OpenAI Partners Color Health to Enhance Cancer Screening & Treatment Using AI

The AI copilot developed by Color Health employs OpenAI APIs, which developers use to access OpenAI models for their applications.

OpenAI has partnered with the California-based Color Health to advance AI use in healthcare, particularly in cancer screening and treatment.

This collaboration marks a significant step towards integrating AI to assist healthcare professionals and ensure timely and accurate diagnosis and treatment plans for patients.

The Wall Street Journal reported that Color Health leverages OpenAI's sophisticated language models to streamline various healthcare processes, ultimately aiming to improve patient outcomes and reduce the workload on medical professionals.

Founded in 2013, by Elad Gil, Nish Bhat, Taylor Sittler, and Othman Laraki, Color Health initially focused on genetic testing. The company has now developed an AI assistant, or "copilot," using OpenAI's GPT-4o model. This AI tool is designed to aid doctors in creating cancer screening plans and pretreatment strategies for patients diagnosed with cancer.

Othman Laraki, cofounder and CEO of Color Health, emphasized that the copilot is intended to assist rather than replace doctors, drawing parallels to how AI copilots support software engineers.

"We call it a copilot because it's very similar to the engineering copilot mindset and model. It's not like copilots replaced [software] engineers," Laraki explained.

Development & Purpose of the AI Copilot

OpenAI and Color Health began working on the copilot last year, culminating in its announcement on Monday. This initiative represents OpenAI's latest foray into the healthcare sector.

In April, OpenAI collaborated with Moderna to use AI to accelerate business processes, including optimizing doses for clinical trials.

Brad Lightcap, OpenAI's chief operating officer, highlighted AI's role in enhancing healthcare delivery efficiency by helping clinicians access and understand medical records, data, and diagnostics more quickly and effectively.

AI Copilot’s Key Attributes

The AI copilot developed by Color Health employs OpenAI APIs, which developers use to access OpenAI models for their applications. Like many developers, Color Health pays OpenAI based on the usage of tokens or word segments processed by its models.

The copilot uses patient data, such as personal risk factors, family history, and clinical guidelines, to create personalized cancer screening plans. This virtual plan helps doctors identify any missing diagnostic tests for their patients.

"Primary care doctors don't tend to either have the time, or sometimes even the expertise, to risk-adjust people's screening guidelines," Laraki noted, underscoring the utility of the copilot in managing routine yet critical tasks.

Addressing Administrative Burdens

The copilot also aids in assembling pretreatment "work-ups" once a cancer diagnosis has been made. These work-ups typically include specialized imaging, lab tests, and securing prior authorization from health insurance for the required tests, which can often take weeks or months.

Studies suggest that a month's delay in treatment can increase mortality by 6% to 13%, highlighting the importance of timely intervention. Karen Knudsen, CEO of the American Cancer Society, stressed the potential of AI to allow oncologists to focus more on patient care by alleviating administrative burdens.

"If this is going to help solve for gathering all the needed information for pre-auth, that will be a win for everyone, not just the patients, but also the clinical teams," Knudsen said.

Despite AI's capabilities, the complexity of the pretreatment work-up process means that it is not meant for complete automation. Numerous decision factors exist for various cancers, and doctors remain in control of the final outputs and decisions. Laraki emphasized that the copilot is a tool to assist, not replace, healthcare professionals in making informed decisions.

Trial & Future Prospects

In trials of the copilot, clinicians could analyze patient records in an average of five minutes.

Alan Ashworth, president of the University of California San Francisco's Helen Diller Family Comprehensive Cancer Center, mentioned that the facility tests the copilot for diagnostic work-ups, comparing retrospective analysis against prospective trials. Reducing the time to treatment by weeks is seen as a significant potential benefit.

Ashworth also highlighted that the most promising application of AI in healthcare is automating routine tasks like paperwork and physician note-taking. He noted that while AI models can "hallucinate" and contain biases, these risks necessitate the continued involvement of doctors in clinical decision-making.

Expanding AI's Role in Healthcare

Color Health's Laraki and OpenAI's Lightcap are firm that doctors must be involved in any clinical decisions using AI tools.

AI's future potential lies in its ability to analyze vast amounts of clinical and real-world data, aiding doctors in quickly identifying clues for even asymptomatic cancers. However, the technology has not yet reached a point where it can completely replace human judgment and expertise.

OpenAI formed a safety and security committee in May to address concerns following a legal dispute over a new voice assistant in its GPT-4o model.

However, customer projects like the one with Color Health fall under standard board oversight.

OpenAI has also recently expanded its reach with several new deals across various industries, including partnerships with Apple for AI functions and PricewaterhouseCoopers to resell its enterprise ChatGPT product.

OpenAI's Lightcap emphasized that the goal is for AI technology to be ubiquitous across different sectors.

"This is a technology that's going to be everywhere. That's kind of our mark for success," he said.