Leveraging large language models to assist with differential diagnosis and clinical plan drafting
Today, Glass Health is thrilled to announce our $5 million seed round, led by Initialized Capital, for our AI-powered clinical decision support platform. Our new funding will accelerate our mission to empower clinicians with generative AI tools that will help end the epidemic of clinician burnout and improve patient outcomes worldwide.
- Parul Singh, Partner, Initialized Capital
We’re proud to count prominent healthcare founders as mentors and investors in this round including Tom Lee, founder of One Medical, Epocrates, and Galileo Health; Connor Landgraf, founder and CEO of Eko Health; and Heather Hasson & Trina Spear, the co-founders of FIGS.
- Heather Hasson & Trina Spear, Co-founders, FIGS
While large language model applications like ChatGPT have encoded impressive medical knowledge as evidenced by their ability to pass the medical licensing exam, we believe that these powerful models must be fine-tuned by physicians and utilize up-to-date, evidence-based, and peer-reviewed clinical recommendations to be truly useful in the clinical environment. We use a technique called retrieval augmented generation to connect a large language model to our database of clinical guidelines created and maintained by the Glass Health Clinical Team of academic physicians.
Our approach is different than others in that we have specifically designed Glass to be doctor-facing rather than patient-facing so that the clinician can closely supervise and review the LLM’s outputs before applying them. Additionally, at this early stage of development and implementation of AI in healthcare, we have designed Glass to produce options that a clinician can consider along with other clinical reference resources rather than directives that would replace a clinician's judgment.
Notably, our retrieval augmented generation strategy also gives us an opportunity to have physicians continually edit and review the clinical guidelines used by our AI system. This gives us the power to keep the recommendations used by our AI system up-to-date and reflective of state-of-the-art management practices. It also allows us to review the guidelines we provide to the AI system for bias. Consistent with our commitment to achieving health equity, shared with the broader medicine community, we currently filter out social identifiers that should not be considered in clinical reasoning. In the future, we plan to offer health equity alerts in our outputs that tell a clinician when healthcare disparities might affect their patients.
We are engaged in efforts to expand the frontiers of generative AI in healthcare through integration with electronic medical records and health system-specific guidelines. Our team works tirelessly to deliver value to our clinician users, and several new features that will deliver tremendous value for clinicians around the world are actively in development.
We extend our heartfelt gratitude to our investors, including Breyer Capital, Y Combinator, and Initialized Capital, our team members, and our clinician users for their unwavering support of our mission.
You can follow Glass on Twitter here.
To contact Glass regarding partnerships with your hospital, health system, or medical training program, please email firstname.lastname@example.org.