Table 2 CDS infrastructure needs and available resources

Centrally managed repositories of computer-processable medical knowledge

Initiatives that can serve as models for creating authoritative knowledge on how genomic interventions should be used in clinical practice (e.g., U.S. Preventive Services Task Force [44], U.S. Evaluation of Genomic Applications in Practice and Prevention initiative [45])

Centrally managed repositories of structured experimental data relevant to genomic medicine (e.g., PharmGKB for pharmacogenetic and pharmacogenomic experimental data and curated knowledge [46]; NCBI dbGaP for data from studies evaluating the interaction between genotypes and phenotypes [47]; and NCBI GEO for high-throughput, experimental genomic data [48])

Various formalisms for representing medical knowledge in a computer-processable format [49, 50] (e.g., Arden Syntax [42], GELLO [41], GLIF [52], PROforma [53], SAGE [54], SEBASTIAN [55])

Large-scale efforts at managing computer-processable medical knowledge (Intermountain Healthcare [56], Partners HealthCare [57])

Standardization of CDS information provided for specific aspects of genomic medicine

Standard mechanisms for defining CDS information to be provided in specified contexts (e.g., HL7 refined message information models [49], openEHR Archetypes [58], HL7 Decision Support Service semantic profiles [59])

Standardized representation of patient data

Standardized information models and terminologies for both genomic and traditional patient data [49, 60] (e.g., HL7 data standards, including HL7 Clinical Genomics data standards [60] and emerging HL7 virtual medical record standard [62]; openEHR Archetypes [58]; SNOMED CT [38]; LOINC [39]; BSML [60]; MAGE-ML [63]; National Cancer Institute/caBIG Common Data Elements [61])

Resources that enable mapping between different terminologies (e.g., Unified Medical Language System [65], National Cancer Institute Enterprise Vocabulary Services [66], HL7 Common Terminology Services standard [67])

Standard approaches for leveraging computer-processable medical knowledge

Various approaches for using computer-processable medical knowledge resources to provide automated CDS [31] (e.g., Arden Syntax [42], SAGE [54], PRODIGY [68], GLIF [69], SEBASTIAN [55], First DataBank Drug Information Framework [70], HL7 Decision Support Service [59])

Standard approach for locating and retrieving patient data

Standardized approaches for resolving patient identities across systems (e.g., HL7 Entity Identification Service draft standard [72] and corresponding OMG technical specification [73]) and for locating and retrieving patient data across systems (e.g., HL7 Retrieve, Locate, and Update Service draft standard [74] and corresponding OMG technical specification [75])

Regional and national initiatives for secure health data exchange (e.g., U.K. National Health Service Connecting for Health [76], U.S. Nationwide Health Information Network prototypes [77], caBIG [78], Indiana Health Information Exchange [79])

All infrastructure components

Initiatives to share machine-processable medical knowledge (e.g., Morningside Initiative [80], U.S. federal CDS Collabatory [81])

Efforts at specifying functional requirements of EHR systems (e.g., HL7 EHR System Functional Model standard [82], CCHIT EHR certification criteria [83])

Efforts at coordinating the use of available health information technology standards (e.g., U.S. Health Information Technology Standards Panel [84], Integrating the Healthcare Enterprise [85])