As hospitals integrate more advanced patient monitoring systems, the ability to process, analyze, and act on real-time medical data has become critical. Traditional monitoring systems generate large amounts of sensor data, but manually interpreting this data in real time is challenging for clinicians.

AI-driven analytics can transform how healthcare providers detect early signs of complications, improve patient outcomes, and optimize hospital operations. However, deploying AI in healthcare comes with significant challenges—ensuring scalability, security, and compliance while maintaining real-time performance.

This blog post explores how to build a pipeline with MATLAB and AWS® for an AI-driven remote patient monitoring system. It also presents a case study where GE® HealthCare used MATLAB and AWS for neonatal monitoring.

 

Challenge: AI Deployment for Hospital Environments

When deploying AI applications to a hospital environment, there are some key challenges to overcome so that the patient monitoring applications can provide meaningful insights and real-time alerts to healthcare providers.

• Data Management: Hospital systems generate high-frequency, multi-sensor data that must be efficiently processed and stored.
• Scalability and Automation: AI models must be deployed to environments where they can automatically operate, scale efficiently, and be able to adapt to varying patient loads.
• Compliance and Governance: Healthcare AI models must meet strict regulatory requirements and be continuously monitored for accuracy and reliability.

To address these challenges, engineers need to create a workflow that integrates AI model development with cloud deployment and automation.

 

Workflow: MATLAB-Based DevOps in AWS

MATLAB provides an end-to-end AI development environment, from data preprocessing and model training to cloud deployment. By leveraging AWS services, AI models can be continuously integrated and monitored within a DevOps framework.

 

Model Development in MATLAB

Engineers use MATLAB to preprocess physiological signals such as ECG, heart rate, and oxygen saturation. They use built-in tools and algorithms for feature extraction and training AI models. These models are then validated using hospital datasets to ensure their accuracy and reliability.

 

CI/CD Pipeline for AI Deployment

A DevOps pipeline automates model deployment in AWS:

• GitHub Actions triggers model retraining in MATLAB and testing when new data or code updates are pushed.
• MATLAB running in the cloud executes automated unit tests, ensuring model robustness before deployment. Learn more here.
• AWS CodePipeline orchestrates model packaging and deployment, ensuring a reliable transition from development to production.

Real-Time AI Inference in AWS

Once deployed, the AI models run on AWS services such as Amazon® SageMaker or AWS Lambda, enabling real-time inference on incoming patient data. Alerts and insights are streamed to hospital dashboards via Amazon Kinesis, providing healthcare providers with actionable intelligence.

 

Ensuring Model Performance and Compliance

Beyond deployment, AI models require continuous monitoring to maintain accuracy and compliance. The pipeline developed in MATLAB and AWS, achieves:

• Automated Performance Monitoring – Model monitoring and drift detection trigger the retraining of AI models to ensure that they remain performant over time.
• Audit Trails and Explainability – Explainable AI techniques help healthcare providers understand and gain trust in the decisions of the AI models.

 

Impact: AI Helping Critical Patient Care

The integration between MATLAB and AWS, enables to build an AI workflow that runs on a cloud environment for remotely monitoring patients. This complete workflow includes building AI models, deploying them to the cloud, and monitoring their operation. As hospitals continue to embrace AI, such solutions have the potential to improve patient outcomes and maximize clinicians’ efficiency.

More specifically, this workflow can bring significant impact to clinical environments with:

• Real-Time Insights: Continuous data processing and AI predictions ensure timely interventions for patients.
• Scalability: Cloud-based deployment allows hospitals to handle increased patient loads dynamically.
• Operational Efficiency: Automated DevOps pipelines reduce manual intervention and errors in operation.

 

Case Study: AI-Driven Neonatal Monitoring

In neonatal intensive care units (NICUs) clinicians are seeking ways to effectively manage patients and provide prompt, high-quality care to critically ill and premature newborns. GE HealthCare used MATLAB and AWS to develop an AI-driven remote neonatal monitoring system for NICUs.

Leveraging AWS cloud services and MATLAB AI capabilities, they created a contactless solution that uses cameras to monitor newborns’ vital signs, reducing reliance on traditional adhesive sensors. This approach improves patient comfort, facilitates remote clinician access, and ensures HIPAA compliance. The scalable system processes high-quality video streams and delivers real-time analytics to support critical neonatal care.

To learn more, see Developing AI-Driven Remote Neonatal Monitoring Using AWS and MathWorks with GE HealthCare.