💕 Power Up Patient Monitoring with the Health and Life Sciences AI Suite on a Single Intel Edge Platform

[stevenhoenisch]

The preview release of the Health and Life Sciences AI Suite and its Multi-Modal Patient Monitoring app demonstrates how to run multiple AI workloads concurrently on a single Intel‑powered edge device without a discrete GPU so that you, as a medical AI developer, can see Intel® Core™ Ultra processors in action.

The Multi-Modal Patient Monitoring application showcases how a single Intel-powered edge system can simultaneously run several AI workloads within one integrated dashboard:

• Physiological signal processing
• Computer vision
• Artificial intelligence inference
• Real-time medical device telemetry

The patient monitoring application proves that heterogeneous workloads -- from 3D human pose estimation with joint tracking to heart and respiratory rate monitoring, AI-based ECG analysis with 12-lead classification, and medical device simulation -— can coexist efficiently on one platform without compromising performance or stability. By accelerating multi-modal AI pipelines with the OpenVINO™ toolkit, the Health and Life Sciences AI Suite empowers you to drive high-performance, AI-powered health care on an Intel® Core™ Ultra platform using CPUs, a built-in GPU, and a Neural Processing Unit, or NPU.

Multi-Modal Patient Monitoring Application

The Multi-Modal Patient Monitoring application shows off how the suite organizes workflows tailored for healthcare and life sciences and eases integration with medical devices. As a collection of healthcare-focused AI applications, libraries, and benchmarking tools, the suite helps medical AI developers, original equipment manufacturer (OEMs), and original design manufacturers (ODMs) build patient monitoring solutions powered by artificial intelligence faster.

With the Multi-Modal Patient Monitoring application, you can view four key patient monitoring workloads side‑by‑side through a graphical dashboard, with each workload displaying the following information:

• MD PnP Open-Source Integrated Clinical Environment (OpenICE) device integration for vital signs and device data
• 3D pose estimation with an OpenVINO webcam demo
• AI‑ECG analysis
• Remote photoplethysmography (rPPG) for contactless vital sign estimates

Outputs from these workloads are consolidated into a two-by-two layout, showing each stream in its own quadrant while sharing a single Intel Core Ultra CPU, iGPU, and NPU platform. Crucially, this combination helps reduce the bill of materials and simplify the deployment by consolidating multi‑modal AI on one cost-effective edge system.

Requirements and Setup for Demonstrating the Multi-Modal AI Capabilities of Intel Core Ultra Platforms

In patient monitoring application demonstrates the multi‑modal AI capabilities of Intel Core Ultra platforms. End-to-end setup and launch (with a single command) is estimated at about 30 minutes on Ubuntu 24.04 with containerized workloads; see the system requirements. Although secure provisioning (for example, with Polaris Peak integration) isn't part of the initial implementation, the architecture is intended to be extensible for security integrations in the future.

At a high level, the system is composed of several microservices that work together to ingest patient signals and video, run AI models on Intel hardware, aggregate results, and expose them to a user interface for clinicians. For details, see How It Works. Here's a glimpse of its architecture:

After you set up the Multi-Modal Patient Monitoring app, you can experiment with different RPPG_DEVICE values to compare CPU, GPU, and NPU behavior. You can also replace the sample video or models with your own assets by updating the models and videos volumes and configuration; see, for instance, the Configure Hardware Target section of the Get Started guide.

Conclusion

In summary, the preview release of the Health and Life Sciences AI Suite and its Multi-Modal Patient Monitoring app demonstrates the following:

• A single Intel®‑powered edge system running multiple heterogeneous healthcare workloads concurrently
• Integrated computer vision, physiological signal processing, AI inference, and real‑time medical device telemetry
• Efficient coexistence of workloads such as pose estimation, heart and respiratory rate extraction, AI‑based ECG analysis, and medical device simulation, all without compromising performance or stability

💕 Check out the Health and Life Sciences AI Suite in its GitHub repository and power up Multi-Modal Patient Monitoring with the Get Started guide.