Arctoris collaborated with researchers at the University of Cambridge and King’s College London on a study demonstrating how artificial intelligence can identify novel drug combinations capable of killing cancer cells.
The research explored whether AI systems could move beyond data analysis to actively generate new therapeutic hypotheses. Using advanced computational approaches, the team identified combinations of existing, non-cancer drugs that could selectively target breast cancer cells while sparing healthy tissue.
These computationally generated hypotheses were then tested experimentally, creating a closed-loop system where in silico predictions informed laboratory experiments, and experimental results were used to refine subsequent predictions.
The Role of High-Quality Experimental Data
High-quality, consistent data is the foundation of AI and machine learning models. Traditional manual datasets are prone to handling errors and operator variability, which can obscure true biological signals.
This modular automation approach produces highly robust data, with Z’ values of 0.8 and higher regularly achieved,reproducible pIC50, and S:B metrics across iterative studies. By removing human error, our workflow empowers machine learning models to accurately identify synergistic effects, accelerating drug discovery and research.
More detailed scientific methodology and results from this work are described in the associated publication.