During his time as a medical student at UC San Francisco, Dereck Paul grew increasingly concerned about the lag in innovation within medical software when compared to other sectors like finance and aerospace. He came to the realization that patients would be better served if doctors had access to cutting-edge technology in the form of software. With this vision in mind, he aspired to establish a company that prioritized the needs of patients and healthcare providers over the interests of hospital administrators and insurance companies.
To turn this vision into reality, Paul partnered with his friend Graham Ramsey, an engineer at Modern Fertility, a women’s health tech company, and together they founded Glass Health in 2021. Glass Health introduced a unique tool for physicians, a digital notebook that allows them to store, organize, and share their diagnostic and treatment approaches throughout their careers. Ramsey described it as a “personal knowledge management system” tailored to the field of medicine.
Paul explained the motivation behind their venture, stating, “During the pandemic, Ramsey and I witnessed the overwhelming burdens on our healthcare system and the worsening crisis of healthcare provider burnout.” Paul had personally experienced burnout during his time as a medical student and later as an internal medicine resident physician at Brigham and Women’s Hospital. This experience, along with their empathy for frontline healthcare providers, inspired them to harness technology to enhance the practice of medicine.
Glass Health initially gained traction on social media, particularly on X (formerly Twitter), among healthcare professionals and those in training. This social following led to Glass Health’s first round of funding, a $1.5 million pre-seed investment led by Breyer Capital in 2022. Subsequently, Glass Health was accepted into Y Combinator’s Winter 2023 batch. However, earlier this year, Paul and Ramsey made the decision to pivot the company’s focus towards generative AI, aligning with the emerging trend in healthcare.
Today, Glass Health offers an AI-powered tool fueled by a large language model (LLM), technology similar to that used in OpenAI’s ChatGPT. This tool generates diagnoses and evidence-based treatment options for patients based on input from physicians. Clinicians can input patient descriptions, such as “71-year-old male with a history of myocardial infarction presents with subacute progressive dyspnea on exertion,” and receive likely prognoses and clinical plans from Glass Health’s AI.
Paul elaborated on the process: “Clinicians enter a patient summary, also known as a problem representation, that describes the relevant demographics, past medical history, signs and symptoms, and descriptions of laboratory and radiology findings related to a patient’s presentation. Glass analyzes the patient summary and recommends five to 10 diagnoses that the clinician may want to consider and further investigate.”
Furthermore, Glass Health can generate a case assessment paragraph for clinicians, complete with explanations about relevant diagnostic studies. These explanations can be edited and incorporated into clinical notes or shared within the Glass Health community.
While Glass Health’s AI tool appears highly valuable in theory, there are concerns about the reliability of AI-generated healthcare advice. Several instances have raised doubts about the effectiveness and safety of such systems. Paul acknowledged these concerns and emphasized that Glass Health’s AI is designed to provide potential diagnoses rather than definitive or prescriptive answers. This distinction helps mitigate legal scrutiny and potential regulation by the FDA.
Paul emphasized, “Glass connects LLMs with clinical guidelines that are created and peer-reviewed by our academic physician team.” This team comprises members from major academic medical centers who contribute part-time to Glass Health, similar to their involvement with medical journals, in creating and refining guidelines for the AI. Clinician users are encouraged to closely supervise the AI’s outputs, treating it as an assistant that offers suggestions but does not replace their clinical judgment.
To enhance its AI’s accuracy, Glass Health collects user data, although this approach may raise privacy concerns. Users have the option to request the deletion of their stored data at any time. Paul emphasized that Glass Health’s AI combines physician-validated clinical guidelines with AI context, differentiating it from LLM applications that rely solely on pre-training and may produce outdated or inaccurate medical information.
Despite these challenges and concerns, Glass Health has experienced substantial adoption, with over 59,000 users to date. The platform offers a direct-to-clinician subscription service, and in the coming year, Glass Health plans to pilot an enterprise offering integrated with electronic health records and HIPAA compliance, with 15 health systems and companies already on the waitlist.
With $6.5 million in funding, Glass Health intends to allocate resources to physician involvement in creating, reviewing, and updating clinical guidelines, fine-tuning its AI, and conducting general research and development. Paul expressed confidence in Glass Health’s four-year runway and its commitment to harnessing AI to advance healthcare while maintaining a strong focus on patient safety and the expertise of healthcare professionals.
