Intel has announced a new competition for developers that will require them to devise a system that can assist with the diagnosis of cervical cancer. The algorithms developed by the competition participants will be used to screen for pre-cancerous lesions. Intel is putting up $100,000 in prizes for the top three solutions, with a first prize of $50,000.
Intel’s vice president, Doug Fisher said “We aim to challenge developers, data scientists and students to develop AI algorithms to help solve real-world challenges in industries including medical and healthcare.”
Treatment of cervical cancer is highly effective, although outcomes depend on early identification of the disease. In locations around the world with low resources, cervical cancer is the leading cause of cancer death in women.
Cervical cancer can take around 20 years to develop, although provided cancer is detected in the first five years, treatment can be administered in a single visit. Furthermore, that treatment is cheap, only costing around $28.
The 5-year survival rate for women diagnosed with stage 0 cervical cancer is 93%, although that figure falls to around 80% for stage 1B, 63% for stage IIA and 35% for stage IIIA.
The use of artificial-intelligence based systems to assist with screening could help detect cervical in the early stages, ensuring patients get treatment promptly. Fisher said, “This will aid the ability to make real-time determinations on treatment.”
The competition is being run by Intel, MobileODT, and Kaggle. MobileODT has a database of tens of thousands of cervical images collected from healthcare providers from more than 20 countries around the world. MobileODT will be providing the data for the competition. The images will be hosted on a framework provided by Kaggle.
The algorithms developed by the competition participants will be used to determine eligibility for treating pre-cancerous lesions using cryoablation. The competition will be split into two stages. Stage one involves reviews of 4,000 patient cases to determine eligibility for treatment. This phase will be overseen by Professor Albert Singer, a leader in Colposcopy education and research. The second phase of the competition will see participants classy the images based on cervical pre-cancer risk.