The Faculty of Medicine

Nano-informatics and Machine-Learning for the prediction of cancer cell behavior

Tumor diversity is a major reason why treatments often fail. Tools that can predict how cancer cells behave could greatly improve treatment and extend patients' lives. By studying how cancer cells interact with nanoparticles of different properties, we can indirectly measure the cells’ mechanical properties and classify them based on their activity. Machine learning (ML) was employed to analyze the complex patterns of particle uptake by individual cells.

In this innovative study, we discovered that human cancer cell subpopulations, which vary in drug resistance or malignancy, can be identified through particle uptake with over 95% accuracy using AI algorithms. This method is particularly important for distinguishing cancer cells that appear similar but function differently. Our study provides the basis for a novel big data type of information, referred to as “Mechanomics,” which offers clinically relevant insights that can help improve personalized therapy in cancer.

Photo: Yoram Aschheim

Prof. Ofra Benny and Yoel Goldstein in the lab