Google's AI determines breast cancer more accurately than doctors

A new study shows that artificial intelligence developed by Google is better able to identify breast cancer than many doctors.

According to The Verge, research published in Nature shows that Google's artificial intelligence has been applied to help doctors detect breast cancer.

According to statistics, when artificial intelligence analyzes X-ray images, the rate of false-negative cancer identification (with cancer but not realizing) is reduced by 9.4%.

A doctor is reading a mammogram to identify cancer.(Photo: Getty).

Breast cancer is the second most dangerous cancer for women, just after lung cancer. Early detection is the best preventive and effective treatment for breast cancer. Mammography, also known as mammography, is a very effective tool for detecting breast cancer but the rate of errors is still high.

"Mammography is highly effective, but there is a problem with both positive and negative results," Shravya Shetty, a Google researcher and research co-author told The Verge.

Research funded by Google used mammography from 28,000 women in the UK and US. Artificial intelligence scans X-ray images, then looks for signs of breast cancer by examining breast changes. The researchers then tested the results from the computer with the doctor's actual conclusions.

The yellow box is where the artificial intelligence system thinks that there is breast cancer.(Photo: Northwestern University).

The research results show that the false negative rate is reduced to 9.4%, while the false positive rate decreases to 5.7% in the US. In the UK, where doctors often test one more time before concluding, Google's model reduced the rate of false negatives by 2.7% and false positives by 1.2%.

"This model outperforms radiologists in both the UK and the US," concluded Christopher Kelly, co-author of the study.

Of course, this model cannot completely replace the doctor. There are still many cases of cancer but the model is ignored. Google believes that this method is clinically applicable. According to the study's co-authors, scientists at Google are looking for ways to apply the model to more cases.

Google also emphasizes that this is just a tool to support doctors."The combination will take advantage of both sides, giving better results , " said Shravya Shetty.

1. AI algorithm is capable of diagnosing skin cancer up to 96%
2. Diagnosis of cervical cancer faster by artificial intelligence