SYDNEY, June 26 (Xinhua) -- Australian and American scientists have developed an artificial intelligence (AI) tool that could revolutionize cancer treatment by mapping cellular diversity within tumors.

The innovation tackles tumor heterogeneity in oncology, where varied cell populations cause treatment resistance and recurrence, according to a release from the Sydney-based Garvan Institute of Medical Research on Wednesday, which co-led the study with the Yale School of Medicine in the United States.

The AAnet AI tool uses deep learning to study gene activity in single cancer cells. It finds five different cell types within tumors, each with its own behavior and risk of spreading. This helps doctors understand cancer better than older methods, which treated all tumor cells as the same, said the multinational research team.

"Heterogeneity is a problem because currently we treat tumors as if they are made up of the same cell. This means we give one therapy that kills most cells in the tumor by targeting a particular mechanism. But not all cancer cells may share that mechanism," said the study's co-senior author, Associate Professor Christine Chaffer from the Garvan Institute.

As a result, some cancer cells survive and the disease can return, Chaffer said, adding that AAnet provides a way to biologically characterize tumor diversity, enabling the design of combination therapies that target all cell groups at once.

Associate Professor Smita Krishnaswamy of Yale University, a co-developer of the AI, indicated that this is the first method to distill cellular complexity into practical archetypes, potentially transforming precision oncology.

The technology is ready for clinical use, with plans to combine AI analysis and traditional diagnostics to create treatments tailored to each tumor's cell types. Validated in breast cancer, it also shows promise for other cancers and autoimmune diseases, marking a shift toward personalized medicine, according to the study published in Cancer Discovery. ■