Microsoft is testing a new product that coordinates AI agents for multi-disciplinary care teams.
In the new era of agentic AI, Microsoft is creating technology to help AI agents work together and integrate them into everyday workflow tools like Microsoft Teams and Microsoft Word.
Its initial framework is designed to study the healthcare AI agent orchestrator for assisting tumor boards, a group of multi-speciality oncologists that review tumors to determine treatments for patients.
The tech company eventually wants to take the technology further by expanding across healthcare use cases and supporting multi-disciplinary teams, according to a blog post by Matthew Lungren, M.D., Microsoft’s chief scientific officer of Health and Life Sciences.
The healthcare agent orchestrator is available through Microsoft’s Azure AI Foundry Agent Catalog. The orchestrator has pre-configured agents, such as a radiology agent, a pathology agent, a cancer staging agent, a clinical guidelines agent and a clinical trials agent.
The agents are customizable, including through open source customization options. The tool can perform multi-agent orchestration and coordinate multimodal data workflows.
Joshua Warner, M.D., a radiologist at the University of Wisconsin Health, said in a statement that the agent orchestrator will help “surface, summarize and take action on relevant multimodal medical information” to decrease the time of tumor board review from hours to minutes.
“As clinical care complexity escalates, the healthcare agent orchestrator empowers developers to confidently navigate the accelerating era of agentic AI, collaborate with clinicians, and democratize precision medicine tools by surfacing these capabilities into existing workflows,” Lungren wrote.
The agent orchestrator is in use by cancer teams at Stanford University, Johns Hopkins, Providence Genomics, Mass General Brigham and the University of Wisconsin School of Medicine and Public Health.
“Stanford Medicine sees 4,000 tumor board patients a year, and our clinicians are already using foundation model-generated summaries in tumor board meetings today (via a PHI safe instance of GPT on Azure),” Mike Pfeffer, M.D., chief information officer at Stanford Health Care and Stanford School of Medicine, said in a statement.
Pfeffer continued: “The new healthcare agent orchestrator has the power to streamline this existing workflow by reducing fragmentation (saving time by avoiding copy-pasting) and enables surfacing new insights from data elements that were challenging to search, such as trial eligibility criteria, treatment guidelines, and real-world evidence. Stanford Health Care is excited to further research the potential of using the healthcare agent orchestrator to build the first generative AI agent solution used in a production setting for real-world care for our cancer patients.”
The agentic AI orchestrator will also allow integration of approved third-party agents. Paige.ai is shipping its agent Alba into preview, which offers real-time, conversational pathology insights. Microsoft says it's the first time a third-party agent has been integrated into an agent orchestrator.
“As we progress towards the routine use of multi-agent systems, the healthcare agent orchestrator demonstrates the power to simplify the integration of various models and agents with productivity tools that clinicians are already using,” Paige.ai CEO Razik Yousfi said in a statement. “The flexible orchestration framework will make it easy for us at Paige to continue to focus on our pathology agents while enabling their integration into the larger cancer care workflow and leverage access to multi-modal data.”