Color Health, a genetic testing company, is using OpenAI's newest, less expensive, large language model to equip doctors with pretreatment workup expertise that could speed up prior authorization requests for cancer screening diagnostics and get patients into treatment faster. 

The company has also partnered with the University of California San Francisco to study how the cancer copilot tool performs in surfacing early warning signs, seemingly incongruous red flags and other pertinent details that may be deeply dispersed throughout electronic health records and other patient information. 

WHY IT MATTERS

While decision factors for different types of cancers vary, a trial of the technology helped providers analyze patient records in five minutes, according to the company.

"Primary care doctors don’t tend to either have the time or sometimes even the expertise, to risk-adjust people’s screening guidelines," Othman Laraki, cofounder and chief executive of Color Health, said in a Wall Street Journal report Monday.

The UCSF Helen Diller Family Comprehensive Cancer Center is testing Color’s copilot for cancer pretreatment diagnostic work-ups by comparing it to retrospective analyses of cancer patient charts. 

Though that study is in the early stages, according to a Color Spokesperson, if AI can ultimately reduce wait times for cancer treatment by connecting the dots, that's a patient care win.

In Color's announcement Monday, Laraki said the company designed the tool to address the supply gap in oncology expertise to decide on a pre-treatment workup for a patient with a confirmed malignancy.

The goal is to offer primary care doctors and other clinicians an AI service that can determine what tests are needed to inform the patient's cancer treatment, without waiting for the patient to see an oncologist before pretreatment diagnostics are ordered and the prior authorization process is initiated, he explained.

"That way, by the time the patient meets her oncologist for the first time, she has a much higher chance of being ready to initiate treatment and, we hope, save weeks of precious time."

Laraki also stressed the clinician's role in decision-making when using the tool. 

"One of the most important design decisions behind our work is that the tools were built from the ground up to be based on a human-in-the-loop model," he said.

The company said it will share the results of the first use case tested – which focuses on automating the analysis of a person’s background risk factors and then applying the guidelines that adjust their screening plan – first with individuals in its cancer program, and then give primary care doctors a chance to review the information.

Color estimated that physicians using the cancer copilot will have supported more than 200,000 patient cases in generating AI personalized care plans by the end of the year.

THE LARGER TREND

Before focusing on tools to help doctors improve cancer patient outcomes, Color launched its model of patient-initiated proactive testing in 2015. The tests focused on genes known to increase an individual’s cancer risk, such as BRCA1 and BRCA2 for breast, ovarian cancer and pancreatic. 

Within a few years, the unicorn, along with 23andMe and other companies, shattered patient barriers to cancer screening not previously possible by offering low-cost, over-the-counter home test kits that could illuminate key genetic risk factors. 

Using AI for a new decision support service that empowers PCPs to get their patients with cancer into treatment faster is a budding area in healthcare AI where automating physician note-taking and reducing clinical administrative burden have made up the majority of mainstream LLM use cases. 

However, applying machine learning to health data is a major opportunity to enhance health outcomes for individuals and populations. 

AI could be instrumental in disease management, said Xin Wang, assistant professor in the University at Albany department of epidemiology and biostatistics.

"By analyzing patient data over time, AI algorithms can predict individual patient risks, suggest personalized treatment plans and even alert healthcare providers to early signs of complications," he told Healthcare IT News in January.

"This proactive approach can lead to earlier interventions, better disease management and, ultimately, improved health outcomes."

ON THE RECORD

"We see a perfect fit for AI technology, for language models," Brad Lightcap, OpenAI’s chief operating officer, said in the WSJ story. "They can give clinicians more tools to understand medical records, to understand data, to understand labs and diagnostics."

Andrea Fox is senior editor of Healthcare IT News.
Email: afox@himss.org
Healthcare IT News is a HIMSS Media publication.

The HIMSS AI in Healthcare Forum is scheduled to take place September 5-6 in Boston. Learn more and register.