AI Multi-Agent Orchestration Drives Personalized Cancer Care

Strategic Insights -

Cancer care is a complex and ever-evolving field, with every patient requiring a unique treatment plan that takes into account their specific tumor type, genetic mutations, and medical history. However, the sheer volume of data and complex analysis required to create personalized care plans makes it challenging for clinicians to spend the necessary time and attention on each patient. The American Society of Clinical Oncology (ASCO) estimates that clinicians spend between 1.5 to 2.5 hours reviewing imaging, pathology slides, clinical notes, and genomic data for each patient. AI multi-agent orchestration holds the potential to reduce administrative friction and further transform care delivery.

Key Takeaways:

  • The healthcare agent orchestrator is a pre-configured agent designed for multi-agent orchestration and open-source customization options that allow developers and researchers to build agents that coordinate multi-disciplinary multimodal healthcare data workflows, such as tumor boards.
  • The orchestrator can manage analysis and reasoning over diverse healthcare data types, including imaging, pathology, genomics, and clinical notes.
  • The healthcare agent orchestrator provides tools that connect enterprise healthcare data through Microsoft Fabric and the Fast Healthcare Interoperability Resources (FHIR) data service, ensuring interoperability and integration into existing workflows.
  • The orchestrator leverages Semantic Kernel and Magnetic-One to coordinate agents, maintain shared memory, and interact with the human in the loop.
  • Researchers and developers at leading cancer care institutions, such as Stanford University, Johns Hopkins, and the University of Wisconsin School of Medicine and Public Health, are currently exploring the healthcare agent orchestrator.
  • The healthcare agent orchestrator is intentionally open-ended, allowing any approved agent to be pulled into a Teams conversational thread, enabling collaboration among clinical providers.
  • The orchestrator has the potential to significantly enhance efficiency and collaboration among clinical providers, providing real-time support to multidisciplinary care teams across the healthcare ecosystem.

Statistics:

  • Every year, 20 million people are diagnosed with cancer globally. (Source: Global cancer statistics 2022)
  • Clinicians spend between 1.5 to 2.5 hours reviewing imaging, pathology slides, clinical notes, and genomic data for each patient. (Source: Adapted Tumor Board Evaluation Tool for Quality Assessment of a Thoracic Multidisciplinary Cancer Conference)
  • The healthcare agent orchestrator can reduce manual work that can take experts over three hours to a mere minutes. (Source: Universal Medical Abstraction)
  • The radiology agent can analyze radiology images for a second read, improving recall by more than double the publicly available Critera2Query baseline. (Source: Scaling Clinical Trial Matching Using Large Language Models)

Sources:

  • Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries, CA: A Cancer Journal for Clinicians, April 4, 2024.
  • Using an Adapted Tumor Board Evaluation Tool for Quality Assessment of a Thoracic Multidisciplinary Cancer Conference: A Pilot Study, JCO Clinical Cancer Informatics, October 5, 2023.
  • Universal Abstraction: Harnessing Frontier Models to Structure Real-World Data at Scale, February 2, 2025
  • MAIRA-2: Grounded Radiology Report Generation, June 6, 2024
  • Nature Medicine, A foundation model for clinical-grade computational pathology and rare cancers detection, July 22, 2024
  • Scaling Clinical Trial Matching Using Large Language Models: A Case Study in Oncology, August 4, 2023