Each year, cardiovascular disease causes more than 30 percent of deaths around the globe. In the United States, someone is diagnosed with the condition once every 43 seconds. Finding a way to enhance and expedite the diagnosis of heart disease requires the attention and support of public and private sector organizations worldwide.
McLean, VA-based Booz Allen Hamilton and Kaggle announced on December 14 that the second annual Data Science Bowl will call on the global data science community to create a set of steps, or algorithms, to help transform diagnosis of heart disease. Through a partnership with the National Institutes of Health (NIH), participants in the 90-day competition will be given MRI images and asked to develop an algorithm to automate the measurements that are key indicators of heart disease.
“This competition will empower data scientists from across the world to improve both the ability to diagnose and the capacity to care for those with heart disease, enabling people to live longer, healthier lives,” said Booz Allen Hamilton President and Chief Executive Officer Horacio Rozanski. “Booz Allen is proud to be a leader in pursuing data science for social good. This project has great potential to demonstrate how data scientists solve problems that affect millions of people each and every day.”
“We hope the Data Science Bowl will be rewarding and fun for our competitors,” said Senior Vice President Josh Sullivan. “But above all, we expect that this competition will significantly progress vital cardiac research through advanced analytics. It is that critical, real intersection of data science and healthcare that makes us most proud.”
The heart’s ability to efficiently pump blood is currently measured through a manual process that takes a specially trained cardiologist approximately 20 minutes to complete – vital time the physician could be spending with his or her patients. The Data Science Bowl’s data set features more than 1,000 images from a broad spectrum of individuals with different ages and genders, ensuring the winning algorithm will be able to handle the varying, yet precise demands of real-world patient-care while drastically cutting the time and cost of the traditional process.
“Combining our rich data set with the collective power of data scientists presents a truly unique opportunity to redefine heart health research in just three months – an almost unbelievably short period of time in this industry,” said Dr. Michael S. Hansen, a NIH specialist in fast MRI techniques for real-time imaging and interventional procedures. “This competition demonstrates advanced analytics’ ability to amplify the impact of the individual analyst, and real-time cardiac measurement promises to significantly improve care where it matters most; at the doctor-patient level.”
At the close of the competition, the algorithm will be released in an open-source format, which will allow cardiologists and NIH researchers to explore its integration with diagnostic technologies. The participants who deliver the most accurate algorithm will be granted $125,000, with the second and third-place teams receiving $50,000 and $25,000, respectively. This year, visual computing leader and special Data Science Bowl sponsor NVIDIA contributed $25,000 in prize money, and will provide all participants with free access to online deep learning courses. NVIDIA will also provide the three winning teams with free passes to its 2016 GPU Technology Conference (April 4-7, 2016, in San Jose, California), where they plan to have the winning teams present their results.
“This year’s Data Science Bowl will be one of the most difficult computer vision problems our community has ever approached,” said Kaggle CEO Anthony Goldbloom. “It also represents an exciting opportunity for data scientists around the world to collaboratively aid and advance heart health research. We’re thrilled to be partnering with Booz Allen Hamilton and the National Institutes of Health in hosting this one of a kind challenge.”
The Data Science Bowl is a component of Booz Allen Hamilton’s continued support of data science education and awareness. The firm has also made a concerted effort to distribute industrywide best practices via its recently updated Field Guide to Data Science, a handbook that captures the discipline born of Booz Allen’s 600+ person data science team. In 2014, Booz Allen and Kaggle’s first annual Data Science Bowl challenged the global data science community to develop an algorithm that could better measure oceanic health at a speed and scale never before possible, in partnership with the Hatfield Marine Science Center at Oregon State University.
Source: Booz Allen Hamilton