NSF Funds Digital Twins, AI, ML Research

To strengthen the connection between the statistical community and National Science Foundation (NSF), we continue the series introduced in the May 2023 issue that poses questions to NSF program officers and awardees. This month, we interview Snigdhansu (Ansu) Chatterjee, whose team was awarded a grant to study neurodegenerative diseases through a new program: Foundations for Digital Twins as Catalyzers of Biomedical Technological Innovation (FDT-BioTech), which is a partnership between NSF, the National Institutes of Health, and the US Food and Drug Administration. If you have suggestions for future Q&As, with either awardees or NSF program officers, send them to ASA Director of Science Policy Steve Pierson.

Snigdhansu (Ansu) Chatterjee

Snigdhansu (Ansu) Chatterjee is the Sinha Ennovate Endowed Chair Professor in the department of mathematics and statistics at the University of Maryland Baltimore County. His research interests include theoretical foundations and explainability of machine learning and artificial intelligence; digital twins; representativeness, fairness, and ethics in artificial intelligence-machine learning (AI-ML) problems; Bayesian and other conditional inferential techniques; and applications of data science techniques in domains such as surveys and small area problems, precision medicine, and climate change.

Snigdhansu Chatterjee was awarded a grant to study neurodegenerative diseases using digital twin modeling. Neurodegenerative diseases (e.g., Alzheimer’s disease, Parkinson’s disease, multiple sclerosis) impact millions of people and result in hundreds of thousands of deaths annually. Digital twin modeling might yield new insights into these diseases and revolutionize their treatment and prevention.

As the principal investigator, Chatterjee will address multiple research problems at the interface of digital twin modeling using AI techniques and substantial amounts of biomedical data on neurodegenerative diseases. The data science topics broadly encompass manifold learning; feature discovery and selection; data assimilation; conditional statistical inference; and verification, validation, and uncertainty quantification of digital twins. He will also address the ethical, legal, and social implications of using digital twin models in the context of health care.

Chatterjee’s interdisciplinary, multi-institution team consists of five investigators, a post-doc, and several students, including the following:

• Animikh Biswas (UMBC, Mathematics and Statistics)
• Karuna Joshi (UMBC, Information Systems)
• Christophe Lenglet (University of Minnesota, Radiology)
• Asim Dey (Texas Tech, Mathematics and Statistics)

The grant amount is just short of $1 million, which was the budgetary cap for grants from this solicitation.

With your grant funded under an NSF Division of Mathematical Science solicitation started in spring 2024, did you approach the proposal writing differently than you might have approached a proposal to an ongoing solicitation?

Fortunately, I have been interested in interdisciplinary research on data science + X, where X is one of several topics in natural or social sciences, for quite some time. So, it was fun to write a proposal combining multiple topics I am excited about.

What advice do you have for others applying for NSF funding?

I suggest the actual research project is something you are excited about and the project description reflects your excitement yet conveys the necessary scientific information. It must be credible in terms of what is promised, not just incremental or routine work. Pay attention to the broader impacts of what you propose. Integrate the parts of the project in a cohesive narrative; program managers and grant reviewers are experts and can easily identify proposals that have not been thought out carefully.

I am a statistician. Why should I be interested in digital twins?

Engineering and other disciplines have used digital twins for the last couple of decades, although this is a relatively new topic in the statistics and data science community. There is tremendous scope for new statistics and machine learning research in the context of digital twins and the potential to apply such models to many practical problems. We should be on board with interesting developments in other disciplines.