AI and advanced data metrics are forging a new era in cancer research

A new project by a team of researchers across the nation analyzes the ways in which digitized health data, artificial intelligence models and other recent technological advancements have changed how cancers are diagnosed, studied and treated.

Led by Indiana University School of Medicine’s Spyridon Bakas, Ph.D., the team’s findings are published in Cancer Research.

“Informatics and AI are being used in every part of the clinical data extracted from cancer patients, from medical imaging to electronic health records, lab results, blood tests and molecular profiling, to expedite diagnoses,” said Bakas, senior corresponding author of the article, an associate professor and division chief in the IU School of Medicine’s Department of Pathology and Laboratory Medicine, and a researcher within the IU Melvin and Bren Simon Comprehensive Cancer Center.

AI models have been trained to identify cancerous cells in tissue that humans may not be able to perceive with the naked eye, while also scanning data repositories for relevant information at a much faster rate than any human researcher, the paper notes.

Finally, AI can identify patterns by analyzing vast quantities of data in ways previously not possible.

But oncologists and human researchers are still needed.

“AI could be used to help with faster triaging of known cases, but the human oncologist will be essential to approve and access rare cases that are not well represented in the AI models’ training,” Bakas said.

Developing standardized methodology for these informatics procedures will be necessary, the article says, as will growing the databases to better train the models and developing software tools that will allow researchers to access the information.

Bakas was joined by nearly two dozen co-authors from universities and institutes around the country, including the University of Maryland, Johns Hopkins University and Harvard University.

“Cancer Informatics is an expanding field with the rapid emergence of increasingly high-resolution, high-dimensional, multiomic datasets,” said Kathleen Noller, Ph.D., lead author and postdoctoral fellow at the Institute for Genome Sciences at the UMD School of Medicine. “This review illuminates cutting-edge methods in cancer biology, as well as their drawbacks and some areas for improvement. Ultimately, these technologies will help in predictive and precision care for cancer patients.”