TABLE 1.
Determinants of the Use of Data Science in Critical Care
| Reddy’s Synthesized Factors | (Updated) Consolidated Framework for Implementation Research Constructs | Relevant Areas for Determinant Measurement in Critical Care |
|---|---|---|
| Data characteristics: volume, velocity, and variety of data; data quality | Innovation complexity | Nature and number of connections and steps involved in real-time analysis of granular data to inform changing clinical conditions in life-or-death treatment scenarios |
| Quality and level of integration of disparate data sources, especially: lifespan data (e.g., pedigree); systems data (e.g., environmental exposome) | ||
| Technology: Infrastructure (hardware and software) | IT infrastructure | Ability to port, access, merge, and analyze critical care clinical (and other) data sources, such as from ICU monitoring systems |
| Organizational structure/model; organizational agility: business process, realignment of work practices, intensity of learning; ambidexterity | Work infrastructure | Organizational architecture and staffing to optimize dynamic, agile learning processes, characterized in particular by: |
| Flexibility to (re)create roles, teams, and administrative procedures to catalyze digital innovation | ||
| Standardized training to drive organizational learning | ||
| Data-driven decisionsInnovativeSustainable | Culture | Support for data-driven decision-making, and especially for open-source common data models (such as OMOP and FHIR) |
| Willingness to enact and sustain IT infrastructure necessary for data science implementation | ||
| Alignment with core strategy: distinct data strategy, (coordination of) stakeholder interests | Mission alignment | Presence of measurable goals and objectives related to the integration of data science into critical care research and practice |
| Resources to support novel data science initiatives amid multiple institutional priorities, in particular through strategic integration with existing priority initiatives (e.g., grants, quality registries, industry collaborations, etc) | ||
| Policies and regulations | Policies and laws | Data security and privacy policies to protect patient and organizational data |
| Governance: control (access), risk management, compliance, privacy, security, sharing, ownership | Data governance processes with respect to the context of data reuse, accuracy, archival, curation, platforms, architecture, and effective sharing within and across ICUs and organizations | |
| Business environment: competitive dynamics, industry structure, partner readiness, consumers (public) | External pressure | Urgency to operationalize real-world data to produce clinical knowledge (e.g., on COVID-19) |
| Market pressures influencing availability of resources for integration of critical care/data science | ||
| Incentive to align public and private strategic partnerships (e.g., electronic medical record/electronic health record industry partners) to facilitate alignment of proprietary data | ||
| Managerial willingness: top management support, trust and acceptance, breaking silos | High-/mid-level leaders | Buy-in from key administrative decision-makers about the acceptability, appropriateness, and feasibility of data science |
| Human-machine teaming trust | ||
| Talent (knowledge and skillset) | Implementation facilitators (SMEs) | Synergistic clinical and informatics expertise related to collecting, analyzing, and operationalizing large datasets to improve clinical practice and facilitate clinical trials |
FHIR = Fast Healthcare Interoperability Resources; IT = information technology; OMOP = Observational Medical Outcomes Partnership; SMEs = subject matter experts.