Artificial intelligence (AI) is not just what we're shown in movies of human-like robots trying to take over the world. So… what is Artificial Intelligence? It is a computer program designed to try to replicate and develop operations of human intelligence.
And can this intelligence help us in the detection of breast cancer? Of course. Although there have been incredible advances in recent years, breast cancer is still diagnosed in advanced stages in many cases. But let's see how this intelligence can be useful…
A couple of years ago, researchers at Massachusetts General Hospital and CSAIL developed a deep learning* model capable of predicting from mammograms whether a woman is likely to develop breast cancer 5 years from now. Starting from patterns of breast tissue that became malignant was how they trained the AI.
*Set of machine learning algorithms that tries to mimic the functioning of the nervous system.
Another objective proposed by these researchers was to provide one more tool to solve racial bias, since current early detection methods have a higher success rate among the Caucasian population. And this affects that African-American women who are the ones who are 43% more likely to die from a late detection of cancer. The results were as expected, since this new detection method works the same in both races. With this we see that when the starting biases are eliminated and it is planned correctly, an investigation can reach results without bias.
This new algorithm called “Mirai”, based on a mammogram, models the possible risks that a patient may have in the future and if her background, both family and personal, and her clinical risks are also available, the algorithm includes it in its analysis. The results were that it is significantly more accurate than previous methods in predicting and identifying high-risk groups. Which is great, as it would improve the reliability of mammograms thanks to artificial intelligence.
However, could we have another use for this AI without the use of mammography? The WHO tells us that mammograms are not the best screening tool, but it is the only one we have today and we must use it. It is an economically expensive, painful and irradiating test. In fact, a high number of women do not undergo screening due to the pain it causes them, because they feel embarrassed or due to lack of time.
The Blue Box that we are developing aims to solve these problems. Using artificial intelligence, but not to improve and perfect mammography, but to offer a cheaper, painless and radiation-free alternative tool. AI can be a powerful tool in the investigation and detection of many pathologies.
Enséñame de ciencia. (23 de marzo de 2022). Gracias a una nueva IA, el cáncer de mama se podrá predecir con años de anticipación. https://ensedeciencia.com/2022/03/23/gracias-a-una-nueva-ia-el-cancer-de-mama-se-podra-predecir-con-anos-de-anticipacion/
Merino, M. (8 de mayo de 2019). Predecir un cáncer de mama cinco años antes de que aparezca, posible gracias a la inteligencia artificial. Xataka. https://www.xataka.com/inteligencia-artificial/predecir-cancer-mama-cinco-anos-antes-que-aparezca-posible-gracias-a-inteligencia-artificial
World Health Organization. (2014). WHO position paper on mammography screening. World Health Organization.