TABLE 2.

Implementation Strategies for the Use of Data Science in Critical Care

Data Science Campaign Domain	(Refined) ERIC Implementation Strategies	(Refined) ERIC Implementation Strategy Definitions	Actors, Actions, Temporalities, Doses, and Justifications
Data harmonization and sharing	Develop resource-sharing agreements	“Develop partnerships with organizations that have resources needed to implement the innovation”	Actors: Clinicians, researchers, administrators, data scientists
	Stage implementation scale-up	“Phase implementation efforts by starting with small pilots or demonstration projects and gradually move to a system wide rollout”
			Actions: Establish principles for data standardization and sharing, develop CDEs for critical care, and perform a pilot project utilizing standardized CDEs
			Temporality: CDEs modified and expanded regularly
			Dose: Pilot: 2 yr
			Justification: Fragmentation of data elements (especially in United States) substantially impedes research and quality improvement efforts
Discovery Critical Care Datahub	Use data warehousing techniques	“Integrate clinical records across facilities and organizations to facilitate implementation across systems”	Actors: Data scientists
	Centralize technical assistance	“Develop and use a centralized system to deliver technical assistance focused on implementation issues”	Actions: Aggregate electronic medical record data on ICU admissions from across sites to develop a cloud-based multipurpose platform for analysis of both clinical and nonclinical information. Enable federated data approaches (i.e., noncentralized data locations usable as a unified whole)
			Temporality: Once developed, always available
			Dose: Continuously maintained datahub, with multiple pathways for further data ingestion over time
			Justification: Current process of manual data extraction and transformation across sites requires unsustainable personnel effort requirements
Society of Critical Care Medicine-sponsored datathons	Use data experts	“Involve, hire, and/or consult experts to acquire, structure, manage, report, and use data generated by implementation efforts”	Actors: Clinicians, researchers, data scientists
	Increase demand	“Attempt to influence the market for the clinical innovation to increase competition intensity and to increase the maturity of the market for the clinical innovation”
			Actions: Host recurring competitions focused on using data science techniques to solve real-world issues
			Temporality: Annually
			Dose: Two-d workshop
			Justification: Pressing need for expedited application of data science in critical care research and practice, especially in emergency situations (e.g., natural disasters, emerging epidemics/pandemics)

CDE = common data element, ERIC = Expert Recommendations for Implementing Change.