The development of a cancer drug is currently long, costly and requires the involvement of
several thousand patients. Yet only 5% of these drugs will be approved for cancer treatment.
RESOLVED2 is a machine-learning method that predicts the approval of a drug at the end of
a Phase I clinical trial. Phase I clinical studies are very early in the development of
a drug, and RESOLVED2 requires an average of only 30 patients to make predictions. When
RESOLVED2 predicts that a treatment will be approved, 73% of the treatments are indeed
approved within 6 years, while the others are during the subsequent years
RESOLVED2 is also correct for 92% of treatments that will not be approved in 6 years.
RESOLVED2 has the potential to reduce the number of patients to be included in clinical
trials, by increasing the likelihood of routine use of treatment from 5% to 73%.
A prospective evaluation is planned to confirm the usefulness of routine use of RESOLVED2.