Archive · 27 Oct 2025
AI Model C2S-Scale Develops Cancer Treatment Hypothesis
Scientists have created an AI model, C2S-Scale, that understands the language of cells.
They trained it with lots of information about cells, like the Human Cell Atlas.
The AI found a way to make cancer cells easier for the body's defenses to spot.
It predicted that a drug called silmitasertib could help, which was confirmed in tests.
The AI model's idea was to make cancer cells show a signal when a certain immune protein is present.
This is a step toward new cancer treatments, but more research is needed to make sure it's safe and works well.
Google DeepMind's C2S-Scale AI model generated a new hypothesis about cancer cell behavior.
The model was trained on data from over 50 million cells to understand gene expression patterns.
C2S-Scale predicted the drug silmitasertib would boost cancer cell visibility to the immune system.
Lab experiments validated the AI's prediction, showing increased visibility in specific conditions.
This could accelerate drug discovery by allowing large-scale experiments inside computers.
- Who
- Google DeepMind and Yale University
- What
- Developed an AI model, C2S-Scale, which generated a hypothesis about cancer cell behavior, later confirmed in lab experiments.
- When
- Recently announced
- Where
- Lab experiments, with collaboration between Google DeepMind and Yale University
- Why
- To develop new cancer treatments by making cancer cells more visible to the immune system.
- AI Model
- C2S-Scale
- Developed by
- Google DeepMind
- Research Partner
- Yale University
- Model Architecture
- Gemma-2
- Parameter Count
- 27 billion
- Drug Identified
- Silmitasertib
- Experiment Type
- In vitro