Each day, 1,500 people in the U.S. are diagnosed with heart failure. And yet, despite decades of medical advancements, assessing cardiac function remains a time-consuming undertaking.
Until, potentially, now.
The results from the 2016 Data Science Bowl could fundamentally transform one of the most important, time-intensive heart assessment procedures. It’s the “gold standard” of heart imaging: the cardiac MRI. Increasingly powerful medical imaging technology generates a detailed picture of the heart. Today’s cardiac MRI is a vast improvement from what patients would have experienced as recently as a decade ago. But after the MRI generates a detailed image of the heart, the process reverts to “manual” assessment by a specially trained cardiologist. The doctor must spend 20-30 minutes reviewing the results before proceeding further.
That time spent is more than inconvenience. It reduces the number of patients the cardiologist can see in a day. It reduces the time available to perform other procedures to understand the patient’s heart health and needs. Most importantly, it takes time away that the cardiologist could be spending with his or her patients.
For cardiologists, the ideal solution is an advanced algorithm that could perform an analysis of the cardiac MRI. Such an algorithm could accomplish in a few seconds what currently takes 20-30 minutes to complete. Yet the search for an algorithm that can analyze cardiac medical imaging – to at or near human- level accuracy – has proven elusive.
This year’s competition sought to tackle the time-intensive nature of cardiac MRI analysis. Can we develop an algorithm that could, if successful, automate the analysis of the MRI-generated data? In doing so, we can empower doctors to help people live longer and spend more times with those that they love.
“The passion and technical expertise of the global data science community that is harnessed by the Data Science Bowl is extraordinary. They have been able to push the boundaries of what’s possible in this extremely complex and important challenge,” commented Josh Sullivan, Booz Allen Senior Vice President, himself a data scientist.
The 2016 Data Science Bowl came to close Monday night. Over the course of the competition 192 teams submitted more than 375 solutions. The winners will be announced soon, once the algorithms from those who top the leaderboard are verified. So far, what’s clear is that participants have created solutions that approach or exceed human-level accuracy for this extremely complex task. To get to this point is exciting; to solve the challenge, historic.
Dr. Michael Hansen, Ph.D., is one of the Principal Investigators from the National Institutes of Health National Heart, Lung, and Blood Institute. He expressed his enthusiasm for how the Data Science Bowl can solve complex problems. “If the cardiology community could have solved this on its own, it would have,” he added. “The Data Science Bowl brings different perspectives, different backgrounds and fresh eyes to the problem. And that is effective.”
For many involved in the competition, improving cardiac care is more than just an academic challenge.
Maia Will, a Senior Consultant with Booz Allen, experienced the importance of cardiac imaging first-hand, when her then three-year old son was diagnosed with a hole in his heart.
“This year’s Data Science Bowl competition represents more than just improving cardiac imagery analytics, it captures and clarifies an experience that is the first step in many journeys, all hopefully ending with a thankful walk out of the hospital by the beneficiaries of your work,” she remarked, in an earlier blog post.
Results from the 2016 Data Science Bowl will be announced on March 21.