WALTHAM, Mass. — BostonGene announced that its artificial intelligence and machine learning capabilities have been independently validated in a blinded, multi-vendor study evaluating HER2 scoring in breast cancer, reinforcing confidence in the company’s clinical-grade AI platform.
The findings were published in the journal Modern Pathology in an article examining agreement across 10 artificial intelligence models assessing human epidermal growth factor receptor 2 expression in breast cancer whole-slide images. The study was conducted in collaboration with Friends of Cancer Research and supported by global pharmaceutical companies and patient advocacy stakeholders.
According to the company, the publication adds to a growing body of external evidence validating the technical rigor, performance, and real-world relevance of its AI-driven approach. By evaluating algorithms under an independent framework, the study mirrors the stringent benchmarks increasingly used by drug developers and regulators to assess advanced analytics technologies.
BostonGene said multi-site, blinded validation is critical for increasing confidence in AI-driven biomarkers and reducing risk in regulatory, clinical, and companion diagnostic development. The company noted that ecosystem-level evaluations such as this confirm its position among a select group of organizations operating at the highest levels of scientific and technical standards for clinical AI.
BostonGene’s platform is built on a foundation model of cancer and immune biology that integrates multiomic, RNA, DNA, T-cell receptor, spatial, and clinical data at scale. This multidimensional approach enables detailed characterization of tumors and the immune microenvironment, supporting decisions across the drug development lifecycle, from early target discovery and research to patient stratification and trial optimization.
As HER2 remains one of the most critical biomarkers in oncology, the company said the terminology and evaluation frameworks outlined in the publication are expected to influence how AI and machine learning tools are assessed by drug developers in the future.
“This is not an isolated result,” said Nathan Fowler, MD, chief medical officer at BostonGene. “We continue to see independent, external validation of the AI and ML algorithms that power our foundation model. These blinded, real-world evaluations provide the high-stakes certainty that drug developers trust when accelerating life-saving therapies.”
Pharmaceutical organizations supporting the study included AstraZeneca, Bristol Myers Squibb, Amgen, Merck, and GlaxoSmithKline. BostonGene said it continues to partner with these and other pharmaceutical companies on programs where its AI-driven insights inform biomarker strategy and clinical execution.
