Here’s How Artificial Intelligence Can Help Predict Breast Cancer Risk
For Dr. Denis Lacombe, CEO of the European Organisation Research Treatment Cancer (EORTC) to fight cancer you need big data.
“Cancer is notoriously complex,” said Lacombe. “Not only is it more than 200 separate diseases, but it can express itself differently in each person and at each stage of its progression.”
Lacombe believes that for patients, big data and machine learning can transform cancer, reducing uncertainty and truly personalising treatment and care, as opposed to a one size fits all approach.
“Finding the most tailored treatment for a person – the one that is going to have the best results – requires researchers to be precise and targeted,” said Lacombe. “Clinical research needs to bring robust evidence across all these different datasets to change practice.”
According to the World Health Organization (WHO), in 2020, 2.3 million women were diagnosed with breast cancer.
“Breast cancer, like many other cancers, is an extremely diverse disease, with multiple different tumour types,” said Lacombe. “The collaboration between human expertise and complex algorithms is vital to help accelerate the pace of progress such as novel therapeutics, more tolerable treatments and better outcomes for patients.”
A new deep learning model
Researchers from the Jameel Clinic at the Massachusetts Institute of Technology (MIT) have been working on a mammography-based deep learning model to help predict breast cancer earlier.
The Mirai model uses an artificial intelligence (AI) algorithm to predict breast cancer risk more accurately based on radiology images.
In a press statement, the researchers said the Mirai model is significantly more accurate and can identify high-risk groups across all three datasets than prior methods in predicting cancer risk. The research published in Science Translational Medicine indicated that the Mirai model identified close to two times more future cancer diagnoses than the Tyrer-Cuzick model, the current standard.
In the research, the Mirai model was more accurate across patients in different age groups, breast density categories and race groups, which are commonly impacted by machine learning algorithms.
Adam Yala, a Ph.D. candidate at the Jameel Clinic, MIT, said that the big picture for patients with the Mirai model is better outcomes. “Our goal is to catch every cancer as early as possible when it is most curable and requires the least aggressive treatment.”
Unlike most breast cancer AI models, Yala says that they designed Mirai to predict if and when a patient will develop cancer in the next five years.
“Mirai can predict future cancer, and as a result, this model offers a unique opportunity to personalize patient screening,” added Yala. “In the future, I’m excited about the potential of Mirai to recommend patients for supplemental screening such as an MRI which can identify cancers not visible on a mammogram.”
Mirai is currently implemented at Massachusetts General Hospital (MGH), and the researchers are working to organize prospective clinical trials of the technology.
Seen on Forbes (Innovation): Article Link