Breast cancer is known to be the second leading cause of cancer death in women. According to the World Health Organization, with early detection, there are more chances of surviving from this dreaded disease. But currently, breast cancer detection tools are only available to the white population.
Researchers from the Massachusetts Institute of Technology somehow came to the rescue by creating a deep learning-based AI prediction model that can analyze mammograms and detect breast cancer development five years in advance, which is “equally accurate for white and black women.” The model used data from more than 90,0000 mammograms from 60,000 patients and discovered that there are breast tissue patterns that are present before the detection of malignant tumors.
The said model proved to be better at projecting risks compared to other methods. Other long-established methods can only place 18 percent of cancer patients in the highest-risk category; this model can precisely place it by 31 percent. Other risk models are based on factors such as age, family, history of breast cancer, breast density, and genetic factors.
Even though these models assisted in early cancer detection, they lack essential data per patient and does not yield accurate results individually.
The team hopes that by using this model, doctors can personalize cancer screening and cancer prevention programs. As a result, cases of late diagnosis won’t occur in the future “…rather than taking a one-size-fits-all approach; we can personalize screening around a woman’s risk of developing cancer. For example, a doctor might recommend that one group of women get a mammogram every other year, while another higher-risk group might get supplemental MRI screening,” says Regina Barzilay, an MIT Professor who’s also the senior author of the paper and a breast cancer survivor. The team is composed of researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) and Massachusetts General Hospital (MGH).
Black/African-American women have more chances of dying from breast cancer compared to white women by more than 42 percent. Another study stated that at least five years after the cancer diagnosis, 88 percent of white women survive compared to 73 percent of African-American women. One main reason behind this sad reality is because current early detection techniques are not accessible or not accurate for black women.
Based on the Susan G. Komen breast cancer foundation, the barriers to breast cancer screening are low financial income, lack of access to healthcare, lack of awareness of breast cancer risks/screening methods and cultural/language differences, to name a few. Also, there are reproductive factors and biological differences between Black/African-American and white women that play a role.
The team stated that the project’s goal is to assess health risks more accurate, especially for racial minorities. Most of the models are created on white populations and are less precise for other cultural races. The model prides itself for being fair and equally accurate for both white and black women. And since this model can detect breast cancer in advance, this can reduce invasive treatments and minimize the cost of medical expenses.
“If you give the right screening to the right person, you can both improve the experience and reduce the harms of mammography but also catch the cancers earlier, which makes a huge difference in treatment decisions, because what you do for early stage and late stage cancer are very different,” says Adam Yala, the paper’s lead author.
Currently, the team is working on having more collaborations with more hospitals to expand their knowledge and serve other groups and make the model even more fair and partial.
In the future, the team hopes that their model can also be used to check if patients are at high-risk of other health problems such as cardiovascular diseases or different types of cancers which have less effective risk models, Barzillay stated.
“Our goal is to make these advancements a part of the standard of care. By predicting who will develop cancer in the future, we can hopefully save lives and catch cancer before symptoms ever arise.”,Yala added.
Aside from MIT’s AI model, other famous companies or institutions have similar projects with the goal to assist or improve the diagnosis and treatment of breast cancer. Some of these organizations are IBM, Google, the Alphabet subsidiary DeepMind, New York University, and Harvard Medical School.