Color Health taps OpenAI to generate screening plans for cancer patients

Color Health is working with OpenAI to test out computer-generated personalized care plans for cancer patients.

The two began working together in 2023 with the goal of using artificial intelligence to improve patients’ access to cancer care. Color developed a new copilot app that leverages GPT-4o to identify missing diagnostics and create tailored workup plans for patients. It relies on APIs to integrate patient data with evidence-based healthcare guidelines and is HIPAA compliant.

The AI model uses patient data like family history and individual risk factors, plus guidelines, to understand what diagnostics might be missing and to generate a personalized screening plan. It also generates documentation necessary to complete any diagnostic workups, such as insurance pre-authorizations. 

“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,” Othman Laraki, CEO of Color Health, said in an announcement. "As a healthcare company, technology that improves access and equity has to go hand-in-hand with technology that supports patient safety and privacy.”

One of the appeals of the model, according to Color, is its ability to extract information that might be buried in pages of inconsistently structured information. It also analyzes clinical guidelines and data from trusted sources to build out the personalized plans, though the data sources were not specifically identified. The copilot's recommendations can be analyzed by a clinician at every step, which will also help refine future iterations, they said. 

Color has served more than 7 million patients since its founding in 2015. It initially started with a focus on gene testing and precision genomics, eventually pivoting to healthcare delivery programs that address key infrastructure systems in the U.S. Those include vaccination and preventive health services and infectious disease management programs.

Last year, the organization partnered with the American Cancer Society to help employers and payers tackle the disease—the second most common cause of death in the U.S. Treatment for cancer is notoriously complicated and screening needs are often highly individualized, Color argues. More than a third of its patients may require earlier or different screening approaches, based on individual risk factors that are not addressed by standardized guidelines. 

“I've witnessed the complexities of developing personalized cancer screening plans for my high-risk patients,” Keegan Duchicela, M.D., a primary care physician at Color, said in the announcement. “The guidelines are constantly evolving, and individual risk factors aren't always immediately clear.” 

To measure the impact of this tool, Color is partnering with the University of California, San Francisco Helen Diller Family Comprehensive Cancer Center. Together, they will conduct a retrospective evaluation following the initial implementation. There may be an opportunity to integrate the copilot into clinical workflows for all new cancer cases at UCSF based on the results. 

Color has started an initial phase-in for its clinicians, applying the tool to a limited number of cases. So far, providers using the copilot are able to identify four times as many missing labs, imaging or biopsy and pathology results than those without the copilot. Using the app also takes on average five minutes for clinicians to analyze patient records and identify gaps. 

Through the second half of 2024, Color intends to use the app to provide AI-generated personalized care plans with physician oversight for more than 200,000 patients.