Artificial Intelligence and Diagnosing Rare Cancers

Milan Radovich, Ph.D., Chief Scientific Officer at Caris Life Sciences, discussed the use of artificial intelligence (AI) to improve the diagnosis of different cancer patients.

As noted by Dr. Radovich, each patient, with their imaging data, electronic health records, exome sequencing data, etc, can lead to a level of data that is too large for a person to comprehend (e.g., a petabyte of data). Fortunately, we are entering a phase where powerful computing technology, power sequencing technology, and the implementation of AI algorithms are emerging to diagnose patients more efficiently.

As an example, Dr. Radovich noted that at Caris Life Sciences, their AI programs are becoming very useful at detecting misdiagnosed cancers. “Every day we get a handful of cases where the computer algorithm says it's a different diagnosis than what came in on the path report, and so what our pathologists will do is, they'll take that case they'll work it up further, with some additional IHC stains and they'll contact the physician to walk them through their the case,” noted Dr. Radovich.

Another example that AI can be useful is to predict brain metastases. As stated by Dr. Radovich, at the American Society of Clinical Oncology annual meeting being held in Chicago (ASCO 2023), Abraham and colleagues presented data showing how AI can possibly be used to predict if a person with cancer is at risk for brain metastasis.