A Florida State University associate professor, Michael Gubanov, has secured the NSF 2026 CAREER Award to build an advanced large language model focused on numerical data properties, aimed at enabling scalable analysis of large healthcare datasets for cancer research.

The project seeks to make AI in clinical practice more trustworthy by reducing numerical errors, with potential applications in personalized cancer treatment decisions such as chemotherapy and immunotherapy, while accounting for individual patient factors like prior treatments.

Support for the project comes from NSF, the Casey DeSantis Cancer Research Programs’ Florida Cancer Innovation Fund, and Amazon Web Services.

FSU officials and collaborators emphasize the significance of the award and Gubanov’s innovative approach to quantitative reasoning in AI, underscoring implications across medicine, government, finance, and science.

The NSF CAREER Award provides five years of support for early-career faculty to pursue research, training, and outreach, recognizing Gubanov’s potential to lead in research and education.

The project aims to develop an AI model that analyzes numeric data by understanding number order and magnitude, addressing current LLM limitations that rely on pattern recognition rather than computation, which leads to numerical inaccuracies.

Gubanov’s background includes a PhD from the University of Washington (2010), postdoctoral work at MIT, an endowed chair at the University of Texas at San Antonio, and joining FSU in 2018.

He collaborates with Moffitt Cancer Center to assess the model’s healthcare impact and its potential for personalized treatment recommendations, and he leads BigLab!, a bridge between AI, data science, and cancer research.