Peter MacCallum Cancer Centre shared on LinkedIn:
”Artificial intelligence is opening new possibilities across medicine and research – helping experts analyse vast amounts of information, identify patterns and potentially deliver more precise care at scale.
Now, two major international studies led by Peter Mac’s Professor Sherene Loi have found AI was just as good as a human pathologist at a task used to predict outcomes for people with breast cancer.
The studies examined data from more than 5,600 patients with triple-negative and HER2-positive breast cancer. Researchers compared different methods of counting tumour-infiltrating lymphocytes, or TILs – immune cells found within tumours that can provide important information about a patient’s prognosis.
AI and pathologists produced similarly useful predictions, with both methods showing that patients with higher levels of TILs generally had better outcomes.
AI was also able to analyse how immune cells were organised within tumours and identify ‘hotspots’ that provided additional information beyond simply counting the cells.
‘Ultimately, combining pathologists and computational approaches may give us more information than either approach alone’
Professor Loi said.
Published simultaneously in The Lancet Group Oncology, the findings show how AI could complement human expertise and help make valuable cancer biomarkers more widely available – particularly in settings where specialist pathology assessment is limited.”
You can also read: Iza-Bren Reaches the Phase III PFS Endpoint in HR+/HER2− Advanced Breast Cancer: Why PANKU-Breast01 Matters
