Denis Newman-Griffis, PhD
Lecturer in Data Science
Information School
University of Sheffield
I'm a researcher interested in language and information, and how we can use computers to study both of them.
As a Lecturer (Assistant Professor) at the University of Sheffield, I study the practical effectiveness and responsible design of AI technologies for medicine and health.
At the University of Pittsburgh, I worked on defining what's involved in building practical natural language processing (NLP) methods for health and disability information.
At the Ohio State University, I studied representation learning technologies for modeling the semantics of new and emerging information domains.
At the National Institutes of Health, I led the development of NLP technologies for function and disability information to help support the US Social Security Administration's disability benefits process.
Statement of Positionality: I am a White, queer, non-binary, non-disabled person from a middle class background.
Pronouns: they/them
News
Videos
Research Interests
I believe the most exciting science is interdisciplinary. My work draws on artificial intelligence, medical informatics, linguistics, clinical medicine, and more to build systems that help real people ask diverse questions about data.
Function and Disability Informatics
Disability in one form or another is a near-universal experience, but it hasn't been a traditional focus area for medical informatics. I study the language we use to talk about functioning and disability and build NLP systems to turn information about function or the impact of disability into actionable data.
Translational NLP
NLP technologies are designed to connect people to information. All too often, the gap between a groundbreaking innovation in basic NLP methods and successfully using that innovation in practice means that new advances get lost in the shuffle and important research opportunities are lost. I study the processes of translating between innovation and application and finding the rich research questions in messy, real-world problems.
Corpus Linguistics meets Representation Learning
Computational tools can help us analyze superhuman amounts of text and capture linguistic patterns in oceans of data. I'm interested in turning NLP technologies, particularly for representation learning, into a lens to ask: what can I learn about this text and the people who wrote it?
Recent Publications
Information Extraction Framework for Disability Determination Using a Mental Functioning Use-Case
Ayah Zirikly, Bart Desmet, Denis Newman-Griffis, Elizabeth E Marfeo, Christine McDonough, Howard Goldman, Leighton Chan
JMIR Medical Informatics (2022) 10(3): e32245
Challenges and opportunities for analyzing mental functioning with NLP in disability policy contexts.
Half the picture: Word frequencies reveal racial differences in clinical documentation, but not their causes
Jacqueline Penn, Denis Newman-Griffis
Proceedings of the 2022 AMIA Informatics Summit
Medical documentation differs in style and content for White vs Black patients.
Digital Scarlet Letters: Sexually Transmitted Infections in the Electronic Medical Record
Sarah Bennett, Denis Newman-Griffis, Mary Catherine Beach, Marielle Gross
Sexually Transmitted Diseases (2022) 49(6):70-74
Ethical analysis of the documentation of STI history in pregnancy in the electronic medical record.
Linking Free Text Documentation of Functioning and Disability to the ICF with Natural Language Processing
Denis Newman-Griffis, Jonathan Camacho Maldonado, Pei-Shu Ho, Maryanne Sacco, Rafael Jimenez Silva, Julia Porcino, Leighton Chan
Frontiers in Rehabilitation Sciences (2021) 2:742702
New systems for linking information about Activities of Daily Living in clinical notes to the ICF.
Improving broad-coverage medical entity linking with semantic type prediction and large-scale datasets
Shikhar Vashishth, Denis Newman-Griffis, Rishabh Joshi, Ritam Dutt, Carolyn P Rosé
Journal of Biomedical Informatics (2021) 121:103880
New, modular approach for semantic type filtering to improve any biomedical information extraction pipeline.