Check Out The Cancer Detection Program That Won Top Prize At Google's Science Fair

bc

Photo: Brittany Wenger/Google Science Fair

Brittany Wenger, 17, won the top prize of the 2012 Google Science Fair by creating an artificial “brain” that can diagnose breast cancer with amazing accuracy.Wenger built a neural network  — a computer program coded so that it “learns” to process data and detect patterns — that is 99.1 per cent sensitive to malignant growths.

The Sarasota, Florida, native wanted to improve the accuracy of the least invasive diagnostic procedure, called fine needle aspirate, so women don’t have to undergo a second biopsy with a bigger needle or even surgery.

She managed to create a program that is better than three commercially available programs at giving correct diagnoses for cancerous lumps.

Here’s how she did it.

See the presentation >
Read more about the artificial “brain” that can accurately diagnose breast cancer  >
See our list of Game Changers: 30 Innovations That Will Change The World >

Cloud4Cancer builds on data collected by the University of Wisconsin in the early 1990s.

Wenger said creating the network was close to her heart because she has family and friends who have been diagnosed with breast cancer.

Currently fine needle aspirates are the least invasive procedure to diagnose breast cancer, but they are often inadequate.

Wenger aimed to improve the accuracy of this test so doctors wouldn't have to resort to more invasive procedures.

Wenger began building neural networks in seventh grade.

Wenger tested her model against three commercially available neural networks.

Her network turned out to be better than all three as its false negative rate — how often it says a growth is cancerous when it isn't — is much lower than that of the other models.

Wenger built a network that could process massive amounts of data.

Currently, the network detects malignancy with 99.1 per cent accuracy.

She published the network on Google's cloud service to allow global submissions.

Anyone can go to her site and insert data.

This is how her service will accept the data and process a diagnosis.

In the end her network was much better than what's currently available.

Her custom neural network achieved predictive success of 97.4 per cent with 99.1 per cent sensitivity to malignancy.

In fact she blew them out of the water.

And the best part is that it is only going to get progressively better.

Wenger thinks that if her network isn't hospital-ready now, it will be soon.

NOW WATCH: Briefing videos

Business Insider Emails & Alerts

Site highlights each day to your inbox.

Follow Business Insider Australia on Facebook, Twitter, LinkedIn, and Instagram.