Color Health has partnered with artificial intelligence (AI) company OpenAI to develop a new copilot application for enhancing cancer screening and treatment.

Powered by OpenAI’s GPT-4o model, this copilot is designed to assist healthcare providers in making evidence-based decisions by detecting missing diagnostics and creating customised workup plans.

The first use case of Color’s copilot tool focuses on automating the understanding and analysis of a person’s background risk factors.

It then applies guidelines to adjust its screening plan based on individualised circumstances. The tool is intended to support doctors by managing patients into treatment and bringing necessary test determination expertise to primary care settings or nurse oncologists.

The copilot extracts and processes patient data, such as family history and individual risk factors, to identify missing diagnostics and create a tailored screening plan.

It also prepares documentation for diagnostic workups, including medical necessity documents and insurance pre-authorisations.

A clinician-in-the-loop assesses the copilot’s output and can edit it, which in turn refines future iterations. Once satisfied, the clinician can integrate the information into the patient’s treatment plan.

Color’s alliance with OpenAI started in 2023, with the aim to enhance cancer patient care and promote health equity through AI.

OpenAI claims that healthcare providers using the copilot can identify four times more missing labs, imaging, or biopsy and pathology results than those without the tool.n

The company further notes that the copilot reduces the time taken to analyse patient records and identify gaps to an average of five minutes, compared to the weeks of delay when data is fragmented without the tool.

To measure the tool’s impact, Color is collaborating with the University of California, San Francisco Helen Diller Family Comprehensive Cancer Center for a retrospective evaluation and a targeted rollout. This could lead to the copilot’s integration into clinical workflows for all new cancer cases at UCSF.

Color is cautiously rolling out the copilot, initially applying it to a limited number of cases with its own clinicians. By the second half of 2024, Color aims to use the copilot to provide AI-generated personalised care plans, with physician oversight, for more than 200,000 patients.