Table 2 Illustration of the application of the proposed LHS architectural framework
From: Architectural frameworks: defining the structures for implementing learning health systems
Layer | Learn From Every Patient initiative [42] | PaTH clinical data research network initiative [21] |
|---|---|---|
Performance | Goal: continuous quality improvement in clinical care for children with cerebral palsy. Measures: changes in healthcare utilization rates and related costs | Goal: informatics-supported infrastructure for cohort identification and data sharing for idiopathic pulmonary fibrosis, atrial fibrillation, and obesity. Measures: not discussed |
Scientific | 12-month study of one cohort, within a series of learning projects for continuous quality improvement | Comparative effectiveness randomized trials with specific research questions informed by clinical data research experts and vetted by informatics group |
Organizational | Program team includes key clinical stakeholders, clinical and information technology teams | Steering Committee includes representatives from each site, three advisory committees (including patients), four working groups (research questions, information technology, methodology, regulations) |
Data | Source: data fields and questions added to institution’s EHR. Data quality: ensured by database manager | Source: Complete set of longitudinal data about target populations taken from site EHRs. Standards: Standardized descriptions of data elements using established standards and vocabularies; use of HL7 by all participating health systems. Quality: manual and automated monitoring by data engineers. |
Information technology | Existing EHRs and related infrastructure | Source data loaded onto centrally-maintained data warehouse; queriable through analytics interface |
Ethics and security | No review board authorization required (all data collected appropriate for standard clinical care) | Not directly discussed; data transformation process includes de-identification prior to loading into warehouse |