最近聽了EMRA*Cast 11月1日這一集，兩位Maimonides Medical Center的PGY2住院醫師Mackenzie Andrews和Tess，訪問Amal Mattu。節目標題叫〈Fear and Electrocardiography〉，從住院醫師「怕自己懂得不夠多」聊起，一路談到怎麼練心電圖、ECG Weekly為什麼要收費、AI判讀心電圖到底能不能信。

這集裡Mattu講了一句：同一個錯，他大概要犯三次才會記住。1

每一張都照同一套流程看

Mackenzie問了一個很實際的問題：寫case題目的時候錯很多題，告訴自己下次就知道了，結果下次遇到還是不會，怎麼辦？2

他說看一次沒記住很正常。每多看一次，就多記住一點。3

他自己出錯，多半是在趕時間、跳過逐步判讀流程的時候。他用的流程就是Rate、Rhythm、Axis、Interval，再來看Ischemia和Infarction。第一次學心電圖到現在大約35年，他還是每一張都照走。練得夠多之後，走完這套流程可以非常快，圖形辨識的能力也會慢慢長出來，但前提是看夠多的病例。4

ECG的morphology沒辦法讓你在看過少數非常typical的案例後，就能抓到每一次的變化。你一定要看到足夠多的normal morphology，才能真正確定自己學到了。也就是說，要看到更多的normal和abnormal ECG morphology，才能在心裡辨識出這兩種之間的差別。

單靠看abnormal的ECG morphology，很難期許自己在未來的職場中，只要看到心電圖就能馬上辨識出來，這是很困難的。

我在這邊教學弟妹看心電圖，也是這樣講：不同的老師會有不同的approach，但最好的方式，就是你固定學了一套完整的方法之後，就一直持續地使用。

要變強，他的建議只有一個：大量看病例和教學檔案。住院醫師可以把班上遇到的有趣心電圖留下來，做成病例檔案，每週一起討論一次。他自己當年是把書店買得到的心電圖教學檔案一份一份做完，那時候這些檔案全是心臟科醫師寫的，急診醫學還沒有人編。5這讓我想起在北榮急診的時候，我們每週都會進行病例討論，看看大家累積收集的case，還蠻有趣。

遮住電腦判讀，先自己下判斷

Mattu所在的University of Maryland，心電圖是主治醫師sign，不是三年級住院醫師sign。但他們會刻意做一件事：把電腦判讀遮起來，拿給資深住院醫師看。6

他會先講清楚：我拿這張給你看，不代表它有問題，可能完全正常，你就看一下、下個判斷。住院醫師說完自己的判讀，再去看電腦的結果，兩個人花30秒聊一下漏了什麼、看到了什麼。7

他的重點是「每次都要對判讀做出承諾」。只看電腦判讀的被動學習，跟自己先下判斷再對答案，學到的差很多。8

如果一張ECG出來還沒判讀就直接先看答案，之後再回來看心電圖，會有盲點。這個盲點就是你會focus在剛剛看完答案的那個地方。

所以，如果真的要考自己的時候，最好一開始先把答案遮住，試著看看自己能列出幾個differential diagnosis。最後再去看答案，看看兩者的差別在哪裡，進一步檢討為什麼對方看的重點是這個，而我看的重點是那個。我覺得這樣比較容易有進步的空間。

住院醫師的心電圖課怎麼排

Mattu的課程設計很樸素：

Table 1｜Mattu的住院醫師心電圖課程安排
時間做法
配對之後、開始訓練之前寄信請新進住院醫師先把基本判讀學起來，上工後不再從怎麼算心率教起
每個月心電圖研討會，用不帶威脅的方式在會議室裡輪流討論；實習醫師答基本題，二年級答難一點的，最難的留給三年級
每年年底心電圖競賽，選出住院醫師心電圖冠軍、頒獎牌

年底競賽對他來說其實是一張成績單。看得出全科哪裡弱：低年級分不清Mobitz I和Mobitz II，就回頭加強；資深住院醫師Brugada判不準，就補syncope相關的心電圖。9

他用學外語來比喻：小孩學語言不是先學文法，是一直聽，聽久了東西就留下來。心電圖也一樣，三到四年的住院醫師訓練一直看，細節自然會累積。不要期待六個月或一年就變專家。10

至於一年要看幾張，他說他不知道有人訂出過一個好的數字。他記得多年前有文獻比較心臟科研究員和急診醫師看過的心電圖數量，兩邊差不多，但沒有人驗證過某個特定數字。11

這讓我想到之前我們的急診老師說，在醫院裡最會看心電圖的兩類人，第一個是心臟科醫生，第二個是急診科醫師。心臟科的fellow大概一年會看三、四千張心電圖。想想看，我們如果一個月上15班，每一個班大概看30個病人，其中有10個會做心電圖，就換算成一個月約150張，一年大概1800張左右。## ECG Weekly為什麼要收費

ECG Weekly起源於2011年9月。一開始只是住院醫師想要每週一次的心電圖研討會，但每月一次的研討會時間不夠。他試過請大家在週三研討會前提早半小時到，第一週來一小群，之後一週比一週少。12

後來他改成貼一個有趣的心電圖病例，配一小段影片講解，寄給有興趣的住院醫師。其他住院醫師、主治醫師、校友陸續要求加入寄送名單，有人建議放上YouTube，就這樣變成ECG Weekly。13

結果這些放在YouTube、公開免費的影片，開始被醫療糾紛的原告律師拿到法庭和證詞程序裡用。律師的說法是：影片就在網路上，Mattu知道，你為什麼不知道？14

所以2015年1月，他們把能下架的YouTube影片全部下架，另外架了ECG Weekly網站，改成訂閱制，價格定在大約每週一杯咖啡。只要有訂閱門檻，就不算公開資料，律師也就拿不到。Mattu說現在訂閱人數已經超過25,000人。15原來ECG Weekly一開始是免費的，後來是因為有律師把內容下載去對付急診科醫生，才改成訂閱制，我終於了解了。

分享missed diagnosis，也要敢發問

Mattu說：從自己的錯誤學習很好，從別人的錯誤學習更好，而醫學界這件事做得不夠。16

他拿航空業比較。航空業對任何失誤都非常公開，結果每個人都學到，航空業也成為世界上最安全的行業之一。醫學界因為自尊，大家不想承認自己漏了什麼。他希望每個住院醫師訓練計畫都有一個完全沒有威脅感的環境，讓人可以站出來說「我漏掉了這個」，就像M&M討論會那樣，把心電圖的missed diagnosis也拿出來討論。17

發問也一樣。他在研討會上常說：如果你有一個很想問的問題卻沒有問，在某個地方就有一位病人死掉。你有疑問的時候，通常還有三、四、五個人有同樣的疑問，只是不敢問。他擔任住院醫師訓練計畫的領導者30年，從沒看過哪個愛發問的住院醫師被同儕看不起，大家反而會把那個人看成領導者。18

主持人Tess也分享了自己的經驗：她在博士班第一週上課舉手發問，被知名教授當著全班說「給那些比較慢的人，答案是這樣」，下課後卻有五個同學跑來說他們也完全聽不懂。她說她喜歡急診醫學的一點，就是這裡比她待過的其他領域更鼓勵發問，因為這個領域實在太廣了。19

急診醫學科要看的東西確實太廣了，外科、內科、小兒科、婦產科、五官科，什麼科都要涉及。各科別的病人有任何不舒服都會來急診，所以我們必須了解的領域實在太多，也因此我們有非常多不知道的地方。

如果不知道也不問也不去查，就只會限縮在你僅有的一小點點知識裡而已。

AI判讀心電圖，Mattu怎麼看

Tess問：AI判讀STEMI有沒有比人可靠？十年後技術可能完美，學習者會受到什麼影響？20

Table 2｜Mattu怎麼看AI判讀心電圖
情境Mattu的判斷21
只看「是不是STEMI」或「是不是OMI」AI很強，可能比一般急診醫師好一點
加上病人的臨床表現熟悉心電圖的醫師在床邊問病史、看病人，大概跟AI一樣好，甚至更好
有疑慮時醫師可以重做一張心電圖看serial changes，AI不會這樣做
心律不整他說沒看過可靠的AI研究或程式
Brugada、hypertrophic cardiomyopathy這類syncope相關的心電圖他說沒看過專門針對這些的AI程式
檢傷也許快到可以在檢傷時用AI判斷STEMI與否，但病人終究要有人去看

他特別提到責任歸屬。AI公司不會為判讀錯誤負任何責任。他協助過一件醫療糾紛，醫師漏看ST上升，AI也判斷沒有STEMI，病人不久後心跳停止死亡。律師去找AI公司幫忙，對方回覆：我們不為任何臨床決策負責。22

所以他把AI看成：有一個很懂心電圖的人幫你看一眼，但他不會把名字sign在病歷上。23

他也舉了醫院的例子：中風病人的head CT由AI先判讀，看到出血就發簡訊給大家，但至少有三分之一到一半的通知最後證實是錯的。24

Tess補了另一個角度：心電圖不只是判斷STEMI、Wellens的工具，也是在理解心臟電訊號怎麼運作。如果有一天主治醫師從來不需要自己判讀心電圖，對整個專科是一種傷害。25

Mattu的結論很直接：AI在心電圖上還有太多地方沒碰到，要很久之後才輪得到討論急診醫師還需不需要學心電圖。現在，絕對需要。26

恐懼是好事

回到開頭Mackenzie的焦慮：已經完成三分之一的訓練，卻覺得自己懂得不夠多。27

Mattu的回答是：你永遠不會懂得夠多，事情也一直在變。十幾年前他對LVAD一無所知，二十年前對超音波一無所知，都是後來才學。但對「懂得不夠多」的恐懼是好事，它會一直推著你往前。28

他自己走上心電圖這條路，也是從恐懼開始。醫學生時被罵過沒有幫躁動的病人做心電圖，從那之後他只是想學會看心電圖，不要傷害病人。這不像學騎腳踏車，沒學會還是能過日子；急診醫學裡有些東西沒學會，就會有人死。29

他接著說了一句我很喜歡的話：如果你已經決定投入這個領域，就等於承諾要學會讓病人活下去所需要的東西。30

我自己在急診醫學的訓練過程當中，也很常會出現這種FOMO的狀況，會覺得自己好像缺少了什麼，然後就想要去鑽研，希望能夠更幫助別人。如果少了些什麼，是不是別人就有可能因為某些原因而受到影響。

我的這種FOMO是fear of missing out。這裡的missing out指的是病人死去，因為太害怕病人死去，所以更想要讓自己的能力變得更強。

學習重點

  1. 每一張心電圖都照Rate、Rhythm、Axis、Interval、Ischemia的順序看。Mattu自己的漏看，多半發生在趕時間跳步驟的時候。
  2. Amal Mattu錯三次才記住是常態，所以大家不要太苛責自己，miss掉一次、miss掉兩次、miss掉三次都沒關係，不會表示自己太笨。ECG的recognition就是要多看、多接觸、多熟悉，這樣才會越來越強；住院醫師可以自己建病例檔案，每週一起討論。
  3. 遮住電腦判讀，先自己下判斷再對答案。主動判讀比被動看電腦結果學到的多。
  4. 把心電圖的missed diagnosis拿出來討論，從別人的錯誤學習。
  5. AI判讀「是不是STEMI／OMI」可能已經很強，但不看臨床表現、不會重做心電圖、不為錯誤負責；心律不整和syncope相關的心電圖，Mattu認為AI還差很遠。

參考資料

內文每個上標數字，滑過去（手機點一下）就看得到出處和原文。整理起來是這1份：

  • Podcast
    • EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026年11月1日（Apple Podcast）

  1. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："I'm going to make a mistake three times before I'll remember it." ↩︎

  2. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01（Mackenzie Andrews提問）。 原文："Because what I struggle with is, when I'm going through these case files, I will get many of them wrong. And I will say, oh, well, now I know for next time. And then the next time I encounter it, it just doesn't stick." ↩︎

  3. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："So if you don't remember it after one time, that's probably not too bad. Every time you see something, a little bit more is going to stick, number one." ↩︎

  4. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："I think I make mistakes if I rush through things and I forget to go through this stepwise approach. … And that's the same process that I try to go through now, 35 years later or whatever it was since I first learned about EKGs." ↩︎

  5. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："And maybe if everybody in the residency got together and just put aside interesting cases, interesting EKGs, created a case file, and every week or so just went through them as a group, I think there's a tremendous amount of learning that you can get. … And so I found some authors that had put together teaching files, and all of them were in cardiology. Nobody in emergency medicine at the time had put together teaching files." ↩︎

  6. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："But we do make an effort to cover up the computer interpretation and show those EKGs to the senior resident." ↩︎

  7. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："It might be totally normal, but I just want you to look at it and make a decision about whatever you think is going on. … So if they're making just a little bit of effort to come up with their own interpretation every time, and then you spend an additional 30 seconds talking about the EKG with the attending, then I think that really adds up to a lot." ↩︎

  8. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："And I think if you're committed to doing that, and therefore the resident has to commit to interpretation every time, they really learn a lot more than the passive kind of learning that people get if they're just looking at the interpretation each time." ↩︎

  9. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："So maybe, you know, I'm surprised the junior residents are not getting Mobitz I versus Mobitz II right. … Or maybe some of the senior residents are not nailing the Brugada or something, you know, and that tells me that we need to go back to some of the syncope EKGs." ↩︎

  10. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："And if you think about the analogy, it's kind of like learning a foreign language. When you learn a foreign language as a kid, nobody's teaching you the specifics of grammar, but you just hear the language over and over and then gradually more and more and more starts to stick." ↩︎

  11. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："Yeah, I don't know that anyone's ever come up with a good number. I recall that there is a publication that came out quite a few years ago that talked about three or four thousand EKGs that fellows, cardiology fellows, look at." ↩︎

  12. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："Well, there's not enough time in our monthly conference to do an hour of EKGs every week. … And just a total aside, I started that in September 2011." ↩︎

  13. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："And then a suggestion came to put it on YouTube. And so we started posting this on YouTube, and it eventually went over to become ECG Weekly." ↩︎

  14. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："used by malpractice attorneys in court and in depositions because they're free." ↩︎

  15. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。訂閱人數為Mattu在節目中的說法。 原文："So in January 2015, we took down all the YouTube videos that we could and we created a separate website called ECG Weekly and made a subscription fee of like 25 cents per week. We wanted to make it like a cup of coffee per week, really cheap. … So it started out with just a few residents and now there's over 25,000 subscribers to that." ↩︎

  16. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："And I think I mentioned it at the conference that it's good to learn from your own mistakes, but it's even better to learn from other people's mistakes." ↩︎

  17. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："And the result is that everybody learns, and the airline industry has really become one of the safest in the world because people are willing to share their mistakes." ↩︎

  18. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："I also tell people in conferences that if you have a question that you really want to ask, but you don't ask, somewhere out there, a patient dies. … I've never seen a resident in 30 years who asked a lot of questions who was looked down upon by the other residents because he or she was asking a lot of questions." ↩︎

  19. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01（主持人Tess）。 原文："And to the entire class, he's like, for those of you that are slow, this is the answer. … I can't speak to surgery as much, but we are very encouraged to ask questions because it's such a broad field." ↩︎

  20. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01（主持人Tess提問）。 原文："And can AI call a STEMI more reliably than a person and some data to support that? And then also the contrast of what does this do to learners if we're approaching an era where we are so reliant on that EKG reading, which now is terrible, but in 10 years might be technically perfect." ↩︎

  21. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。表中為Mattu的個人看法。 原文："I think from the standpoint of is it a STEMI or not STEMI, the AI is really good. … It probably is a bit better than the average emergency physician, but what AI doesn't take into account is the patient's presentation, really. … And the other thing that you can do when in doubt is just get a repeat EKG and look for serial changes. … I haven't seen any good AI studies out there on arrhythmias at all. … I haven't seen AI programs that are really focused on syncope things like Brugada, hypertrophic cardiomyopathy, and so on." ↩︎

  22. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："AI companies are not going to take any liability for any mistakes that end up occurring." ↩︎

  23. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："So I think of AI as having somebody who knows a lot about EKGs look at it as a favor to you, but they're not going to put their name on the chart." ↩︎

  24. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："And it seems like at least a third to a half the time, we'll get a message from AI, which turns out to be incorrect." ↩︎

  25. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01（主持人Tess）。 原文："But there's also what the EKG is telling you about how the electrical signals in the heart work. … And if at a certain point we reach a point where AI is the gold standard and we have attendings who have never had to interpret an EKG, I think that does a disservice to the field." ↩︎

  26. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："But again, I think that there's so much that AI hasn't touched yet with electrocardiography that I think it's going to be plenty of time before we even get to seriously debating whether emergency physicians need to keep learning EKGs or not. Right now, there's absolutely no question." ↩︎

  27. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01（Mackenzie Andrews）。 原文："We're in a three-year program. … I'm supposed to be 33% of the way done with all the things I ever have to know as an emergency medicine doctor, and I feel inadequate. I just don't think I've learned enough." ↩︎

  28. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："You're never going to know everything, and things are constantly changing. … You know, 10, 15 years ago, I didn't know anything about LVADs and I had to learn about that. You know, 20 years ago, I didn't know anything about ultrasound. … But the fear that you have about not knowing enough is really good, because that's going to keep you pushing yourself." ↩︎

  29. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："I just wanted to learn about EKGs so I wouldn't hurt people. … Well, fear is good, especially because if you don't learn something, then people die." ↩︎

  30. EMRA*Cast〈Fear and Electrocardiography with Dr. Amal Mattu〉，2026-11-01。 原文："And if you've made a commitment to do this field, then you've made a commitment to learn what's necessary to keep people going." ↩︎

◆ ◆ ◆