Researchers at the Massachusetts Institute of Technology in Cambridge and Boston-based Massachusetts General Hospital used the known cancer outcomes and nearly 89,000 mammograms of 39,571 women to train the model to recognize subtle patterns in breast tissue that indicate the formation of malignant tumors.
As a result, the model accurately characterized 31 percent of cancer patients as high-risk, while traditional models only identified 18 percent as such. The model was also equally accurate for patients of all races, notable because many traditional models have only been trained on white patients’ data, leading to lower rates of detection in non-white populations.
More articles about AI:
AI detects anxiety, depression in children’s speech patterns
IBM Watson imaging AI goes live in 1st health system
Recursion Pharmaceuticals releases open-source 300-gigabyte dataset to improve AI use in biotech
At the Becker's 11th Annual IT + Revenue Cycle Conference: The Future of AI & Digital Health, taking place September 14–17 in Chicago, healthcare executives and digital leaders from across the country will come together to explore how AI, interoperability, cybersecurity, and revenue cycle innovation are transforming care delivery, strengthening financial performance, and driving the next era of digital health. Apply for complimentary registration now.
