healthcare

Google Claims Its AI Breast Cancer Screening Outperforms Human Experts

So does Dr. Google have its license to practice?
article cover

Francis Scialabba

· less than 3 min read

Keep up with the innovative tech transforming business

Tech Brew keeps business leaders up-to-date on the latest innovations, automation advances, policy shifts, and more, so they can make informed decisions about tech.

Google Health and Alphabet subsidiary DeepMind published research Wednesday showing impressive advances in AI-based breast cancer screening.

Let's unpack the findings

Working with U.K. and U.S. clinical researchers, Google's team claims its AI system outperformed expert radiologists in most screening mammograms.

To train the AI system, the research team used over 90,000 de-identified mammograms, then evaluated its performance on mammograms from 25,856 British women and 3,097 American women. They compared the AI model's predictions against the actual cases of biopsy-proven breast cancer.

The AI system reduced false positives by 5.7% and false negatives by 9.4% for U.S. patients.

  • False positives: When results show you have a disease, but you don't actually have it.
  • False negatives: When the test does not detect the presence of a disease you actually have.

The patient POV

The American Cancer Society says one in five screening mammograms do not find breast cancer (false negative). And roughly half of all women receiving annual mammograms will encounter a false positive finding over a 10-year period.

If AI screening models can increase the accuracy of breast cancer detection, it could improve healthcare outcomes and save lives. Pundits often argue that radiologists and diagnosticians could be automated out of a job, but Google's researchers say AI screening systems would complement professionals and provide a second opinion...not replace them.

So does Dr. Google have its license to practice?

Not so fast. While promising, these results aren't necessarily reproducible across a wider population or applicable to mammography techniques not included in Google's research. The team needs regulatory approval and more time to refine the AI model before any clinical deployments.

Plus, Google's healthcare push could be haunted by the ghosts of prior privacy and data protection indiscretions. Judging by the reaction to Project Nightingale, Google's partnership with the the U.S.' second largest health systems provider, the information-gathering company may want to be exceedingly transparent about healthcare projects upfront.

Healthcare is a heavily regulated, no-nonsense industry. By comparison, Google's fortés (ad targeting and search engines) are the Wild West.

Keep up with the innovative tech transforming business

Tech Brew keeps business leaders up-to-date on the latest innovations, automation advances, policy shifts, and more, so they can make informed decisions about tech.