Currently, oncology
staff must manually search thousands of patient health records to find the
right trials or care pathways for their patients. Triomics has announced that it has raised $15m to help cancer centres streamline these workflows and process technology data at scale by applying their framework to build, institution-tuned larger language models (OncoLLM) and use case-specific software.
The company has raised from several Silicon Valley firms making
pioneering investments in generative AI and healthcare, including Lightspeed,
Nexus Venture Partners, General Catalyst and Y Combinator.
Manual chart review
can take hours per patient, and many health systems face significant backlogs
in completing key oncology-related workflows for thousands of patients. This
workload leads to clinical delays, such as patients missing out on clinical trials
or biomarker-driven treatments, lagging quality reporting, and provider
dissatisfaction and turnover.
Triomics co-founders
Sarim Khan (CEO) and Hrituraj Singh (CTO) were college friends who later worked
as an MIT biotech researcher and Adobe AI researcher, respectively. They knew
software existed to quickly analyze the ~20% of medical data that is stored in
a uniform, structured manner, like a patient's demographics or
laboratory values. However, they realized recent advances in generative AI
created the possibility of similarly analyzing the ~80% of medical data that
exists in an unstructured format, like a doctor's free-text
note.
"Hrituraj and I decided to partner to build solutions leveraging the advances in the field of generative AI and LLMs to help hospital staff," said Sarim Khan, CEO of Triomics. "We want our solutions to reason and sound like experts in oncology."
After developing an
OncoLLM™ with Medical College of Wisconsin researchers, Triomics found that, in
just minutes, it found 90% of eligible patients for clinical trials, which
would have taken days or weeks for qualified nurses. It also extracted
structured data points from unstructured notes at similar or higher accuracy to
proprietary models like GPT4 or Claude while being 40 times cheaper. Triomics
recently also published the results of its information retrieval engine for
oncology, which they found to be 1.5-2 times better than other state-of-the-art
retrieval models.
"Most of the solutions on the market
today claim they use GenAI, but many lack published evidence. Triomics is
setting themselves apart by taking a truly collaborative approach to
co-developing these models," said Bradley Taylor, Chief Research Informatics Officer at the Medical College
of Wisconsin and Director of the CTSI Center for Biomedical Informatics.
Anai Kothari, a surgical oncologist at the Medical College of Wisconsin
Cancer Center added:
"The ability to quickly and
accurately convert complex cancer data into a format that can be used to
improve patient care is crucial. Triomics has been a great partner in
integrating our suggestions and rigorously studying their approach to ensure
safety."
OncoLLM™ powers
proprietary Triomics software that integrates with health system EHRs to
complete specific clinical and administrative tasks. For example, Triomics
Prism aids in patient-trial matching by prescreening oncology patients with
upcoming appointments to find relevant clinical trials. Triomics Harmony
curates EHR data to support quality reporting, cohort analysis and precision
oncology.
Hrituraj Singh, CTO at
Triomics,
commented: "Our investments in our core areas of focus have been deliberate.
We have successfully merged expertise in two complex functional areas: our AI
researchers, who are specialized in customizing language models to specific
domains, and our clinical staff, who have decades of oncology-specific
experience. As a result, our software can complement the strengths of these
advanced models while also proactively addressing potential flaws, all with the
intricacies of cancer research and care in mind."
Given the heightened
importance of accuracy for oncology data, Triomics partners with leading
academic cancer centers and researchers to develop generative AI performance
and safety benchmarks and best practices. Partners include the Collaboration
for Oncology-focused LLM Training (COLT), a consortium of leaders from a dozen
NCI-designated cancer centers, and the Cancer Informatics for Cancer Centers
(CI4CC) Society.
"We differentiate
ourselves by building tailored models specifically for oncology and pairing
them with GenAI native workflows," said Sarim Khan. "While other solutions address some of the use
cases we're working on, like patient-trial matching, they are broad based
solutions that use or modify legacy technologies that have proven not to have
the scalability or ROI the industry is requesting."
Triomics next plans to
publish additional data on OncoLLM efficacy across a diversity of settings and
patient populations, and develop software that powers additional use cases.
"Triomics
is leveraging existing healthcare datasets and Generative AI to empower
hospital staff to automate clinical trials and streamline cancer center
workflows," said Dev Khare,
partner at Lightspeed. "We are excited to back Triomics
in this important mission."
"With robust early results for their
proprietary oncology specific LLMs and partnerships with leading cancer care
and research centers, Triomics is well poised to deliver significant value to
cancer care providers and patients in the U.S. and globally," said Jishnu Bhattacharjee, managing director at Nexus Venture Partners. "We are
thrilled to partner with Sarim and Hrituraj to help build a remarkably
impactful company!"
