New Model Enhances Cancer Drug-Response Ranking With Sparse Data

VERA Brief

AI-generated. Grounded in the article and its cited sources.

Researchers have developed a new Reverse Item Response Theory (IRT) model to improve the accuracy of ranking cancer drug responses, especially with incomplete data. This method treats cancer types as subjects and drugs as items, estimating resistance and activity on a shared scale.

Key facts

• A new Reverse Item Response Theory (IRT) model has been developed to enhance the accuracy of ranking cancer drug responses.
• The model conceptualizes cancer types as latent subjects with resistance abilities and drugs as items with evasion difficulties.
• The model was applied to drug sensitivity measurements from the Genomics of Drug Sensitivity in Cancer (GDSC2) database.
• Reverse IRT outperformed simple averaging in recovering full-data latent rankings under missingness conditions.
• The IRT model achieved the best Brier score among five evaluated methods in held-out prediction.