From intensive care monitors to cloud environments: a structured data pipeline for advanced clinical decision support

Area of Science:

• Health Informatics
• Data Science in Healthcare
• Clinical Data Management

Background:

• Clinical decision-making increasingly relies on data-driven approaches and large datasets for developing diagnostic and predictive algorithms.
• Intensive care units (ICUs) generate high-frequency, unstructured monitor data, posing challenges for traditional data management.
• A successful data pipeline has been developed to transfer patient data to a cloud environment for structured storage.

Purpose of the Study:

• To showcase a data pipeline for efficient transfer and structured storage of ICU patient data in a cloud environment.
• To support individual patient analysis and large-scale retrospective research.
• To enable the development of data-driven algorithms for healthcare applications.

Main Methods:

• ICU data from Amsterdam UMC collected since June 2021 and stored in a third-party cloud environment on virtual servers.
• Demonstration of pipeline feasibility using available data for research and clinical use cases.
• Consideration of privacy, safety, data quality, and environmental impact during the cloud storage transition.

Main Results:

• Over two years, data from more than 9000 patients have been successfully stored in the cloud.
• Cloud environment offers availability, agility, computational power, high uptime, and streaming data pipelines for retrospective analyses and real-time predictions.
• Machine learning algorithms can be implemented for real-time prediction of critical events, with access via keyword search in natural language data.

Conclusions:

• Cloud environments provide essential features for developing and implementing predictive algorithms.
• The data pipeline facilitates healthcare evaluation and enhances individual patient care.
• Structured cloud storage of ICU data supports advanced research and clinical decision-making.