Nat Commun: Scientists hope to use artificial intelligence to predict a combination of drugs that can effectively kill cancer cells

December 5, 2020 // -- In a recent study published in the international journal Nature Communications, scientists from the University of Alto and others said that artificial intelligence technology could be used to predict which combinations of drugs could effectively kill cancer cells.
When clinicians treat patients with advanced cancer, they often need to use a combination of anticancer therapies, and in addition to cancer surgery, patients often receive radiation, medication, or both.
Picture: Matti Ahlgren, Aalto University drugs can be used in combination with cancer-specific drugs, and if the dose of a single drug can be reduced, combination drug therapy can usually improve the efficiency of treatment and reduce harmful side effects of the drug, however, experimental screening of drug combinations is often very slow and expensive, so researchers often fail to see the full benefits of combination therapy; 'We've developed a machine learning model that accurately predicts how a combination of multiple cancer drugs kills multiple types of cancer cells,' said
researcher Juho Rousu. 'We were able to train this new AI model with a large amount of data from previous studies that focused on the association between drugs and cancer cells, and machine learning models are actually a polynthon function similar to school math, but very complex.'
The model was able to detect a link between drugs and cancer cells that had not previously been observed by researchers, and it was able to give very precise results, such as a value of more than 0.9 in experiments, indicating very high reliability, where the correlation coefficient of 0.8-0.9 was considered a very reliable result.
researcher Tyro Aittokallio said the new model could accurately predict how drug combinations selectively inhibit specific types of cancer cells when their effects on a particular type of cancer have not yet been tested, which could help cancer researchers select the best combination of drugs from thousands of combinations to conduct in-depth research.
The same machine learning method can also be used as non-cancerous cells, in which case researchers will have to retrain new models using disease-related data, such as how different combinations of antibiotics affect bacterial infections, or how different combinations of drugs can effectively kill host cells infected with SARS-CoV-2.
() Original source: Julkunen, H., Cichonska, A., Gautam, P. et al. Leveraging multi-way interactions for systematic prediction of pre-clinical drug combination effects. Nat Commun 11, 6136 (2020). doi:10.1038/s41467-020-19950-z。