Oxford University using AI to develop personalised cancer treatments

The approach utilises AI combined with mathematical modelling in an attempt to increase the efficacy of cancer treatments.

Professor Philip Maini, director of the Wolfson Centre for Mathematical Biology at Oxford’s Mathematical Institute, is a senior authors on the study.

He said: "This study illustrates how combining mathematical modelling with the power of AI could have significant impact on the clinical treatment of cancer, increasing effectiveness and reducing cost."

The new method may be more effective and less expensive (Image: Getty)

Traditional strategies for cancer treatment revolve around the delivery of maximum tolerated doses, killing as many cancer cells as possible.

However, these frequently fail against metastatic cancers, those which spread to other parts of the body, due to the emergence of drug resistance.

Adaptive therapy strategies adjusts treatment to reduce the growth of drug-resistant tumours.

However, the lack of personalized approaches that account for patient variation limits their efficacy.

In the study, using deep reinforcement learning - an form of AI - the researchers created personalised adaptive therapy schedules.

The results indicate that the new adaptive approach could potentially double the time to relapse compared to maximum tolerated dose or non-personalised treatment breaks.

Kit Gallagher, a DPhil student at Oxford’s Mathematical Institute and first author on the study, said: "In our computational simulations, these schedules consistently outperform clinical standard-of-care protocols for cancer treatment as well as generic adaptive therapy, demonstrating how these results could be translated to support clinical decision-making."

The researchers demonstrated that interpretable treatment strategies could be extracted from the ‘black-box’ deep learning network, in a form which a clinician would be able to understand and prescribe to their patients.

He added: "Interpretability has long been a significant hurdle to integrating machine learning approaches into clinical practice.

"When these frameworks are a black box, and we can’t understand how they derive treatment recommendations, we can’t be confident applying these in the clinic.

"But our new study shows that this hurdle can be overcome."

The researchers proposed using the deep reinforcement learning model to develop treatment plans for patients new to a particular drug regime, by creating a 'virtual twin' based on their initial treatment data.

This method was robust to changes or uncertainty in both the patient’s treatment response and the time interval between treatments, crucial for the real-world application of this approach.

The researchers are planning further studies to refine this method and explore its application to other forms of cancer.

The study ‘Mathematical Model-Driven Deep Learning Enables Personalized Adaptive Therapy’ has been published in Cancer Research.