Projects

My ongoing funded projects include:

• Extended Methods and Software Development for Health NLP (NIH NIGMS R01GM114355; funded since 2016), in which we are designing machine learning algorithms that leverage large collections of text to improve information extraction tasks such as mapping text to ontologies and discovering links between health events.
• Learning science concepts through metaphor comprehension, production, and conversation: Behavioral, neural and artificial intelligence measures (NSF BCS 2140897; funded since 2022), in which we are using language models to help students generate metaphors and improve their science knowledge.

My past funded projects include:

• Temporal Relation Discovery for Clinical Text (NIH NLM R01LM010090; funded 2010-2023), in which we were designing machine learning algorithms to extract timelines from clinical text and integrate those with structured data from the electronic medical record.
• Using Natural Language Processing to Determine Predictors of Healthy Diet and Physical Activity Behavior Change in Ovarian Cancer Survivors (NIH NCI R21CA256680; funded 2021-2022), in which we were designing machine-learning algorithms to analyze conversations in behavioral interventions, with the goal of improving patient outcomes by improving how patients are coached.
• A Data Science Platform and Mechanisms for Its Sustainability (NSF SMA RIDIR 1831551; funded 2018-2022), in which we were designing information extraction algorithms to make it easier to search 50 years of environmental policy documents.
• Global Reading and Assembly for Semantic, Probabilistic World Models (DARPA W911NF-18-1-0014; funded 2017-2022), in which we were designing machine learning algorithms to infer from text the times and locations over which a causal relation is valid, with the goal of modeling complex interactions in domains like food security.
• Voice Assistant for Data Entry and Recording (VADER; USAF SBIR FA8649-21-P-0834; funded 2021-2022), in which we were designing speech-to-text algorithms to assist aircraft maintainers in logging their activities.
• Automated Domain Adaptation for Clinical Natural Language Processing (NIH NLM R01LM012918; funded 2018-2021), in which we were designing algorithms that can improve machine learning models trained in one medical institution with data from another institution without any need for sharing of patient data.