A new artificial intelligence tool has been developed to predict if patients with breast cancer would benefit from chemotherapy.
Engineers at the University of Waterloo who have developed the new tool believe it will better help unsuitable candidates avoid the major side effects of chemotherapy, as well as giving those who are suitable better surgical outcomes.
The initiative is part of the open source Cancer-Net initiative.
“Determining the right treatment for a given breast cancer patient is very difficult right now, and it is crucial to avoid unnecessary side effects from using treatments that are unlikely to have real benefit for that patient,” says Dr Alexander Wong, a professor of systems design engineering and leader of Cancer-Net.
“An AI system that can help predict if a patient is likely to respond well to a given treatment gives doctors the tool needed to prescribe the best personalized treatment for a patient to improve recovery and survival.”
The AI was “trained” on images of breast cancer produced through a novel magnetic resonance technique invented by Wong’s team called synthetic correlated diffusion imaging (CDI).
CDI images of old breast cancer cases and information on their outcomes led the AI to come up with predictions about whether pre-operative chemotherapy would be beneficial for new patients based on their images.
The researchers found a prediction accuracy of 87.75 percent, more than three percent higher than “the next best gold-standard” involving invasive and expensive MRIs.
“I’m quite optimistic about this technology as deep-learning AI has the potential to see and discover patterns that relate to whether a patient will benefit from a given treatment,” says Wong.
Breast cancer is now the most commonly diagnosed cancer, passing lung cancer. One in eight cancer diagnoses are breast cancer patients.
In 2020, 2.3 million breast cancer diagnoses and 685,000 deaths were recorded globally according the World Health Organization. More than 20,000 new cases of breast cancer were reported in Australia in 2022 according to Cancer Australia, with about 3,000 deaths from the disease.
As a result, there has been considerable effort in helping in the early diagnosis and treatment of breast cancer. And much of that has involved AI.
But it hasn’t been smooth sailing.
In the US in 1998, an early AI tool began being used to detecting breast cancer but a 2015 study of the process showed mammography sensitivity was actually reduced when the tool was in use, putting lives in danger.
Read more: Promise and problems: how do we implement artificial intelligence in clinical settings and ensure patient safety?
In this most recent work, detailed in a paper recently presented at the major international AI conference NeurIPS 2022, the researchers warn of potential issues.
“Potential negative societal impacts include over-reliance on technology and misuse of collected data,” the authors write.
They say that, though promising, the tool would have to be retrained consistently. They also note that “misuse of collected data could occur as the proposed method could be used to forecast future medical fees that insurance companies could leverage to increase their premiums on insurance plans.”