Cancer cells are remarkably adaptable and able to make molecular changes that allow them to evade even the best-laid treatment plan. If you plan treatment protocols for patients, you’ll be interested in learning about this new machine learning approach that could help improve the targeting of your treatment regimen and avoid drug resistance.

With funding from NCI’s Cancer Systems Biology Consortium, researchers developed a deep reinforcement learning (DRL) framework that blends mathematical modeling with machine learning.

Using the DRL approach, the researchers were able to analytically model changes in a tumor’s dynamics alongside treatment timing. According to the authors, this approach allowed them to identify a clear treatment threshold so they could time the frequency of treatment to when it would be most effective.

In the future, this technology could create a virtual patient treatment model, allowing you to track tumor-medication interactions in real-time to give our patients the best care.