REAL-TIME HEART MONITORING OF THOUSANDS OF PATIENTS
CardioHPC improves existing DL solutions with advanced HPC services, addressing real-time ECG data streaming and computational-intensive tasks. It simulates 10K patients in a heart-monitoring center, requiring a cloud-based HPC system and sophisticated architecture for efficient parallel processing.
Start date: 01/06/2021
Duration in months: 18
Problem Description
Usually patients have to stay in the hospital for ECG monitoring. Innovation Dooel's ViewECG uses real-time ECG data from wearable sensors and Signal Processing and Machine Learning techniques to monitor heart arrhythmias, allowing patients to follow their routines while maintaining their health.
Goals
Reduce time to solution
Challenges
The ML arrhythmia classification model requires extensive benchmark dataset training and high computing capacity for efficient processing of concurrent ECG data streams, which Innovation Dooel's infrastructure cannot meet. Scaling up to more users requires this capability.
Innovation results
The consortium utilized HPC to address business challenges, training a new machine learning model for heart arrhythmia detection. The HPC solution efficiently processed thousands of ECG streams, using containers and a service model.
Business impact
Innovation Dooel expects to double their revenue by selling the new service which now scales to tens of thousands of users, a 50% error reduction and a 90% accuracy, a 25% increase of efficiency and profit due to reduced costs for processing an incremented workload with reduced software administration.