While combinatorial drug therapies have been shown to increase the efficacy of cancer treatment and alleviate harmful side-effects by reducing the dosages needed by patients, the expensive and painstaking nature of experimental screening for combinations has been an obstacle to realising their full benefits.
Trained with a wealth of data obtained from previous studies, the machine learning model can accurately predict how different combinations of cancer drugs eliminate various types of cancer cells.
The researchers have demonstrated that the model is capable of discovering previously unobserved associations between drugs and cancer cells and predicting accurately how a combination of drugs would inhibit certain cancer cells. The results have been published in Nature Communications, an open-access peer-reviewed multidisciplinary journal.
“This will help cancer researchers to prioritise which drug combinations to choose from thousands of options for future research,” said Tero Aittokallio, a researcher at the University of Helsinki’s Institute for Molecular Medicine Finland.
The model can also be re-calibrated for other diseases by re-teaching it with data related to them.