$1M in Prize Awards for Data Science Bowl 2017: Helping Radiologists Lower the False-Positive Lung Cancer Scans

In January 2016, Vice President Joe Biden’s created the Cancer Moonshot initiative to help promote 10 years’ worth of advances in cancer research, prevention and treatment within just five years. One of the main goals of Moonshot is to bring the expertise of diverse fields together to solve pressing cancer problems. Following VP Biden’s lead Kaggle, a company that promotes the use of data science in all fields, announced this year’s Data Science Bowl competition will be aimed at improving the data science behind diagnosing lung cancers, bringing together data scientists and medical providers, including radiologists.

Currently, the rate of false positives for lung cancer is at 95 percent. Across all industries this is unacceptably high. To improve this technology, Kaggle is hosting this year’s Data Science Bowl with Booz Allen Hamilton, a provider of management and technology consulting services. Further, the Laura and John Arnold Foundation joined in to provide largest prize in Data Science Bowl history—$1M in prizes. This year the first place winner will be awarded $500,000; the second place winner, $200,000; third place, $100,000; and each of the following seven runners-up will receive $25,000.

Using high-resolution lung imaging provided by National Cancer Institute, the competitors will be tasked with developing an algorithm that can accurately identify which lung lesions are cancerous. The data set includes over 1,000 low-dose CT scans from high-risk patients. Depending on the type of machine used to take the scans, every image contains a variable number of 2D slices. The hope is that by giving data scientist access to these scans which are very familiar to radiologists, a new technology will be developed. They hope the advancements made through this competition will provide the medical community with the ability to lower the false-positive rates and help patients get the care and intervention they need during the earliest stages of lung cancer. The development of this algorithm would also would enable radiologists to spend more time with their patients.

The American Cancer Society estimates that in 2017 there will be about 222,500 new cases of lung cancer in the United States alone. If you think you have have the know-how to improve lung cancer detection, there is still time. The competition is open to new competitors (and merged teams) until April 12, 2017. To learn more or to look at the CT scan data set yourself go the Data Scientist Bowl 2017 information page.