Fund effort seeks to ID lung cancer

August 10, 2017 GMT

A crowdfunding campaign aimed at building artificial intelligence into tools used to spot lung cancer is offering $100,000 to coders, engineers and researchers who will build the software that could identify the world’s most deadly cancer earlier and more accurately.

The Bonnie J. Addario Lung Cancer Foundation, which put up the funds for the collaborative effort, wants to develop software that will improve the ability of CT scans to pinpoint lung cancer while it’s most treatable. And the goal by April is to have a program that can be delivered to clinics.

“We wanted to focus on big data, machine learning,” Guneet Walia, the foundation’s senior director of research and medical affairs, told the Herald. “We wanted something that wasn’t just some code sitting on some engineer’s computer, but something that could work in a clinic.”

Cancer experts, big data scientists, engineers and others will contribute patches and improvements to a diagnostic tool. Participants will get points for adding lines of code to move the project forward and cash awards for gaining the most points in certain categories.

The foundation wants the push for earlier diagnoses to help it reach its ultimate goal of having lung cancer be a chronically managed disease by 2023 — “something you die with, not die of,” Walia said.

Lung cancer, which kills more people per year than any other cancer, is typically diagnosed late and after it has spread to other parts of the body. One of the problems with spotting lung cancer early, experts say, is it often doesn’t show symptoms until later stages when it has spread.

“Most people who have symptoms — a cough, coughing up blood, shortness of breath — they already have advanced disease,” said Dr. Christopher Lathan, of the Dana-Farber Cancer Institute. “Finding lung cancer when it is asymptomatic is the best way to cure people.”