AI Tool PERCEPTION Predicts Cancer Therapy Response at Single-Cell Resolution

Key Insights

• A novel AI tool, PERCEPTION, leverages single-cell transcriptomics to predict patient response to cancer drugs, offering a personalized approach to treatment.
• PERCEPTION uses transfer learning to overcome limited clinical single-cell data, pre-training models with bulk-gene expression data from tumors.
• Validated in multiple myeloma, breast, and lung cancer clinical trials, PERCEPTION accurately stratified patients into responder and non-responder categories.

An artificial intelligence (AI) tool named PERCEPTION can predict patient responses to cancer therapy by analyzing information from individual cells within a tumor. This new approach, published in Nature Cancer, leverages single-cell transcriptomics to offer a more personalized approach to cancer treatment.

How PERCEPTION Works

PERCEPTION, short for PERsonalized Single-Cell Expression-Based Planning for Treatments in Oncology, uses transfer learning to overcome the challenge of limited clinical single-cell data. The AI model is pre-trained using bulk-gene expression data from tumors and then fine-tuned with single-cell data from cell lines and patients.

Sanju Sinha, PhD, assistant professor at Sanford Burnham Prebys and first author of the study, emphasized the tool's ability to monitor the emergence of resistance. "The ability to monitor the emergence of resistance is the most exciting part for me. It has the potential to allow us to adapt to the evolution of cancer cells and even modify our treatment strategy," Sinha said.

Validation in Clinical Trials

PERCEPTION was validated by predicting the response to monotherapy and combination treatment in three independent clinical trials for multiple myeloma, breast cancer, and lung cancer. In each case, the AI tool correctly stratified patients into responder and non-responder categories. Notably, in lung cancer, PERCEPTION captured the development of drug resistance as the disease progressed.

Future Directions

While PERCEPTION is not yet ready for clinical use, the researchers hope that it will encourage the adoption of single-cell information in clinics to generate more data, which can be used to further develop and refine the technology. "The quality of the prediction rises with the quality and quantity of the data serving as its foundation," Sinha noted. "Our goal is to create a clinical tool that can predict the treatment response of individual cancer patients in a systematic, data-driven manner."

The researchers aim to spur more data and studies to accelerate the development of this technology for clinical use.