Credit: VentureBeat made with Midjourney V6
Video AI generator models have been hogging the news cycle lately, but there are plenty of new initiatives happening on the large language model (LLM) side of AI, too.
Case in point: yesterday, OpenAI announced it had partnered with Color Health, an 11-year-old Burlingame, California startup, to launch a copilot app for cancer specialists to help them understand what tests to order up for their most at-risk patients.
The app uses OpenAI’s newest multimodal model GPT-4o to compare medical data of individual patients, and their risk factors, to the constantly changing list of healthcare guidelines — developing personalized cancer screening plans for said patients, and pinging doctors with what outstanding tests may still be needed.
What tests should which patients get, and how soon? The hope and early indications are that GPT-4o and OpenAI’s models more generally can help doctors answer these questions faster than they could have before, while maintaining high accuracy, ideally catching cancer or pre-cancerous conditions earlier and enabling for faster treatment and improved patient outcomes.
Doctors come forward in support of Color’s GPT-4o powered copilot app
In OpenAI’s post on the collaboration, Dr. Keegan Duchicela, a primary care physician at Color, highlighted the complexities of developing personalized cancer screening plans for high-risk patients. “I’ve witnessed the complexities of developing personalized cancer screening plans for my high-risk patients. The guidelines are constantly evolving, and individual risk factors aren’t always immediately clear.”
A Wall Street Journal story on the partnership mentions that a trail of the app found that doctors were able to analyze patient records in 5 minutes using Color’s GPT-4o-powered copilot app — a process that can sometimes take hours to weeks otherwise.
Dr. Allison Kurian, a professor at Stanford University School of Medicine, attested as much in OpenAI’s blog post, stating: “Many of my patients require weeks to complete all of the tests and evaluations necessary to provide appropriate treatment, during which precious time is lost and additional administrative burden is placed on clinicians.”
Cancer is the second leading cause of death in the United States behind heart disease, and a major driver of healthcare costs, so any effort to expedite and improve screening and treatment will no doubt be welcome.
What is Color Health?
Founded in 2013 by Othman Laraki, Elad Gil, Nish Bhat and Taylor Sittler as a kind of 23andMe genetic testing competitor to evaluate cancer risks, Color Health has since focused on overall cancer patient experience including sample collection, clinical support, and care to 7 million patients in all 50 states.
The company partnered with the American Cancer Society last year to provide comprehensive cancer care from screening and prevention to effective management and was mentioned as part of the White House’s “Cancer Moonshot” initiative in May 2024 for its plan, in conjunction with ACS, to offer free at-home colorectal cancer screenings for under- and uninsured individuals starting this month, enhancing early detection and reducing cancer-related healthcare costs.
It also helps patients find care for infectious and cardiometabolic diseases.
“Color’s vision is to make cancer expertise accessible at the point and time when it can have the greatest impact on a patient’s healthcare decisions,” said Othman Laraki, CEO of Color Health, in a statement posted on OpenAI’s official company blog.
A vision for accessible expertise
Through his X account, Laraki elaborated on the importance of timely access to healthcare, stating, “Healthcare being available doesn’t make it accessible. Making services and information accessible at the right time often outweighs anything else we can do.”
Through our work with 7mm patients & @AmericanCancer, we’ve seen the impact of access firsthand.
Simply put: Healthcare being available doesn’t make it accessible. Making services and information accessible at the right time often outweighs anything else we can do.
— Othman Laraki (@othman) June 17, 2024
Laraki further voiced what he saw as an opportunity for gen AI in healthcare: “Our bet is that the gen AI revolution is the biggest opportunity to improve access and immediacy to critical health information & services. Given the complexity and breadth of healthcare, a constant challenge is access to expertise at critical moments in a patient’s journey.”
Our bet is that the GenAI revolution is the biggest opportunity to improve access and immediacy to critical health information & services.
Given the complexity and breadth of healthcare, a constant challenge is access to expertise at critical moments in a patient’s journey.
— Othman Laraki (@othman) June 17, 2024
Proof-of-concept
Color Health’s collaboration with OpenAI began well before the release of GPT-4o last year, focusing on interpreting inconsistently formatted patient data, analyzing dense healthcare guidelines, and protecting patient data privacy.
OpenAI guided Color to prototype clinical workflows using the standard ChatGPT interface and generate sample cases using custom GPT models. This approach enabled Color to develop proofs of concept before it committed engineering resources to ideas that weren’t helpful or workable.
Laraki emphasized the focused approach on X, stating, “With new tech like gen AI, it’s tempting to fall into the trap of trying to boil the ocean. Rather than going broad, we went deep scaling the ability of clinicians to support a few specific, but highly impactful use cases/scenarios that are constrained by access to experts.”
With new tech like GenAI, it’s tempting to fall into the trap of trying to boil the ocean.
Rather than going broad, we went deep scaling the ability of clinicians to support a few specific, but highly impactful use cases/scenarios that are constrained by access to experts.
— Othman Laraki (@othman) June 17, 2024
Laraki also shouted out how the OpenAI-powered copilot app was unique from other gen AI medical approaches, linking doctors with patients rather than relying on the GPT-4o model to make assessments on its own.
3. We took a “copilot” product design approach. Our goal is for a regular doctor to not experience the tool as a new piece of software, but rather something that merges into their current workflow.
— Othman Laraki (@othman) June 17, 2024
Partnership with UCSF for Impact Evaluation
While early results indicate that clinicians using the copilot can identify significantly more missing labs and imaging results and analyze patient records much faster than those without the copilot, Color Health and OpenAI are keen to have outside experts evaluate the results of their collaboration.
To measure the impact of the new OpenAI GPT-4o powered copilot application, Color is partnering with the University of California, San Francisco Helen Diller Family Comprehensive Cancer Center (UCSF HDFCCC).
These medical experts will conduct a retrospective evaluation of the copilot’s recommendations, followed by a targeted rollout.
Dr. Alan Ashworth, President of UCSF HDFCCC, expressed enthusiasm for the collaboration in OpenAI’s blog post, stating: “We are interested in tools that can improve the efficiency and accuracy of pre-visit charting and avoid costly delays in treatment initiation for cancer patients at UCSF.”
Dr. Karen Knudsen, CEO of the American Cancer Society, praised the integration of AI technologies with clinical workflows. “The idea of combining AI technologies with digitally-enabled clinical workflows to expedite that process would be a positive advancement for all parties involved – the patient and their clinicians, as well as the payer covering the cost of treatment,” she stated.
Looking ahead
Ultimately, if all keeps going well with the initial trails, Color plants to expand the copilot app to more than 200,000 patients throughout 2024.
Laraki underscored Color’s commitments on X, writing: “Our top priority is quality and safety, which results in key decisions like: Clinician-in-the-loop model, use cases that are deeply guidelines-based, and collaboration with UCSF on validation/phase-in.”
Color writes on its blog that physicians or healthcare professionals who are interested in the copilot can reach out via email at: [email protected].
VB Daily
Stay in the know! Get the latest news in your inbox daily
By subscribing, you agree to VentureBeat’s Terms of Service.
Thanks for subscribing. Check out more VB newsletters here.
An error occured.
