The tools include access controls and approval settings and are offered through its API and healthcare products.
Photo Credit: OpenAI
OpenAI is working with four healthcare organisations
OpenAI has introduced its computer-use technology to healthcare teams, where AI agents can work across electronic health records, insurance portals and other clinical systems. The agents can help staff prepare medical orders, submit prior authorisation requests and retrieve patient information from different platforms. OpenAI has also published the results of tests assessing how safely the agents handle clinical tasks and how well they perform administrative work. The company has shared examples of healthcare organisations using the technology, which is available through its API and ChatGPT products for clinicians.
OpenAI's computer-use agents are being tested for a range of administrative tasks in healthcare, from preparing medical orders to handling insurance paperwork, the company said in a blog post. The agents can work across existing electronic health record systems, insurance portals and other clinical software. OpenAI is working with healthcare organisations including Abridge, Ambience, Cigna Healthcare and CVS Health to bring the technology into their workflows.
To demonstrate how this could work, OpenAI used synthetic patient data to show an agent preparing for an MRI referral. It reviewed the patient's medical records, physical therapy notes and previous imaging before drafting an MRI order and a clinical note for the doctor to review. The agent also prepared an insurance authorisation request for a specialist to check.
CVS Health has been using the technology to manage referrals and send documents. One agent cleared a backlog of document-sending tasks that had accumulated over several months in a single weekend, OpenAI said. Cigna Healthcare is using the models to pull information from more than 10 clinical and administrative systems, helping care managers gather the details they need before reaching out to patients.
OpenAI also tested GPT-6 Astra on tasks commonly handled through electronic health record inboxes, including reviewing patient histories and preparing clinical documentation. In its evaluation, 99 percent of 325 tasks reviewed by clinicians were rated clinically safe. Doctors rated the model's responses very helpful or extremely helpful in 93 percent of 646 assessments.
On HealthAdminBench, a Stanford-developed benchmark for healthcare administrative tasks, GPT-6 Astra scored 94.3 percent, compared with 91.8 percent for GPT-6 Luna. OpenAI said GPT-6 Luna cost 8.5 cents per task, 98 percent less than GPT-6 Astra.
According to the company, healthcare organisations can control which applications and patient records agents access, restrict the actions they can take and require staff approval for certain tasks. ChatGPT for Healthcare also includes role-based access controls and audit logs. OpenAI said it does not use organisational data to train its foundation models.
The computer-use features are available through the OpenAI API, ChatGPT for Healthcare and ChatGPT for Clinicians. Existing customers can ask their workspace administrators to enable them in ChatGPT Work and Codex. Clinicians in the US can also sign up for ChatGPT for Clinicians.
Get your daily dose of tech news, reviews, and insights, in under 80 characters on Gadgets 360 Turbo. Connect with fellow tech lovers on our Forum. Follow us on X, Facebook, WhatsApp, Threads and Google News for instant updates. Catch all the action on our YouTube channel.
Apple Watch Series 12
Starts from ₹56,900
Samsung Galaxy Watch Ultra 2
Starts from ₹58,949
Samsung Galaxy Watch 9 (44mm, LTE)
Starts from ₹41,999
Samsung Galaxy Watch 9 (40mm, LTE)
Starts from ₹38,999
Samsung Galaxy Watch 9 (40mm)
Starts from ₹36,999
Samsung Galaxy Watch 9 (44mm)
Starts from ₹40,999