Re: Regression for MTBF (Medical Device: Repairs vs Install Base) with Reliability Da
I know this sounds counterintuitive but to get the average pbar you do accumulate (or sum) the install base of all of the months. This yields a large number and you are counting the same instruments many times, BUT you need to think about the instruments as 'independent' opportunities each month...from a physics standpoint they aren't but by plotting the failure rate per month, you are treating them as if they are. and if you really think about it you are using the number of devices as opportunities and there is more to the opportunity than just the device. This is what we do for our instruments and the bottom line is that it works.
the better way is to use the number of uses in a month as the area of opportunity (which we are beginning to do) but this data is not easy to get. so many people use number of devices as a surrogate...