Editor’s note: Ethan Weiss, a cardiologist at UCSF, reflects on a recent patient who may well be a forerunner of many more patients in the near future.
I saw a man in his early 40’s. He’s super healthy. He had a Ziopatch as part of a general medical workup (not really important why). He was having some mild palpitations that were not terribly symptomatic.
The Zio showed a few symptomatic extra systoles and a few runs of what the computer called “supraventricular tachycardia”.
The cardiologist over-reading the study said: “Agree except at least 2 of the runs of SVT were actually atrial fib”
Here’s an example:
The patient saw a cardiologist at another hospital who thought it was AF and prescribed aspirin. The patient got another opinion at another hospital and the doctor there said it was not AF. The patient came to me for a 3rd opinion.
I think we are going to be seeing a lot of this. (The Apple Watch and devices like the AliveCor KardiaBand will add to the volume.)
Frankly I’m not sure what it is. It’s probably not AF, but I’m also not sure it matters.
The scientist in me wonders if this is what early lone AF might look like. The doctor in me feels like we have done this patient a disservice. The economist in me wants to cry.
—by Ethan Weiss
And the device companies are sauntering to the bank…
Looks like Wolf Parkinson White
That is a Flutter probably 3 to 1
I have been using the alivecor sensor on my patients for the last 4 years and can happily say it has cured a few hypochondriacs of their palpitations and diagnosed a few with Sustained AF , PVCs and transient heartblocks.
Dr N Basir
Cardiologist in Karachi
Pakistan
Perhaps eventually this will change the medical view of what is “normal”.
Though what exactly doctors mean by “normal” puzzles me. I once asked a doctor “do you mean ‘usual’ or ‘desirable’?” His jaw dropped – he’d obviously never asked himself.
IF there is a reasonable indication to find out more than the few seconds of office ECG can reveal about rhythm, ZIO is far better than cumbersome Holter. Two comfortable weeks of continuous recording vs 24 awkward hours.
Given that, it is only a machine and its output must be interpreted by an experienced cardiologist/electrophysiologist for it to be of any real worth.
Or perhaps it is only a machine and its output must be interpreted by a suitable trained robot for it to be of any real worth.
By which I mean this may well be a job well suited to AI.
The people at ZIO tell me that the data is reviewed by their “algorithm”, producing a preliminary report. Then clinical technicians review that, producing the final report.
AI, yes. But human techs are the final ZIO arbiters.
Then, of course, your own doc.
But what if experience were to teach that AI did a better job, on average, than your own doc?
The experience so far (past 40 years) with computer interp of ECGs has been a failure. Despite multiple generations of software it has yet to gain on the docs.
It’s difficult to tell what this is. It would be hard even on a 12 lead, let alone this type of recording. The RRs look pretty regular. Regularity however is difficult to assess at that rate with with such a brief run. There is electrical alternans which can be seen in some SVTs. If this is pulmonary vein tachycardia it could be a precursor to a fib. If you are working up a cryptogenic stroke this type of testing is legit and the finding might be clinically important. In a healthy person, more than anything else it leads to psychological harm for the patient and economic harm due to the downstream testing and referrals.