Large
Language Models
SeRP in collaboration with its partner tenant DPUK are exploring Large Language Models (LLM) to develop data discovery & project feasibility tools that will have great utility across the TRE’s within SeRP.
With so many different cohort datasets within TREs, sometimes with thousands of variables in each, finding the right datasets can often be challenging for researchers.
Our LLM innovations aim to streamline this process by enabling researchers to query our data with natural language to identify which datasets contain the data that is needed for a given project. It will also allow researchers to dig deeper into the data to find out whether there is enough participants with a certain disease, or if the data contains other modalities such as neuroimaging and genomics.
This new innovative approach allows researchers to go beyond the typical data discovery and metadata methods to allow a deeper understanding of the datasets before applying for access.