The end result of the project, the researchers wrote, will be models that can aid in decision making “by clustering patients prior to the start of immunotherapy through a unique combination of somatic and germinal traits.”

The inclusion of modeling is what makes this trial potentially exciting, Dr Cooper said.

“Machine learning is not very intelligent and cannot distinguish between causal correlation and regular correlation,” Dr Cooper said. “The real novelty here is the modeling, which will try to describe mechanisms and explain how things work, allowing [the models] to test hypotheses.”

Is incorporation of artificial intelligence into cancer clinical trials the wave of the future? According to Kun-Hsing Yu, MD, PhD, assistant professor of biomedical informatics at Harvard Medical School, Boston, incorporating machine learning or a biologically agnostic method would be helpful to identify previously unrecognized associations or signals.

Continue Reading

However, right now the majority of clinical trials include a relatively small data set, which is often a roadblock to the application of machine learning or artificial intelligence methods. These methods have more potential in getting nuanced or subtle signals out of a gigantic dataset, he said.

“These methods could potentially be used with real-world data … where we include people who are treated in the real-world settings after a drug approval by using electronic health records, insurance claims, or any other data set across different hospitals that are routinely collected,” Dr Yu said.

Indeed, in the past 5 years, machine learning and artificial intelligence–based research methods have been used on a wider scale, Dr Cooper said, but these technologies should continue to be incorporated into research cautiously.


Ciccolini J, Benzekry S, Barlesi F. Deciphering the response and resistance to immune-checkpoint inhibitors in lung cancer with artificial intelligence-based analysis: when PIONeeR meets QUANTIC [published online June 16, 2020]. Br J Cancer. doi: 10.1038/s41416-020-0918-3