Tech in Cardiology
On a recent flight from San Francisco, I found myself sitting in a dreaded middle seat. To my left was a programmer typing way in Python, and to my right was an oncologist flipping through a slide set on chemotherapy trials. While this may sound like the beginning of a bad joke, I remember this moment because it got me thinking about the influence of tech on medicine. The purpose of my trip, by the way, was to interview for a fellowship position in cardiology, a specialty with arguably some of the most impressive tech.
Not to discount advances in medical devices (e.g. leadless pacemakers, bioprosthetic valves), the emergence of consumer-facing wearable devices is as trendy as ever. Google recently collaborated with AHA to build its fitness app (Google Fit), which uses algorithms to quantify physical activity in terms of “heart points.”1 The Apple Health app now incorporates EKG capabilities, allowing patients to record episodes of arrhythmias—something I have certainly witnessed in cardiology clinic.2
Big data is an increasingly prominent component of clinical research, and a number of joint ventures with medical and tech leaders have emerged. One Brave Idea3 is a research collaboration between AHA and Verily (Alphabet’s life sciences division) which uses genomics to study coronary artery disease. Meanwhile, Verily’s Project Baseline4 is a massive longitudinal observational study—a modern version of the Framingham Heart Study.
AI could eventually play a prominent role in medical diagnosis and decision-making. The Stanford Machine Learning Group5 has developed a neural network that outperforms cardiologists in diagnosing arrhythmias on EKG—a significant improvement on existing algorithms which are often unreliable. AI also carries vast potential in radiologic interpretation. Already, Veril is using machine learning to interpret retinal images not only to detect diabetic retinopathy and macular edema but also to extrapolate information about cardiovascular risk.6
Electronic medical records represent an obvious space for tech innovation. Fast Healthcare Interoperability Resources (FHIR) are making it easier to share health information across our disjointed EMR systems. Providers are now able to push health data directly to patients’ iPhones using Apple Health Records.7 One can only speculate whether we will see a legacy software giant compete directly in the EMR space.
Cardiology and the rest of medicine has long excelled at basic science and translational research, but digital tech is increasingly creeping in. We are in a tech zeitgeist, and this is good for both patients and providers.
Thirteen years ago in my first anatomy class of Med School, the instructor asked us to make sure our learning is “deep.”
“You need to memorize the names of every single tiny nerve and muscle, because they all will be on your exam. One day you could be surgeons and if you cut out the wrong structure, you can kill someone!” he asserted as we all stood there in fear.
Later in Med School, we were told that half of what we’re learning will be wrong by the time we practice. But one thing we were not told is that the way we learn and the entire premise of what makes a good doctor would also change. For example, “deep learning” itself now means something different to me and to most healthcare professionals. If you are reading this article so far, then you are likely to have seen the term floating around in medical journals.
Deep learning is a type of machine learning in which the computer uses multiple layers of processing to extract features from otherwise vague data input, such as an ECG or a slice of an MRI. Each layer uses outputs from the previous layer. Through deep learning, the computer simulates the neural network of the brain and is able to learn and make sense of abstraction.
Working at the Broad Institute of Harvard and MIT allowed me to recently be part of a team that uses deep learning to solve important problems in cardiology. Over several weeks, cardiologists, scientists, and machine learning experts worked in teams to train computers on deep learning models so that they understand data such as medications, ECGs, genomic architecture, and cardiac MRIs of tens to hundreds of thousands of people.
The insights I gained were incredible. Just like medical students perform better on their exam if they learn “deeper,” the longer you train a computer model, the more it learns and the better it performs in predicting – but it does so at much faster rates than any doctor could ever match. For example, in only two days, we trained a computer to read the ejection fraction from a cardiac MRI as good as a doctor would. Using the MRI, the computer could also predict with reasonable precision the presence of hypertension and coronary artery disease, without knowing anything else about the patient. The power of computer vision is beyond imagination. While it could take you a full day to read 100 ECGs, a well-trained deep learning model could read them in only few seconds. It could also identify patterns in the data that the human eye could not discern, which might or might not be biologically or clinically relevant.
As data availability and computing power continue to grow, we will be seeing more and more applications of deep learning in cardiology. While we do so, we should stay mindful that human supervision and our role as doctors in charge of our patient’s health is more important than ever. This requires us to understand how computers work and how those models are built through working with multidisciplinary teams. If we do this right, we can probably do less deep learning ourselves by delegating to computers, and gain a whole lot of extra time that we can invest in taking care of our patients.
This morning I strayed from my usual hangout in the basic sciences sessions to investigate the hottest new products in the Health Innovation Pavilion at the Health Tech Competition. In this event, 8 highly talented applicant companies were allotted 3 minutes to pitch their company or product, followed by 5 minutes of Q&A from a panel of venture capitalists and AHA VIPs. Contestants were scored on novelty, innovation, potential patient outcomes, ability to address patient and provider needs, and strategy to launch and sell, among other criteria.
A focus emerged early in the competition: data aggregation. We’re seeing start-up companies developing digital platforms that collect massive amounts of patient data and process it to improve cardiovascular health outcomes. The target consumer varied from individual to hospital system, and the aims and applications stretched from prevention to detection and diagnosis. I particularly enjoyed hearing about Seqster’s software that integrates health records, DNA and fitness data in one place, though I thought all the panelists in the “real-life Shark Tank” had interesting and educational pitches.
The judges’ deliberations illuminated potential advances and pitfalls facing the field, and I think we need to ask ourselves a few things as both scientists and consumers. How might we respond if an adverse event is detected, and what are the consequences if something is missed? How efficient and accurate are the technologies? Who owns the data?
The pitch competition today demonstrated that health technologies hold great potential. It’s clear that as our tools evolve to improve patient health, the direction and guidance provided by our congregated cardiovascular experts, like those at Session 2018, will be invaluable.
Annie Roessler is a PhD Candidate at Loyola University in Chicago, IL. Her research focuses on the neurobiology and molecular mechanisms of electrically-induced cardioprotection. She tweets @ThePilotStudy and blogs at flaskhalffull.com