Ci4CC Initiative Invitation

"Onco-LLM"

Multi Center Oncology AI Initiative & Collaboration

Building The Nations first Federated Clinical Trials LLM

Designing a Community-Driven, Oncology-Focused Large Language Clinical Trials Language Model and Retrieval Pipeline

Project Launch June 2023

The bulk of clinical data in the field of oncology resides within clinical notes, making it challenging to analyze and interpret on a large scale. Currently, manual abstraction remains the prevailing and most accurate method for extracting and quality-controlling clinical information from these notes. However, as we are all aware, this approach is labor-intensive, costly, and lacks scalability,

thereby constraining the full potential of advanced clinical research.

In response to this challenge, Triomics, in collaboration with the Cancer Center Informatics Society (Ci4CC), has embarked on a groundbreaking initiative: the development of an oncology-focused Clinical Trials Language Model (LLM) designed to revolutionize the management of cancer data in the context of clinical research. ("COLT", the Collaboration for Oncology focussed LLM Training.)  As part of this endeavor, Triomics is training a Language Model with over 30 billion parameters using structured and unstructured oncology datasets.

While the scale of this model is indeed impressive, its true strength lies in its precision. By concentrating exclusively on oncology and being trained on clinical datasets, it promises to bring about a significant transformation in clinical research—a feat that generic Language Models currently cannot accomplish. This initiative was recently launched by Sarim Khan (Triomics) and Sorena Nadaf-Rahrov (Ci4CC) at the 2023 AACI-CRI National Forum.

In the spirit of genuine collaboration, we extend an invitation to interested institutions to participate in various capacities: aiding in the construction of the foundational dictionary, contributing de-identified datasets for training and alignment, and more. Participating institutions will gain complimentary customized access to this Language Model for internal use. However, due to the complexity of the project, the number of collaboration opportunities is limited.

If you wish to learn more and join the initiative, please fill out the form below.  Please highlight "Onco-LLM" in the Subject Line