Healthcare organizations are increasingly focused on adopting artificial intelligence to enhance data utilization and understanding. This trend is expected to drive the integration of AI into more medical devices and robotics systems. In specific scenarios, these devices must process data locally, delivering immediate responses, which is key for real-time healthcare applications.
Healthcare organizations are looking to manage edge AI devices successfully and balance the need for edge analytics with the benefits of the cloud. David Niewolny, senior director and global head of business development for healthcare and medical at NVIDIA, notes that healthcare is a real-time, sensor-rich environment, making edge computing important. He explains that each individual medical device on the edge has its own unique computing architecture, from hardware and software to operating system and applications.
NVIDIA is working to democratize across different technologies, including emerging areas of spatial intelligence and robotics. A common platform is needed where AI-powered software offerings can live, ideally a cloud-native platform where data and applications can operate both at the edge and in the cloud. This platform will enable healthcare organizations to run AI applications where they are needed, whether at the point of care or in the cloud.
Edge-to-Cloud AI in Healthcare
Niewolny defines “edge-to-cloud AI” as the ability to run AI applications where they are needed, whether at the point of care or in the cloud. The exercise is to determine the workload, how fast a response is needed, and how sensitive the data is. Real-time perception, device control, alerts, and procedural support usually belong at the point of care on the edge. Institution-wide inferences, local model services, and large-scale training or simulation can generally be done in either a cloud or an on-prem data center.
Niewolny emphasizes the importance of considering latency, data residency, and network availability when deciding where to run AI workloads. The technology and governance foundation needed to deploy edge-to-cloud AI includes accelerated compute, reliable networking, storage, standardized containers, and application programming interfaces. Governance models are being developed, and regulatory agencies are outlining expectations for device manufacturers and software providers.
Healthcare organizations need to balance the need to keep data local with the benefits of a centralized AI model. Niewolny advises erring on the side of caution and keeping raw patient data on-site, while centralizing approved model artifacts, de-identified data, and aggregated telemetry data. When deploying edge-to-cloud AI, healthcare organizations should understand the different AI applications they plan to run and ask how edge compute is being delivered.
NVIDIA is working with healthcare organizations to build a true infrastructure from the ground up, providing edge architecture based on the software and workloads needed. They are instrumental in working with med tech and health tech companies on their AI model development, using open models and initiatives related to robotic surgery and spatial intelligence.
Security and Governance
Security is a critical component of edge-to-cloud AI, and healthcare organizations need to ensure the security of these tools. Niewolny notes that security needs to cover the full life cycle of an edge device, including hardware root of trust, secure boot, signed software, and network segmentation. Organizations need to complete a full inventory of devices and applications, monitoring all models, versions, and data flow.
This should be managed as part of the IT network, with continuous observation of latency, resource health, input quality, output distribution, model drift, and outcomes. NVIDIA provides medical-grade hardware needed to deploy at the edge in healthcare and accelerates model development and testing with simulation platforms. The company sees an opportunity for AI and automation to help solve serious capacity problems related to staffing shortages and cost of care.
