System Requirement - FDA - Software Validation on various OSs

[rstrina]

To all,

Who could give me advice on what the technical explanation of writing a system requirement such as

Software xyz has to run on

Windows 98 SE
Windows ME
Windows 2000
Windows XP Pro
Windows XP Home

REALLY MEANS form a Software Testing point of view? How can the Software test for these system requirements, IF THERE ARE NO SPECIFIC SOFTWARE requirments?

If anyone could point me in the right direction to get a technical explanation as to why testing Windows ME will suffice and include Windows 98 SE or if the software works on Windows XP Pro, it WORKS on Windows XP Home. Who or where can I find A RATIONALE that will uphold to an FDA inspection.

Any hints would be greatly appreciated.

Sincerely, Renate Strina

 

[Marc]

Is this from a written requirement or what? You say "...IF THERE ARE NO SPECIFIC SOFTWARE requirments..." so I'm confused.

 

[Al Rosen]

Is this from a written requirement or what? You say "...IF THERE ARE NO SPECIFIC SOFTWARE requirments..." so I'm confused.

Marc, I think that Renata is alluding to the regulatory requirements for software validation. If you have software that is supposed to run on all windows systems, you need to validate it on all systems or justify why the validation on all systems is not required. I think that we need to have more information before anyone can comment intelligently. Assuming that I am correct, although there is no software requirements there are system requirements (i.e. end use or user defined) that the software must be validated to.

 

[Marc]

I think you're right. What perplexed me was the 'no requirements' statement. My 'assumption' would be that the software would have to be validated on any system they advertise, or otherwise claim, it will run on.

It appears to me the person is writing software and is being required to validate it on all the listed systems. Is that how you interpret it?

rstrina - Can you give us some clarification?

 

[Al Rosen]

I think you're right. What perplexed me was the 'no requirements' statement. My 'assumption' would be that the software would have to be validated on any system they advertise, or otherwise claim, it will run on.

It appears to me the person is writing software and is being required to validate it on all the listed systems. Is that how you interpret it?

Precisely!

 

[rstrina]

System requirements - Software requirements, verification, validation

I want to clarify my post:

The system requirements are that the software (medical software) is supposed to run all the operating systems listed, but so far the programmers haven't written ANY software requirements for the operating systems. Of course they have written other software requirements specific to the software itself.

And my question now is, how and more specifically WHY I would have to test the software on Windows ME AND Windwos 98 SE, when the operating systems are so similar and one was built on the other?

The same with Windows XP Pro and Windows XP Home. Why test a software on both operating systems, when Windows XP Pro is the more sophisticatd OS system anyway.

What rationale do I need for the FDA to argue that it's a waste of time and resource to test the software on Windows ME and Windows 98 SE, ONE should suffice. And that goes for Windows XP Pro, test it on Win XP Pro, but no again on Windows XP Home.

If someone could explain the technical differences for me to understand the logic of testing it ON ALL OS systems or if someone could explain, why technically it wouldn't have to be done?

I'm still new to the medical field, I'm new to testing in the medical field and worst of all, I only have experience with manual testing, so it's double, three times, or more the work !

It would be great if we were to find a way to not make MORE work just because the FDA says so, I want substantial evidence so to speak!

Thanks for your input!
Regards, Renate Strina

 

[Marc]

Well, the bottom line is there are differences, however small, between the different OS versions. From the FDA's point of view testing on all OSs is important because life / health is involved. It's one thing for a game or something to crash a system (for example, I have *heard* that Windows ME is not very stable), it's another thing for software which is related to 'health' to crash a system.

Maybe one of the others can give you a better discussion on this - It's not my forte. But it doesn't surprise me that they want the software 'qualified' on all OSs it would be used on.

If it was me, I would wrote up a test procedure which addresses all the critical functions of the software and use that test on each OS. But again, this is not my forte so maybe one of the others can better help.

 

[Wes Bucey]

I want to clarify my post:

The system requirements are that the software (medical software) is supposed to run all the operating systems listed, but so far the programmers haven't written ANY software requirements for the operating systems. Of course they have written other software requirements specific to the software itself.

And my question now is, how and more specifically WHY I would have to test the software on Windows ME AND Windwos 98 SE, when the operating systems are so similar and one was built on the other?

The same with Windows XP Pro and Windows XP Home. Why test a software on both operating systems, when Windows XP Pro is the more sophisticatd OS system anyway.

What rationale do I need for the FDA to argue that it's a waste of time and resource to test the software on Windows ME and Windows 98 SE, ONE should suffice. And that goes for Windows XP Pro, test it on Win XP Pro, but no again on Windows XP Home.

If someone could explain the technical differences for me to understand the logic of testing it ON ALL OS systems or if someone could explain, why technically it wouldn't have to be done?

I'm still new to the medical field, I'm new to testing in the medical field and worst of all, I only have experience with manual testing, so it's double, three times, or more the work !

It would be great if we were to find a way to not make MORE work just because the FDA says so, I want substantial evidence so to speak!

Thanks for your input!
Regards, Renate Strina

It is interesting to come across this thread because it involves an issue many folks are unclear about - Configuration Management.

The entire point of Configuration Management is to assure all issues of compatibility between versions of a document or product are addressed and noted. It is an especially big deal for Electrical Engineers who hook their equipment into networks which all are connected and interact with each other.

In the present situation, FDA realizes that the spectrum of potential users of a software product ranges from backwater operations still using MS-DOS up to and including those using XP, Windows 2000, Unix, Linux, Mac OS-??, and so on. FDA says the burden of proving compatibility with all these systems should rest on the organization looking to introduce new software into the system rather than on the individual user to test and determine for compatibility with his individual operating system.

Consider if you had a Pacemaker which only worked on folks without diabetes (the equivalent of a software system which only works on Windows XP) - there wouldn't be much value in the marketplace for such a system, would there? In addition, there would be horrible liability problems for installations where someone missed a diagnosis of diabetes (a system running on Win 98.)

 

[lindal]

I want to clarify my post:

And my question now is, how and more specifically WHY I would have to test the software on Windows ME AND Windwos 98 SE, when the operating systems are so similar and one was built on the other?

Windows 98SE plug and play doesn't work as well as ME, so the user will need to browse to the setup file with 98SE and usually XP Home where as ME, 2k and XP pro will actually automatically load the software.

The same with Windows XP Pro and Windows XP Home. Why test a software on both operating systems, when Windows XP Pro is the more sophisticatd OS system anyway.

I don't remember the specifics, but XP Home has some pretty big quirks that XP Pro doesn't have.

What rationale do I need for the FDA to argue that it's a waste of time and resource to test the software on Windows ME and Windows 98 SE, ONE should suffice. And that goes for Windows XP Pro, test it on Win XP Pro, but no again on Windows XP Home.

If you are preloading the software on a dumb box that will only run on your selected OS with your software, you are installing it at the site, and validating it at the site you might be able to argue limited testing.

If someone could explain the technical differences for me to understand the logic of testing it ON ALL OS systems or if someone could explain, why technically it wouldn't have to be done?

I can't explain it technically, but when I was performing (not writing!) validations, the operating systems all acted a lot differently with the same software. I dreaded testing things on XP home and pro. If you don't want to do it, change the claims so that you aren't guaranteeing the software on systems you don't want to test it on. (How many customers use XP Home? ME had a pretty small marketshare also, and I think most companies that used 2k upgraded to XP Pro) Your marketing group should be able to give you a better idea of what OS the software will really be used on. Depending on what they say, you can limit the customer requirements and product spec, there by reducing your testing. Of course this may have already happened, in which case you're stuck

It would be great if we were to find a way to not make MORE work just because the FDA says so, I want substantial evidence so to speak!

I don't think the FDA has requirements as to the OS you test it on, but that you validate the software on the OS that you claim it works on.

Off-the-shelf operating systems need not be validated as a separate program. However, system-level validation testing of the application software should address all the operating system services used, including maximum loading conditions, file operations, handling of system error conditions, and memory constraints that may be applicable to the intended use of the application program.

If the operating system requirements are based on market research, I would do the testing because you want the software to work for every user. If not, change the requirements to accurately reflect the user's needs. There will be a much larger headache if you are constantly getting complaints because your software doesn't work on every customer's system, when you told them it would.


HTH,
L

 

[Al Rosen]

I want to clarify my post:

The system requirements are that the software (medical software) is supposed to run all the operating systems listed, but so far the programmers haven't written ANY software requirements for the operating systems. Of course they have written other software requirements specific to the software itself.

And my question now is, how and more specifically WHY I would have to test the software on Windows ME AND Windwos 98 SE, when the operating systems are so similar and one was built on the other?

The same with Windows XP Pro and Windows XP Home. Why test a software on both operating systems, when Windows XP Pro is the more sophisticatd OS system anyway.

What rationale do I need for the FDA to argue that it's a waste of time and resource to test the software on Windows ME and Windows 98 SE, ONE should suffice. And that goes for Windows XP Pro, test it on Win XP Pro, but no again on Windows XP Home.

If someone could explain the technical differences for me to understand the logic of testing it ON ALL OS systems or if someone could explain, why technically it wouldn't have to be done?

I'm still new to the medical field, I'm new to testing in the medical field and worst of all, I only have experience with manual testing, so it's double, three times, or more the work !

It would be great if we were to find a way to not make MORE work just because the FDA says so, I want substantial evidence so to speak!

Thanks for your input!
Regards, Renate Strina

Renate, I agree with Marc on this subject. Although there are minor differences, it is well known that software will operate on each OS differently and might have a bug related to one OS but not an other.

On the issue of requirements, the software needs to be both verified and validated. I find that many people do not fully understand the difference and interchange the terms.

Verification is the interim testing of the software code to assure the output meets the requirements for a phase in development or of a module of code.

Validation is the process of assuring that the user requirements and intended use for the device are consistently satisfied. This is part of the Design Validation of the finished device.

Based on what you have described, it appears that you are at the stage where you can verify your software, but cannot validate it until you have a finished device.

There are many factors that affect the amount of verification and validation including the intended use and class of the software/device. I know that you may not be able to discuss the details, but it is important information for making a decision. Since you should not reveal what might be proprietary details and haven't the experience in this area, I think you should seek the advice of a consultant who has experience with Medical Device Software and FDA requirements. In the mean time download General Principles of Software Validation; Final Guidance for Industry and FDA Staff and Design Control Guidance for Medical Device Manaufacturers. Both of these FDA Guidance Documents are attached in other threads that you may want to read through.

After reviewing these, you may find some answers or have some specific questions to post here.