Cancer is not made up of identical cells. Tumors often contain a mix of different cell types, each behaving in its own way. This complexity, known as cellular heterogeneity, makes it difficult to predict how a patient will respond to treatment and is a major reason why some therapies fail.

Researchers Northwestern Polytechnical University in China describe a new concept called the in silico patient, which uses computational models to simulate how an individual’s tumor may respond to different drugs. This approach combines multiple types of data, with single cell RNA sequencing playing a central role.

Single cell RNA sequencing allows scientists to measure gene expression in individual cells within a tumor. This provides a detailed view of the tumor ecosystem, including cancer cells and surrounding cells in the tumor microenvironment. By understanding how these different cell populations behave, researchers can gain insight into how a tumor might respond to therapy.

The in silico patient framework brings together several key components. First, it uses large single cell atlases as reference maps to identify different cell types. Second, it incorporates pharmacogenomic databases, which contain information about how cells respond to various drugs. Finally, it integrates patient specific RNA sequencing data to make personalized predictions.

Artificial intelligence is a critical part of this system. Deep learning models can analyze complex patterns in gene expression and apply knowledge learned from experimental models to real patient data. Techniques such as transfer learning help bridge the gap between laboratory studies and clinical applications.

Another important aspect is the inclusion of the tumor microenvironment, which consists of immune cells, stromal cells, and other components that influence tumor behavior. The framework also uses advanced experimental models, such as tissue slice cultures, to quickly test predictions and refine the models in a continuous cycle of prediction, validation, and optimization.

This approach represents a step toward more personalized cancer treatment. Instead of relying on general treatment guidelines, researchers aim to predict how a specific patient’s tumor will respond to different therapies before treatment begins.

While challenges remain, including integrating additional data types such as spatial and multi omics information, the concept of in silico patients highlights how combining RNA sequencing with artificial intelligence could improve decision making in precision oncology.