Hybrid Workflow Policy Management for Heart Disease Identification

Dong-Hyun Kim, Woo Ram Jung, Chan-Hyun YounOpenReview 2010 · OpenReview

Published in the International Journal of Engineering Science and Technology (2009), this research introduces a "hybrid" approach to managing complex medical workflows within Grid computing environments. By bridging the gap between user-centric design tools and high-performance execution engines, the authors aim to make sophisticated heart disease diagnostics both more accessible and computationally efficient.

The Challenge: Scientific vs. Enterprise Workflows

As medical applications—such as genomic analysis and heart simulations—become increasingly data-intensive, they require the distributed power of Grid computing. However, a significant gap exists in existing software solutions:

  1. Scientific Workflow Management Systems (WMS): Tools like Pegasus excel at scheduling tasks across distributed resources but often lack intuitive user interfaces.
  2. Enterprise WMS: These systems are user-friendly but struggle with the massive data processing and "heavy lifting" required by scientific research.
  3. Service Level Agreements (SLAs): Medical professionals have varying priorities, such as requiring a quick diagnosis versus a low-cost one. Most existing systems lack the mechanism to guarantee these specific requirements.

A Hybrid Architecture for High-Performance Diagnostics

The researchers developed the "Physio-Grid" platform to address these issues by integrating two specialized tools governed by a policy-based management layer. The system uses Kepler for high-level workflow composition (the user interface where logic is defined) and Pegasus for low-level execution (the engine that runs jobs on the Grid).

The three-tier Physio-Grid architecture integrating mobile applications, hybrid middleware, and grid resources.

At the heart of this system is the PQRM Middleware (Policy Quorum based Resource Management). This layer acts as the "glue," featuring a Policy Decision Point (PDP) that evaluates user requirements against available resources and a Policy Enforcement Point (PEP) that dynamically adjusts system behavior. This allows the system to translate "soft requirements" (like "High Performance") into mathematical weights for resource allocation.

Intelligent Medical Workflows: From ECG to Simulation

The system is designed to handle various data types, including real-time Electrocardiogram (ECG) and Magnetocardiogram (MCG) signals. A key innovation is the workflow's inherent intelligence: it begins with simple, low-cost ECG analysis and only escalates to resource-intensive simulations when necessary.

The heart disease identification workflow logic, showing the conditional trigger for Virtual Heart Simulation (VHS).

As shown in the workflow logic, if the initial analysis detects symptoms such as fibrillation or ischemia, the system automatically triggers a Virtual Heart Simulation (VHS). While these simulations require massive computing power, they provide the high-fidelity data needed for an accurate diagnosis in complex cases. This tiered approach ensures that high-performance resources are only utilized when the medical condition warrants the cost.

Performance and Policy-Driven Results

The authors evaluated the system by running heart disease identification workflows under various policy strategies. The results demonstrated that policy-driven management significantly improves both speed and predictability:

  • Optimization: Using a full suite of optimized policies (Resource, Workflow, and High-level), the total elapsed time for the workflow was reduced to approximately 35–40 minutes.
  • Stability: In contrast, a "Random" resource selection strategy resulted in completion times of up to 70 minutes with high variance, making the system unpredictable for clinical use.
  • Cost Efficiency: The PQRM middleware successfully maintained SLA guarantees even as network conditions changed, identifying the most efficient resource sets dynamically.

Conclusion and Future Work

The hybrid approach effectively separates the logic of medical diagnostics from the complexities of Grid scheduling. While the system successfully automates the transition from simple ECG analysis to complex simulations, the authors note that future work is needed to improve real-time visualization within the design interface and further reduce the latency of massive batch-heavy simulations. Ultimately, this framework provides a blueprint for making high-performance scientific computing a practical tool for the medical community.

Summary updated

Abstract

As science technology grows, medical application is becoming more complex to solve the physiological problems within expected time. Workflow management systems (WMS) in Grid computing are promising solution to solve the sophisticated problem such as genomic analysis, drug discovery, disease identification, etc. Although existing WMS can provide basic management functionality in Grid environment, consideration of user requirements such as performance, reliability and interaction with user is missing. In this paper, we propose hybrid workflow management system for heart disease identification and discuss how to guarantee different user requirements according to user SLA. The proposed system is applied to Physio-Grid e-health platform to identify human heart disease with ECG analysis and Virtual Heart Simulation (VHS) workflow applications.
BibTeX
@misc{kim2010hybrid,
  author = {Dong-Hyun Kim and Woo Ram Jung and Chan-Hyun Youn},
  title = {{Hybrid Workflow Policy Management for Heart Disease Identification}},
  year = {2010},
  eprint = {1001.4199},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1001.4199},
}
Figure