick large swaths of products to investigate.... even if it is a simple as asking the manufacturer to send the agency SBOMs for all the versions of the software that have been distributed.