In research, I'm not just a PI, I'm a Patient-turned-Researcher with firsthand experience to the conditions that we study in our lab. After years of trying to understand certain conditions that I have lived with, I came to realize that the best way to understand and provide better solutions to the conditions that I interface with was to take a comprehensive approach to research that crosses the boundaries of disciplines. As such, I'm inspired by the vast amount of data collected by scientists over the world and work to find new insights into their data by combining and crossing data from a variety of different sources and disciplines.

-- Sam Keating, Principal Investigator 

What We Do:

(1) We use LLM era tools as the methodology powering text summarization (primarily focused on biomedical literature, with the biggest emphasis placed on populating the metadata columns of the human readable table that we discussed)

(2) We have Natural Language Processing (NLP) frameworks separate from LLMs conduct:

(2a) extractive summarization (and ExtractiveQA, extractive question answering) over abstractive summarization

(2b) Parts of Speech tagging (NLP POS), which is capable of summarizing by highlighting existing text (without having to synthesize or generate any new text, bypassing hallucination risk by lacking said propensity entirely)

(2c) Temporal Knowledge Graphs (TKGs)

(3) We implement biomedical publication specialized text mining tools: emphasizing deployment and pipeline building, rather than starting a new codebase from scratch (see some key terms in this literature including GROBiD and Bio-RED, for starters)