While there may be some reasons to be wary of artificial intelligence, it’s an increasingly effective tool in the health industry.
Researchers from the University of British Columbia and BC Cancer have developed an AI model that predicts a cancer patient’s survival rate with over 80 per cent accuracy, according to new research released from the university.
The AI model – which can understand complex human language – looks at characteristics of the cancer diagnosis by examining an oncologist’s notes, effectively predicting six-month, 36-month and 60-month survival rates.
The model will then give appropriate recommendations, such as getting a quicker referral or starting a more aggressive treatment.
In most settings, cancer treatment recommendations have been established retrospectively and “categorized by only a few generic factors such as cancer site and tissue type,” according to a UBC news release. Despite familiarity with the disease, the unique complexities affecting individual cancer cases can make it difficult for oncologists to predict survival. AI changes this.
The AI model is currently only trained on B.C. data; however, it’s only a matter of time before the tool can be used across Canada, changing and improving the lives of hundreds of thousands of lives.
The data from 47,625 B.C. patients were used to train the model. The study was published in Jama Network Open on Feb. 27.
Nearly one-half of Canadians will be diagnosed with cancer in their lifetime, according to national statistics. Cancer screening is a critical step in prevention.
BC Cancer recommends mammograms for women over 40 years old, cervical screenings every three years for those with a cervix aged 25 to 69, colon screenings for ages 50 and up and lung screenings for those 55 and older or with a smoking history.
Do you have a comment about this story? email:
kim.kimberlin@blackpress.ca
