OpenAI has joined forces with healthcare startup Color Health to embed GPT-4 into an AI-powered “copilot” that assists doctors in developing personalized cancer care plans.
The copilot, developed by Color Health, leverages OpenAI’s models to analyze patient data, including personal risk factors and family history, alongside clinical guidelines.
By identifying missing diagnostic tests and generating bespoke screening and pretreatment plans, the AI assistant supports healthcare providers in making evidence-based decisions.
“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.
The copilot’s potential to streamline cancer care is noteworthy, as screening, diagnosis, and treatment delays can have severe consequences for patients.
Studies show that a month’s delay in treatment can increase mortality by 6% to 13%.
Color Health’s trial of the copilot has already demonstrated promising results in lowering that figure. Clinicians can analyze patient records in an average of just five minutes, compared to the weeks it can take without the AI assistant.
“I’ve witnessed the complexities of developing personalized cancer screening plans for my high-risk patients,” says Dr. Keegan Duchicela, a primary care physician at Color. “The guidelines are constantly evolving, and individual risk factors aren’t always immediately clear.”
OpenAI and Color Health began collaborating in 2023, aiming to use AI to improve cancer patient care and health equity. They call it a “clinician-in-the-loop workflow,” which essentially supports clinical decision-making without in any way replacing it.
“We see a perfect fit for AI technology, for language models, because they can really help on every one of those dimensions,” said Brad Lightcap, OpenAI’s chief operating officer (COO).
“They can bring in relevant information to the surface faster. They can give clinicians more tools to understand medical records, to understand data, to understand labs and diagnostics.”
To measure the impact of the copilot, Color Health is partnering with the University of California, San Francisco Helen Diller Family Comprehensive Cancer Center (UCSF HDFCCC).
The partnership will conduct layers of evaluation, followed by a targeted rollout, with the potential to integrate the copilot into clinical workflows for all new cancer cases at UCSF.
“UCSF is a leader in implementing cutting-edge technology to improve patient care,” says Dr. Alan Ashworth, President of the UCSF HDFCCC.
“Patients frequently come to primary oncologists with incomplete diagnostic workups, and the time it takes to collate and accurately identify the completion of those workups prevents providers from working at the top of their license. 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.”
Color Health intends to slowly roll out the copilot, starting with an initial phase-in for its own clinicians and applying several layers of quality assurance.
Through the second half of 2024, the company intends to use the copilot application to provide AI-generated personalized care plans, with physician oversight, for over 200,000 patients.
AI’s promise in combating disease
The key here is using language models to democratize specialist knowledge. It’s become far easier to fine-tune models for different purposes, enabling researchers to build domain-specific models with medical data.
Similarly, in eye health, a chatbot was designed to provide clinicians with information on retinal issues and glaucoma. It met or exceeded expert advice in responding to clinical questions, again showing how AI can democratize specialist knowledge.
AI’s promise in detecting, diagnosing, and treating diseases, including cancer, is otherwise well-established.
AI tools are beating doctors at identifying complex forms of cancer and accelerating drug discovery and AI-identified drugs are even heading for clinical trial.
