Real-World Integration of a Sepsis Deep Learning Technology Into Routine Clinical Care: Implementation Study.

Area of Science:

• Clinical informatics
• Machine learning in healthcare
• Quality improvement initiatives

Background:

• Machine learning integration into clinical practice is infrequent.
• Barriers to adopting machine learning in healthcare are not well-documented.

Purpose of the Study:

• To detail a quality improvement initiative for integrating a deep learning sepsis platform (Sepsis Watch) into clinical workflows.
• To address the rarity of successful machine learning implementations in routine care.

Main Methods:

• A multidisciplinary team (statisticians, data scientists, engineers, clinicians) was formed in 2016.
• The learning health system framework guided the assessment, design, development, implementation, and evaluation of Sepsis Watch.
• Frontline clinical staff actively participated in workflow and model development.

Main Results:

• Sepsis Watch was successfully integrated into routine clinical care, influencing local machine learning project execution.
• Development of novel machine learning methods for early sepsis detection required robust infrastructure.
• Significant investment in stakeholder alignment, trust-building, role definition, and staff training was crucial.

Conclusions:

• Successful integration of machine learning into clinical decision-making remains a challenge.
• Learnings from the Sepsis Watch integration offer insights for other healthcare systems implementing deep learning technologies.