Why are we still interpreting cardiac rhythms?

EpiEMS said:
do we think it is OK that the algorithm misses nearly 20% of AMI? Is that an appropriate level of risk to take? I don't know - I would say no, but that's a gut response. On a system level, is that right? Do you need a clinician to interpret? Probably.
Click to expand...
That's interesting... and yet, we trust the algorithm/machine to identify vfib/vtach on 100% of the AEDs... I wonder what the failure rate is for those?
 
DrParasite said:
Why are they wrong? is not the critera for a stemi the same on the east coast as on the west coast ( as per https://ecgweekly.com/classic-stemi-criteria/[/URL ...{snip}... While I understand having human oversight (and agree it's needed), and not treating strictly off a machine, shouldn't the machines be able to objectively weird stuff with a heart rhythm?
Click to expand...
[/QUOTE]

Curious, have you heard of STEMI mimics?

“The mnemonic “ELEVATION”, can help you remember STEMI mimics


There are some rhythms that act and seem to be STEMI but aren’t. The algorithm the monitor uses may see the traditional STEMI rhythm but no other artifacts or artifacts that aren’t programmed into the algorithm. As such, a working knowledge of cardiac rhythms could prevent a disaster.

Using a mechanical analogy, I actually own a $6k troubleshooting computer from a high end tool company. Computer says misfire on cylinder 1, solution is new plugs, new wires and possibly a new coil. Replace all that, still have a cylinder 1 misfire. Why? Because it didn’t detect the carbon build up on the intake valve causing severely reduced compression thus not creating the environment in the cylinder The engine needs to properly fire cylinder 1. (Yes, this is an actual issue with the Chevy 5.3L).

No matter what anyone does in life, remember, machines can be fallible as its creator is fallible. If any medic relies solely on the monitor, regardless of how much it costs, I worry for the safety of their pt.
 
Knuckles said:
Curious, have you heard of STEMI mimics?
Click to expand...
no need to medicsplain, we are all mature adults here (although some more than others)

Knuckles said:
There are some rhythms that act and seem to be STEMI but aren’t. The algorithm the monitor uses may see the traditional STEMI rhythm but no other artifacts or artifacts that aren’t programmed into the algorithm. As such, a working knowledge of cardiac rhythms could prevent a disaster.
Click to expand...
so, if you are just looking for the ST elevation, I would agree with you. but why not just program in the stemi mimics into the algorithm? if it's objective criteria, I'm sure some programmer would be able to add it to the cardiac monitors.
 
DrParasite said:
no need to medicsplain, we are all mature adults here (although some more than others)


so, if you are just looking for the ST elevation, I would agree with you. but why not just program in the stemi mimics into the algorithm? if it's objective criteria, I'm sure some programmer would be able to add it to the cardiac monitors.
Click to expand...

Chaos theory. Doctors call medicine a “science” but there is still so much that can happen within the human body that, when multiple issues arise at once, can throw even the best doctor off.

Eat too much, get gas, but it presents as a cardiac issue. And then there’s the fact that everyone is different. Take me. Just about every textbook will tell you I have to be diabetic with high blood pressure and a cardiac issue as I have quite a few candles on my cake, enjoy Tim Horton’s and don’t belong to a gym. However, my BP is typically 110-120 systolic (114 at my last FF physical), my body processes sugars like nuts and my EKG at my last physical is normal sinus (by machine AND cardiologist interpretation). My PCPs constantly tell me my numbers can’t be right even though they took my BP, EKG, etc., but they are.

Algorithms cannot be programmed for the infinite number of variables, but they can guide us. It’s up to the medic to put the puzzle pieces together. And any EMS instructor worth their salt will teach to “treat the patient, not the numbers.”
 
Knuckles said:
Chaos theory. Doctors call medicine a “science” but there is still so much that can happen within the human body that, when multiple issues arise at once, can throw even the best doctor off.

Eat too much, get gas, but it presents as a cardiac issue. And then there’s the fact that everyone is different. Take me. Just about every textbook will tell you I have to be diabetic with high blood pressure and a cardiac issue as I have quite a few candles on my cake, enjoy Tim Horton’s and don’t belong to a gym. However, my BP is typically 110-120 systolic (114 at my last FF physical), my body processes sugars like nuts and my EKG at my last physical is normal sinus (by machine AND cardiologist interpretation). My PCPs constantly tell me my numbers can’t be right even though they took my BP, EKG, etc., but they are.

Algorithms cannot be programmed for the infinite number of variables, but they can guide us. It’s up to the medic to put the puzzle pieces together. And any EMS instructor worth their salt will teach to “treat the patient, not the numbers.”
Click to expand...
I don't get the part about your history and why it signals DM.
 
Knuckles said:
Algorithms cannot be programmed for the infinite number of variables, but they can guide us. It’s up to the medic to put the puzzle pieces together. And any EMS instructor worth their salt will teach to “treat the patient, not the numbers.”
Click to expand...
tenor.gif


No one is saying treat solely off the monitor. and I already agree with you that there are plenty of stemi mimics, but since you have already identified several of them, shouldn't you be able to incorporate them into the monitor's analysis?

Your example of gas being a cardiac issue is a great one... if someone has gas, are you activating the cath lab? or should you be using the monitor (which is an objective way to evaluate the heart electrical activity) to determine if it's a stemi or not? I mean if it's normal sinus, nice easy side, but if it's ST elevation (without chest pain), let's have the cardiologist double check and go to the lab?

I mean, we (EMS) have all these fancy and expensive pieces of equipment on the ambulance, shouldn't we use them to accurately identify what is going on with the patient? or would you consider that treating the numbers, not the patient?
 
mgr22 said:
I don't get the part about your history and why it signals DM.
Click to expand...
I work In an area that has a high turnover of brand new “doctors”, nurses and PAs. They finish school, come here for quick and easy residency and then bail. But the book says...” mentality has tainted patients and, once that med pro disappears, the pt gets a new doctor. Many of the new med pros go strictly by the book that fat=diabetic=hypertension=etc. Every newbie I’ve seen immediately send me for glucose tolerance tests, bloodwork aimed mainly to determine hyper/hypoglycemia and ignores previous testing because ”the book says...”

When they finish their testing for DM, hypertension, cardiac issues, etc. and come up empty, they scratch their heads and tell me the tests must be tainted. Putting full faith in a monitor to tell you exactly what’s wrong with your patient is the same, it’s programmed by the book and you end up treating numbers, not the patient.
 
Knuckles said:
I work In an area that has a high turnover of brand new “doctors”, nurses and PAs. They finish school, come here for quick and easy residency and then bail. But the book says...” mentality has tainted patients and, once that med pro disappears, the pt gets a new doctor. Many of the new med pros go strictly by the book that fat=diabetic=hypertension=etc. Every newbie I’ve seen immediately send me for glucose tolerance tests, bloodwork aimed mainly to determine hyper/hypoglycemia and ignores previous testing because ”the book says...”

When they finish their testing for DM, hypertension, cardiac issues, etc. and come up empty, they scratch their heads and tell me the tests must be tainted. Putting full faith in a monitor to tell you exactly what’s wrong with your patient is the same, it’s programmed by the book and you end up treating numbers, not the patient.
Click to expand...
Your experiences with doctors, books, and medicine are much different from mine.
 
DrParasite said:
I mean, we (EMS) have all these fancy and expensive pieces of equipment on the ambulance, shouldn't we use them to accurately identify what is going on with the patient? or would you consider that treating the numbers, not the patient?
Click to expand...

I’m not saying not to use the monitor, I’m saying there are too many things they would have to program in that the monitor will become worse at diagnosis. Learning the rhythms is part of “treating the patient” as, like I said, everyone is different. Being able to verify what the monitor says is a necessity. Going strictly on what the monitor tells you? That’s “treating the numbers, not the patient.” We need to know rhythms to, as you said, accurately identify what is going on with the patient. There are signs and symptoms monitors cannot see. Any chimp that can read can do what a monitor tells them. What if there’s more going on that the monitor doesn’t know and cannot be programmed for?
 
mgr22 said:
Your experiences with doctors, books, and medicine are much different from mine.
Click to expand...

I’m not trying to say it’s like that everywhere, but there’s 4 hospitals in my area and several medical groups where new doctors show up, do their time and disappear. My wife has lived with low K levels most of her life. Her body has negative side effects when they push the med and, because the new doctor was taught “this is the range the level must be at”, they push it. Even after her explaining the issue. I’ve personally been told by several new doctors that, since I’m overweight, I must have hypertension and borderline diabetic before even running tests. When the tests come back negative, they doubt the tests. Several others I know personally has had similar experiences. But again, I’m not saying this occurs everywhere, but old school doctors that actually practice medicine are getting harder to find near me.
 
Knuckles said:
I’m not trying to say it’s like that everywhere, but there’s 4 hospitals in my area and several medical groups where new doctors show up, do their time and disappear. My wife has lived with low K levels most of her life. Her body has negative side effects when they push the med and, because the new doctor was taught “this is the range the level must be at”, they push it. Even after her explaining the issue. I’ve personally been told by several new doctors that, since I’m overweight, I must have hypertension and borderline diabetic before even running tests. When the tests come back negative, they doubt the tests. Several others I know personally has had similar experiences. But again, I’m not saying this occurs everywhere, but old school doctors that actually practice medicine are getting harder to find near me.
Click to expand...
I can certainly understand how your personal experiences would shape your views. I sympathize with what you and your wife have been through. I hope you find doctors you're comfortable with.
 
DrParasite said:
That's interesting... and yet, we trust the algorithm/machine to identify vfib/vtach on 100% of the AEDs... I wonder what the failure rate is for those?
Click to expand...

Fair point - and not sure. That said, and I haven’t done a comprehensive review of literature, seems like they are quite good. For example: http://hqmeded-ecg.blogspot.com/2015/04/pulseless-ventricular-tachycardia-why.html?m=1 and even as of 20 years ago, they were very very good: https://pubmed.ncbi.nlm.nih.gov/11524645/
 
Didn't have a chance to read all the replies so I apologize if I'm mentioning something that has already been said:

My experience with Cardiac Monitor's is exclusive to LP12 and LP15 but here's my experience: They are pretty darn good at the vertical access, but not so good at the horizontal axis. I'm more likely to trust it's diagnostics of STE than I am of rhythm.

Of interest years ago - 2007 or so, the service I worked for participated in a study by Tufts Medical Center regarding the use of a GIK (Glucose, Insulin, Potassium) in the setting of ACS. The idea was that an early administration of the "Metabolic Myocardial Enhancement" preparation could slow the progression of ACS. The study was double blind, placebo controlled but was reliant on ECG findings. Our LPs were programmed to interpret the percentage of likelihood that the patient was experiencing cardiac ischemia by the use of an acute cardiac ischemia time insensitive predictive instrument (ACI-TIPI). So basically, when your 12 lead would print out it would look pretty much like it always did, with one exception, it included a percentage of likelihood that the 12 lead demonstrated ischemic abnormalities. This aided in paramedic's ability to determine whether or not the patient met inclusion criteria. What was amazing to me was that I would have a patient with an ACS like presentation, obtain my 12 lead but refuse to look at the percentage without first performing my own assessment. The number of times I looked at what appeared to be a completely normal 12 ECG that also had a high percentage indicator was shocking (totally possible that my ECG skills weren't so good back then but I don't remember, ha!). Several times, subsequent 12 leads demonstrated clear criteria for STEMI. I always wondered, "what was the computer seeing that I missed". Some studies of the use of ACI-TIPI conclude that this can be helpful in ECG diagnosis.

For me, AI in ECG interpretation is helpful or supplementary at best but is no exception for clinical judgement. There is also the garden variety of patient's experiencing NSTEMI that may not demonstrate significant ischemic changes on 12 lead. Robots and AI could be invented to perform all sorts of tasks, but I'll end with one of my favorite quotes - "Scientists were so preoccupied with whether or not they could, they didn't stop to think if they should" - Ian Malcom/Jeff Goldblum Jurassic Park.
 
Dr Parasite: Yes we allow the AED to make the decision on V-Tach and V-Fib, but we don't allow them to make the decision to shock the patient we still have to push the shock button.
Which I reminded my medical director of during a HCP CPR class about 15 years ago when he told me they were fool proof, and he allowed me to put a regular AED on him, not the trainer; and then shake the pads that were on his chest: when the AED analyzed and announced "shock advised, charging" he sat up, yelling NO when I announced Clear, shocking, while reaching for the shock button, stating "the AED is correct you 'Must be in V-Fib". I tried that again about 3 years ago with a co-worker and got the same results.

On LP-15's and Zoll's we get the same results on dirt roads (and we have 100's of miles of dirt roads in both my FT and PT jobs coverage areas) when they are on patients, they will alarm and tell us that the patient is in a bad rhythm, but I try hard to resist the urge to de-fibrillate the patient that is awake and talking to me, even though he may be clinically dead and his body just hasn't caught up to the monitor.

So I fall back on the teaching from paramedic (and EMT-Basic) School from 17 and 25 years ago: Treat the patient, not the monitor.
 
Plus as a paramedic co-worker found from 20 years ago that tried to get his new EMT partner fired because she refused to start CPR on a patient because the rhythm strip he printed out looked so much like a bad rhythm, until it was pointed out that it was upside down. She refused because the patient was talking. Treat the patient
 
I feel like "treat the patient, not the monitor" is an abused aphorism we use. It makes sense, but from what I've seen, this is applied incorrectly more often than not. Of course we should be questioning the AED, monitor, or whatever, when it contradicts our assessment, or question our assessment. That being said, we are not using these tools for fun. They serve a purpose beyond being required by our companies or protocols. As much as we'd like to think out assessment is good enough for identifying STEMI, it is not. Quite often, we'll get cool, pale, diaphoretic patients with epigastric patient, elderly, nauseous, and you think this is the big one. GERD. You get patients feel anxious, nothing else really, STEMI. Some rhythms you can probably guess eg irregular pulse is probably atrial fibrillation, otherwise probably a sinus rhythm. You guys are intentionally creating artifacts or there is clearly something wrong, maybe loose electrodes, and these machines can be fooled. At the same time, we can be fooled too. Use the machines to augment your assessment and provide the best care for the patient. @JPINFV once wrote about this aphorism, treat the patient and not the monitor. If there is a contradiction, question yourself and the machine. Find out why. Awake and talking patient? Probably don't need CPR. Probably have time before making a decision on giving a drug or shocking. Is there artifact? Did you miss something? If the machine got a blood pressure different than you, did you hear it right? Are you certain about your blood pressure or heart rate you got?
 
So 1. I'm late to the conversation and 2. I'm going to speak out of ignorance of being a EMT-B, and a lightly 'seasoned' one at that.
As for Computers, AI, and ML thats more my language. After reading the thread I do agree with a fair bit of what @EpiEMS states. But, the original merit shouldn't be lost, AI is never *meant* to be like in the movies where robots control themselves. One major use is to catch patterns humans can't (lets say in binary for computers, human eyes can't look at X million bits of "1's and 0's" and assert "thats random"), but large data sets w/ sufficient AI/ML can.

It seems to me that for this particular use case it shouldn't be "AI can determine cardiac rhythms", but it should be confirming "we got that rhythm right". The "tech" should also be able to interpret things that may alter any diagnosis and alert on it. This quickly turns into an ethics discussion very quickly though. I can assert as far as cyber goes, if AI (lets say on a dynamic/stateful firewall) detects a large pattern associated w/ a specific nefarious actor it instantly takes action before warning for user intervention to stop.

As a personal observation (sorry Epi, no statistics here) there seems to be a much shorter tolerance for computers to fail vs a human. Simple example of: Computer reads rhythm wrong vs *insert medical provider here* reads rhythm wrong.

This Mayo clinlic article seems to be fairly representative of the capabilities IMO
 
Because id Like to think that i was trained to treat my pt, and not the monitor. The interpretation of a 12 lead (computer) is a TOOL that i read as well and then verify. We've ALL had artifact that reads as a stemi and We can easily say "nope, thats wrong, do it again"

and Maybe computer reads STEMI, and im not 100%! Thats what ER consults are for.
 
MMiz said:
Whenever I see a cardiac strip posted online, in this community or elsewhere, I note how different the interpretations are by so many qualified people.

With AI, deep learning models, and so much technology, why aren't we better leveraging technology to assess cardiac rhythms? Why in 2021 are paramedics still interpreting cardiac rhythms?
Click to expand...
My first thought when I read this is, there is no such thing as AI, just machine learning. I have a masters in IT (Dont ask, job just sucks) and deep learning is just use of a wider group of sampling for the algorithm. For a machine to interpret a rhythm is like asking who John Smith is... too many possibilities and without human talent (read arts) it won't be possible to accurately use for a long time. Think Tesla and "it'll be ready in ten years" every ten years.
Sorry for the rant, just not a fan of the term AI.
 
NPO said:
Nor do we have the ability to check it in the field (usually).
Click to expand...
Although that be changing with some of the new POC lab units like iStat (Not an endorsement or fan)
 
You must log in or register to reply here.
Back
Top