Artificial intelligence predicts drug combinations that kill cancer cells most effectively

A machine learning model developed in Finland can help us treat cancer more effectively.

When healthcare professionals treat patients with advanced cancers, they usually need to use a combination of different therapies. In addition to cancer surgery, patients are often treated with radiation therapy, drugs, or both.

The drugs can be combined with different drugs that act on different cancer cells. Combinatorial drug therapies often improve treatment efficacy and can reduce harmful side effects if the dosage of individual drugs can be reduced. However, experimental screening of drug combinations is very slow and expensive and therefore often fails to uncover all the benefits of combination therapy. With the help of a novel machine learning method, the best combinations could be identified to selectively kill cancer cells with a specific genetic or functional composition.

Researchers from Aalto University, the University of Helsinki and the University of Turku in Finland have developed a machine learning model that accurately predicts how combinations of different anticancer drugs kill various types of cancer cells. The new artificial intelligence model was trained with a large data set from previous studies that had studied the association between drugs and cancer cells. “The model learned from the machine is actually a polynomial function familiar from school mathematics, but very complex,” says Professor Juho Rousu of the University of Aalto.

The research results were published in the prestigious journal Nature Communications, showing that the model found associations between drugs and cancer cells that had not previously been observed. ‘The model gives very accurate results. For example, the values ​​of the so-called correlation coefficient were more than 0.9 in our experiments, which indicates excellent reliability, “says Professor Rousu. In experimental measurements, a correlation coefficient of 0.8-0.9 it is considered reliable.

The model accurately predicts how a drug combination selectively inhibits particular cancer cells when the drug combination’s effect on that type of cancer has not previously been tested. “This will help cancer researchers prioritize drug combinations to choose from among thousands of options for further research,” says researcher Tero Aittokallio of the Institute of Molecular Medicine of Finland (FIMM) at the University of Helsinki.

The same machine learning approach could be used for non-cancerous diseases. In this case, the model should be re-taught with data related to that disease. For example, the model could be used to investigate how different combinations of antibiotics affect bacterial infections or how effectively different drug combinations kill cells that have been infected with the SARS-Cov-2 coronavirus.

Reference: 1 December 2020, Nature Communications.

DOI: 10.1038 / s41467-020-19950-z