Cardiology & AI

Still With Us: Treating ST-Elevation Myocardial Infarction Remains an Urgent Focus. Finding Hope in the Rise of Telemedicine and AI

CLD talks with Sameer Mehta, MD, FACC, MBA, Chairman, Lumen Foundation, Miami, Florida.

CLD talks with Sameer Mehta, MD, FACC, MBA, Chairman, Lumen Foundation, Miami, Florida.

[Note to readers: Download the accompanying pdf of this article for extra information that includes a map of the STEMI hub-and-spoke networks in Colombia, and two abstracts presented at the European Society of Cardiology (ESC) 2019 meeting: "Base Camp to Mount Everest – Milestones in Advancing a Single-Lead for STEMI Detection" and "The Continued Proficiency of Artificial Intelligence for Interpreting EKG: Single Lead EKG for STEMI Culprit Lesion Localization".]

What is the status of ST-elevation myocardial infarction (STEMI) today?

In the United States, we have achieved so much. Heart attack is no longer the number-one killer because of the combined efforts of hospital administrators, the American Heart Association, American College of Cardiology, Society for Cardiovascular Angiography and Interventions, and of course, the interventional cardiologists. We have created such a fantastic system that it has helped save the young and even the elderly patients who used to die from heart attacks. Progress continues to be made, but beyond that, we need to take a hard look at the last decade, which I think has been a lost decade in STEMI interventions. We have not addressed the issue of delayed presentation, although patient awareness has improved in North America and in Europe. Patients are presenting early, but by and large, we have not addressed patient awareness systematically with a step-by-step educational approach.

The larger issue involves looking beyond our borders to examine what is happening in the rest of the world. The gains have been pathetic in major parts of the world, particularly in regions of Southeast Asia, the Middle East, and Africa, where a STEMI goes unrecognized, untreated, and with extremely high and unacceptable rates of morbidity and mortality. In some areas, patients do not even have access to streptokinase. Delayed presentation is endemic; infrastructure is lacking. In terms of the geographical distribution of cath labs, there is a complete maldistribution. They all seem to be located in big geographical centers. In India, for example, there is a tremendous lack of infrastructure nationwide, including cath labs, and yet in New Delhi there are 200 cath labs, probably more per square mile than anywhere else in the world. In most parts of the world, there is a complete lack of a systems of care approach. The care for women is absolutely pathetic. Women continue to have far worse outcomes. They are uniformly presenting late. Access to care for women having a heart attack is burdened by barriers that are personal, social, racial, cultural, financial, and religious. It’s a shame that in so many parts of the world, the care for our sisters and mothers is so awful.

How is telemedicine helping STEMI patients worldwide?

Over the last near-decade, our STEMI endeavors with the Lumen Foundation were focused on creating STEMI networks that were successfully constructed in 32 countries, with some notable examples being Puerto Rico, China, Malaysia, Thailand, and of course, Central and South America with our telemedicine endeavors. In the Lumen Foundation’s efforts to reduce the disparities of heart attack management in both developed and developing countries, the Latin telemedicine program has been a beacon. It has lit the path ahead. This is guidelines-based, population-based STEMI care coverage using the pillars of telemedicine. We began with the assumptions that telemedicine could hugely increase access; it could be an accurate way to diagnose, it could provide and guide comprehensive STEMI management, and of course it could be cost effective. The first three premises we were able to demonstrate in the first two years. The last premise, of telemedicine being cost effective, took almost a decade to demonstrate, but we have shown it to be the case. At the most recent European Society of Cardiology meeting, we had a dozen presentations about the various aspects of our telemedicine program1, focusing on how Latin America has provided a global template, along with ten presentations at TCT. We answered the following questions: (1) How to create a program; (2) How to create the telemedicine structure; (3) How to make it cost effective; (4) How the telemedicine program impacts the morbidity and mortality; and (5) How to increase stakeholder participation, including the involvement of national cardiology societies. At the moment, the telemedicine program is operational in four countries: Brazil, Colombia, Argentina, and Mexico, and it provides an umbrella of heart attack protection to 100 million patients. We have seen 850,000 patients access the telemedicine service and approximately 9200 STEMIs diagnosed and treated, with the majority being treated with primary percutaneous coronary intervention (PCI). For door-to-balloon times, 74% are now ≤51 minutes, which should make us extremely proud — but that would be a mistake. We can reduce door-to-balloon times, but the true total ischemic time remains increasingly high. There are times I wonder, truly, how genuine are all the results we present? Door-to-balloon time was an extremely meaningful metric to follow perhaps a decade ago. We need to look beyond door-to-balloon time and instead consider the true ischemic time.

Our telemedicine work has demonstrated an ability to take expert cardiologists and have them guide treatment within the telemedicine structure. We created a new metric known as TTD or “time to telemedicine diagnosis”, comprised of several characteristics. An electrocardiogram (ECG) is done at a “spoke” (we utilize a hub-and-spoke model), uploaded, downloaded at the telemedicine command center, the artifacts are removed, and the cardiologist is immediately notified. The cardiologist makes the diagnosis, the diagnosis is uploaded, and three immediate alerts are coordinated: to the hub, to the spoke, and to the interventional cardiologist on call. The time to telemedicine diagnosis from a center located hundreds of miles away from the originating patient is now 3.5 minutes. It takes us 3.5 minutes to make a diagnosis and to trigger the STEMI pathway. Such is the power of telemedicine today. It requires broadband technology placement at each site, cloud computing, and the incorporation of business intelligence tools at the command centers. But with some investment, these amazing results can be obtained. Some centers are run entirely within individual nations and others transcend national borders. We created 143 centers in Brazil that are monitored and guided within Brazil. The 108 centers in Columbia are guided within Columbia. However, the 82 centers of Mexico are guided through Columbia and the centers in Argentina are guided through Chile. In comparing results between the two approaches, we saw a uniformly decreased time to telemedicine diagnosis, affirming that telemedicine is a strategy without borders.

As you noted, total ischemic time remains a challenge, however. How is this being addressed?

The next area of focus for the Lumen Foundation is to reduce the delayed time to presentation. Recent publications at the European Society of Cardiology meeting from the U.K. and Ireland, demonstrated that even today in developed countries, we are facing a two-and-a-half hour delay between patients recognizing symptoms and seeking care. There is probably a very large number of acute MIs that do not have a typical presentation. Women classically will not have the typical symptoms and diabetics may not have any symptoms. The elderly can have confusing symptoms. Plus, when you are having chest discomfort at four in the morning, chances are that self-denial creeps in and people relate it to the spicy food they had for dinner or the anxiety of facing the boss in the morning. Even in the United States, there are data demonstrating patients’ reluctance to trigger the system because of the cost of the ambulance care, which could reach $600. In other parts of the world, there is no ambulance care. So at four in the morning, where do you go and get a test done? The present system is faulty and fraught with challenges.

I think the amazing 12-lead electrocardiogram (ECG) needs to be abandoned. A simple 12-lead ECG, although it is the only diagnostic tool required to determine a STEMI, is actually not so simple. It requires a complex machine that needs 12 leads, shaving chest hair, breasts can be an obstacle, and it requires a trained technician and expert cardiologist. Over the last several years, it became obvious to us that there must be a better way for a patient to self-diagnose, via the self-administration of an ECG. It was no great brain exercise to recognize that artificial intelligence was the way to go. With our partners at ITMS Chile and the brilliant cardiologist Dr. Roberto Botelho in Brazil, we started further exploring the role of artificial intelligence. The true value of artificial intelligence (AI) is in creating sophisticated and effective protocols, but the largest determinant of that is the quality of the data behind it. We realized that as a result of our telemedicine programs, the Lumen Foundation has access to the world’s biggest repository of classified, annotated ECGs with angiographic correlation. We have 19 million classified ECGs and half a million with angiographic correlates. We saw the possibility of creating an augmented AI methodology to read an electrocardiogram.

The first step was to see whether AI today can interpret an ECG and of course, the answer was yes. The next step was to find out if it could differentiate a STEMI from a non-STEMI. AI was easily able to do so. Then we asked, is it possible that a single lead — as an example, what you see on an Apple Watch, which is able to decipher rhythms — could be used to detect a STEMI? There were possibilities: the lead V2 has 97% accuracy in detecting a STEMI. We started exploring whether this single lead could be converted into a wearable sensor device, such as a patch, a ring, a watch, or a multimodality device that could be placed on other devices such as a Holter monitor or a blood pressure instrument. As we explored further, we realized that although lead V2 was having the highest success rate, it does require five electrodes. Going from 12 electrodes to 5 electrodes may be better, but is not the final destination. We decide to take lead 1, which has a 91.7% accuracy in diagnosing and detecting STEMI, and created a multimodality pathway. We now have prototypes on the Apple Watch and have a tool which needs to be taken to the patient. It can be self-administered and could be accessed by patients who are at high risk, those who had a previous percutaneous coronary intervention (PCI), or diabetics. It could be obtained by a patient and/or could be provided by payers to their higher risk population. The goal of this tool is to detect STEMI. The system can be made so sophisticated that with GPS positioning, it could even trigger an ambulance as well as activation of the prehospital STEMI pathway.

The real benefit is in developing countries where people are presenting much later and STEMI is not detected, because getting an ECG is fraught with barriers and is complicated. To use the detector in these parts of the world, all that is necessary is two conductive surfaces. We have even tried putting it on the steering wheel of a car and on a Holter machine. It could be placed where AEDs are located. It could be put in a 7-Eleven. You have discomfort, go and touch it, and you get an immediate notification as to whether it is a heart attack. Of course, in the United States, we are exploring ways to take the detector through the FDA approval process and compiling data, and it is my hope that one day such a system will become part of the guidelines; that STEMI detection probably only requires a single lead. The exercise of creating STEMI networks has now been expanded, with a focus on reducing the disparities between developed and developing countries, and on trying to find an AI solution for delayed presentation and for empowering a patient to detect their own STEMI.

What are the plans for testing the single-lead AI STEMI detector?

We have done initial validation with actual patients in Brazil, testing it on 25 patients presenting with an acute MI. A 12-lead ECG is done and correlated with the single-lead AI-guided Apple Watch. Thus far, it has shown perfect concordance with the coronary angiogram. At this point, we are convinced that it needs to be brought to much larger validation and are discussing further steps with the FDA.

You have set up telemedicine networks in Latin American countries, but in the U.S., we have seen small, rural hospitals shutting down as well. Even in the U.S., there is a need.

At one of our TCT presentations on telemedicine, two physicians, one from Alabama and another from Arizona, similarly expressed that there is a need for such systems in the United States. Another technology that is extremely promising for heart attack systems around the world is robotic intervention. The CorPath by Corindus (recently acquired by Siemens) offers a very elegant pathway that could be a game changer. An interventionalist with broadband connectivity in one location could potentially perform STEMI intervention in another location where there is no expert available. Our telemedicine system was created between spokes and hubs, with “spokes” located in remote areas where there is not even a cardiologist or any physician to read an ECG. Robotic intervention systems may permit us to one day perform a STEMI intervention at one of these spokes. This may be exactly the tool that is required to perform STEMI interventions and reduce barriers in parts of Africa, for example. Corindus actually looked at how long the lag phase is, and at TCT, data were presented that are very promising, showing that the lag phase has now moved from 400 milliseconds down to 50 milliseconds. This is just the beginning of this incredible, very exciting field and it will have huge applications in STEMI interventions. 

Dr. Sameer Mehta recently reached the milestone of 20 years in a STEMI-only practice. He says, “The best gain has been for my own self: for my own body, heart, and soul. Working such a challenging discipline has kept me healthy. It has kept me mentally and spiritually happy, having saved lives. It’s been a very meaningful career and I thank God for giving me the opportunity.”

As part of the Lumen Foundation, Dr. Mehta has helped create STEMI networks in 32 countries, including China, India, Myanmar, Cambodia, Sri Lanka, Pakistan, Bangladesh, Nepal, and Indonesia. He has written several STEMI textbooks for interventional cardiologists as well as the layperson, with his most recent being “Preparing for Your Heart Attack …and Surviving It”.

Sameer Mehta, MD, FACC, MBA, can be contacted at


Further Reading/Viewing

  1. Mehta S. Working to improve STEMI care around the globe. Cath Lab Digest. 2018 Sept; 26(9): 46-54. Available online at Accessed December 11, 2019.
  2. Mehta S, Acosta MI, Lopez C, et al. A primer on artificial intelligence (AI). Plus: is your job on the line? Cath Lab Digest. 2018 Dec; 26(12): 32-35. Available online at Accessed December 11, 2019.
  3. Mehta S, Vieira D, Torres MA, et al. Artificial intelligence: refining STEMI interventions. Cath Lab Digest. 2019 May; 27(5): 1-25. Available online at Accessed December 11, 2019.
  4. A list of Sameer Mehta and colleagues’ European Society of Cardiology 2019 presentations can be found at Accessed December 11, 2019.
  5. Dr. C. Michael Gibson Talks With Dr. Sameer Mehta and Francisco Fernandez About Wearables: The Future is Now. TCT 2019. Available online at Accessed December 11, 2019.