Monday, March 4, 2024

When the conventional algorithm diagnoses the ECG as COMPLETELY NORMAL, but there is in fact OMI, what does the Queen of Hearts PM Cardio AI app say? (with 10 case examples)

I have often written about how an ECG interpreted as "normal" by a conventional algorithm may well be manifesting OMI, or even long QT or hyperkalemia.

I have collected 51 OMI cases that were diagnosed as completely normal by the conventional automated computer algorithm.

Shifa Karim and Gabe Keller helped with a project to assess all these ECGs with the Queen of Hearts.

Part of the result is back:

Of those 51 cases of OMI with a "completely normal" ECG (not even "nonspecific ST-T abnormalities, but completely normal!), 35 were diagnosed by the Queen of Hearts as OMI.

I wanted to show some of the cases here.

You can read all the details in the links if you want.

The Queen of Hearts PM Cardio App is now available in the European Union (CE approved) the App Store and on Google Play.  For Americans, you need to wait for the FDA.  But in the meantime:

YOU HAVE THE OPPORTUNITY TO GET EARLY ACCESS TO THE PM Cardio AI BOT!!  (THE PM CARDIO OMI AI APP)

If you want this bot to help you make the early diagnosis of OMI and save your patient and his/her myocardium, you can sign up to get an early beta version of the bot here.  It is not yet available, but this is your way to get on the list.

Or use the QR code:


10 Cases:

Case 1

Do you want to be interrupted to view what the computer calls normal or nonspecific ECGs? 2 cases at once!

RCA Occlusion (OMI), called normal by conventional Veritas algorithm.
There are inferior hyperacute T-waves, with reciprocal STD and TWI in aVL.

Algorithm: Veritas (on Mortara machines)


The Queen gets it right



Case 2

A middle-aged woman with chest pain and a "normal" ECG in triage.  The OMI was not seen and she arrested in triage.  

LAD Occlusion called Normal by the conventional algorithm
There are hyperacute T-waves in V3-V5.  LAD OMI.

GE Marquette 12 SL algorithm


The Queen gets it right



Case 3

Should Emergency Physicians be interrupted by ECGs that are read as "Normal" by the computer?

Inferolateral OMI diagnosed as "Normal" by the conventional algorithm
There is STE in lead III and aVF, with reciprocal STD in aVL
Subtle STE in V5 and V6.

Unknown algorithm


The Queen gets it right


Case 4

How unreliable are computer algorithms in the Diagnosis of STEMI?

Pain was resolving.  Diagnosed as Normal by the computer.  Troponin negative.   The patient's prehospital ECG showed that there was massive STEMI and these are hyperacute T-waves "on the way down" as they normalize.
This was called normal but has hyperacute T-waves in V2-V5. 
LAD OMI.

Unknown algorithm


The Queen gets it right


Case 5

Subtle Dynamic T-waves, Followed by LAD Occlusion and Arrest

These are Wellens' waves in V2-V4, which represent an LAD thrombus that is open (reperfused OMI) at this time, but could close at any time. The computer called it "normal." the physicians did not see it. 
He did re-occlude and arrest but was resuscitated.

Unknown algorithm


The Queen gets it right
The Queen has been taught the difference between ACTIVE occlusion and REPERFUSED.  So she knows that this is reperfused. 
But in Version 1 the output is only OMI vs. Not OMI, with confidence level.
(Version 1 does not report Active vs. Reperfused)


Case 6

An Elderly Male with "Indigestion"

This is diagnostic of LAD Occlusion with Terminal QRS distortion and hyperacute T-waves in V3.

Algorithm is either Glasgow or Marquette 12 SL


The Queen gets it right


Case 7

Chest pain relieved by Maalox and viscous lidocaineDiagnosed as Normal by the computer.  Troponin negative. Patient was being discharged with a diagnosis of GERD when he arrested.

These are hyperacute T-waves.

Unknown algorithm

The Queen gets it right


Case 8

Another Inadequate Paper Published on Triage ECGs, whose Conclusions Need Scrutiny.

Proximal LAD Occlusion with STE in I and aVL, and hyperacute T-waves in V2-V6.

Algorithm: Marquette 12 SL (GE)

The Queen gets it right


Case 9 (prehospital and ED ECGs).  



Prehospital ECG:
There are hyperacute T-waves in V3-V5.


The Queen gets it right


First ED ECG:
Hyperacute T-waves persist.  

Called normal again!

 Algorithm: Veritas (on Mortara machines)

The Queen gets it right



Case 10: 
There is minimal STE in lead III with some T-wave inversion, and reciprocal STD in aVL, also minimal but clearly real.  There is also STD in V2-V4.  This inferior-posterior OMI.

Called normal again!

 Algorithm: Veritas (on Mortara machines)


The Queen gets it Right


Get Queen of Hearts PM Cardio app or Telegram version of the Queen:

App

If you live in the EU, the Queen of Hearts is integrated into the entire PM Cardio app and is CE approved.  The full app has the whole range of ECG diagnoses.  You will get 5 free uses, then will need to pay  [If you live outside the EU, you can also get this if, during registration, you state that you live in the EU.]




Telegram

If you live in the U.S., it is not yet FDA approved and you will need to fill out this form to get access to the Queen as it is housed on Telegram (but without all the other features).  You can use it an unlimited amount.

Or use QR code to get the telegram version:






==================================
My Comment by KEN GRAUER, MD (3/4/2024):
==================================
In today's Blog post — Dr. Smith justifiably laments the inaccuracy of conventional ECG interpretation programs for recognizing acute OMI.
  • As we have so frequently documented on Dr. Smith's Blog for well over a decade now — all-too-many clinicians (including all-too-many cardiologists) — remain "stuck" in the STEMI Paradigm, as well as ignoring increasing data in support of cardiac cath-validated studies showing at least 30% of acute OMIs (Occlusion Myocardial Infarctions) are missed if one depends solely on millimeter-based criteria from the outdated STEMI Paradigm.

As a result — it is not in the least surprising that conventional ECG interpretation programs are inadequate for identifying acute OMIs
  • The above said — the fault is not in the computer — since computer programs merely do what they are programmed to do.

  • Conventional ECG interpretation programs (ie, virtually all computer interpretation programs prior to the QOH AI Bot app) have been programmed according to STEMI criteriawithout consideration of clinical context (ie, correlation of the presence and relative severity of symptoms with each of the serial ECGs that are done) — and without consideration of the additional ECG findings that we routinely emphasize on this ECG Blog that correlate with OMI (See Figure-1).

  • As a result — We should not expect conventional ECG interpretation programs to manifest even a reasonable degree of accuracy for recognizing acute OMI. And, as a result — We should not be using conventional ECG interpretation programs for this purpose.

  • Finally — Clinicians (including cardiologists) who in 2024 remain "stuck" in the STEMI Paradigm (and who continue to ignore other ECG findings shown to more reliably identify acute OMI, as noted in Figure-1— such clinicians will continue to delay (if not completely overlook) the need for prompt cardiac catheterization of acute OMIs that could (and should) benefit from prompt revascularization.

===========================

The clinical examples of OMIs missed by conventional ECG interpretation programs that Dr. Smith provides in his discussion today — confirms our need to ignore these conventional algorithms.
  • Optimal clinical ECG interpretation for identification of acute OMI can be attained by attention to the principles reviewed in the 90-minute Webinar by Drs. Smith, Meyers and Herman (in the December 5, 2023 post of Dr. Smith's ECG Blog) — many of which are summarized in Figure-1 — and, all of which are routinely discussed in cases presented on this ECG Blog.

  • The QOH (Queen OHearts) PMcardio OMI AI model offers proof that the fault of conventional ECG interpretation programs is not the fault of the computer. On the contrary — because QOH has been programmed through guidance by Drs. Smith and Meyers (with an ever expanding data base of more than 18,000 tracings with cath-finding validation) — QOH already manifests superior accuracy for identifying acute OMI compared to the use of standard STEMI critieria (Herman, Meyers, Smith et al — Eur Heart J Digital Health — November, 2023).

  • MY Thought: Optimal clinical ECG interpretation will probably be BEST attained by a combination of expertise from experienced clinicians with the already impressive, but continually improving accuracy of QOH. Some degree of clinician oversight will probably always be needed to ensure optimal performance of QOH — but the additional opinion QOH provides can clearly be of great assistance to clinicians as a teaching tool, and for increasing clinician confidence and accuracy for rapid identification of which patients need prompt cath.


Figure-1: ECG findings to look for when your patient with new-onset cardiac symptoms does not manifest STEMI-criteria ST elevation on ECG. 
For my clarifying Figure illustrating T-QRS-D (2nd bullet) — See My Comment at the bottom of the page in Dr. Smith’s November 14, 2019 post.


=========================
MY Editorial NOTE re Computer Interpretations:
As one who has followed the evolution of conventional ECG interpretation programs over recent decades — I have offered my thoughts in a number of prior posts in Dr. Smith's ECG Blog, regarding what these conventional ECG interpretation algorithms can and cannot do. For those interested — Check out My Comments at the bottom of the page in the following posts:


BOTTOM Line: 
I find the following true regardless of whether your availability is with a conventional ECG interpretation program or QOH:
  • Never look at what the computer says until after you have forced yourself to interpret the patient's ECG in conjunction with available clinical information. 
  • Only THEN — Look at what the computer said. Do you agree?
  • Realize that IF the program available to you is based on a conventional ECG algorithm — that these are not reliable programs for identification of acute OMI (ie, Far too many false positives and false negatives!).

  • IF you do have availability of QOH — You can feel reassured IF your clinical impression matches that of QOH.
  • IF on the other hand, your interpretation differs from that rendered by QOH — GO BACK to the drawing board, realizing that unless you can come up with a valid reason for why QOH may be mistaken (ie, artifact, your availability of serial or prior tracings) — that the QOH AI app is usually accurate.
  • Keep in mind that the QOH PMcardio OMI AI model will continue to get better as it gets programmed with thousands of additional ECGs (all with cath-validated data) — and as QOH becomes capable, as it will with time — of integrating prior and serial tracings, as well as historical information into its interpretations.
 



No comments:

Post a Comment

DEAR READER: I have loved receiving your comments, but I am no longer able to moderate them. Since the vast majority are SPAM, I need to moderate them all. Therefore, comments will rarely be published any more. So Sorry.