Page 26 - AIH-1-3

Artificial Intelligence in Health Optimizing EHRs to support AI

Table 1. Ecosystem high-level building blocks enabling optimally connected digital health
Legislation, policy, and Policy directions and associated legislation enabling the provision and compliance monitoring of health service
compliance 3 funding mechanisms enabling universal, person-centered healthcare through an optimized primary care
infrastructure supported by a well-specified and mandated digital health infrastructure. This includes privacy and
cyber security legislative and regulatory requirements.
Ecosystem system An ecosystem architecture needs to include an open platform able to meet the needs of all stakeholders and support
architecture 20,92 the data supply chain.
“Better data and regular data use will create a data use culture, leading to better decisions, an improved health system,
and improved health outcomes.” 93
Ecosystem data architecture 94 A representation of concepts and their relationships. The data architecture defines concepts, constraints, and rules,
which provide safe consistent data collection and use which retains meaning throughout the data supply chain. The
domain or discourse contribute to the architectural requirements and select data from the data ecosystem based on
their use case. Such data collections result in data that are structurally independent, simpler, and safer to share. The
resulting lack of data silos enables advanced data analytics.
Concept representation Key health concepts need to be represented in the same manner throughout any digital health ecosystems to ensure
standards 18 data accuracy, enable consistent quality data collection at every level, optimizing data analytics, and reducing data
collection burden. Consistent representation of key health concepts enables evidence-based decision-making at all
levels and is best achieved by adopting a multilevel modeling approach and an open platform.
Data/information governance 48 Specification of decision rights and an accountability framework to ensure appropriate behavior in the evaluation,
creation, storage, processing, use, archiving, and deletion of information. Coordinated data governance applies to all
points along the data supply chain.
Data access control Legislation is required to indicate who can have access to identifiable and non-identifiable data for what purpose.
Legislative mandates and regulatory requirements need to be considered in the light of ethical data use, and “use case”
specific privacy and confidentiality, and continuity of care considerations.
Unique identifiers An essential pre-requisite to ensure data collected can be linked to care recipients as well as to organizational and
individual providers.
Cybersecurity Minimizing risk of cyberattacks by protecting systems, servers, networks, and mobile devices. Adopt and maintain
programs that educate the workforce, and manage and monitor unauthorized data access.
Vendor/technology-neutral The separation of systems and storage delivering scalable cost-effective data access and flexible systems for all users
federated data storage across the health-care network. Separating data from applications as used by the openEHR community were found to
support persistent and transient data as well as real-time local and remote data access.
Electricity and broadband A fundamental pre-requisite for all living in this digital era, irrespective of time, and location.
(Internet access) for everyone

and biobanks. The patients’ own contribution to their whom do not have full understanding about healthcare.
EHR, and the increasing use of mobile devices and sensors, Therefore, adoption of the openEHR standard is key for
are also important. They can add valuable insights about designing and building next-generation EHR/EMRs
environmental and behavioral factors as well (e.g., food, air systems and other applications that deal with clinical data.
quality, exercise, and mood).
The emerging trend of using of HL7 FHIR beyond its
Both data- and terminology‐level standards are purpose to represent the full breadth of clinical data in an
reasonably mature, although there is considerable overlap EHR is not scalable and costly in terms of time required
among certain terminology and ontology systems such as to develop and maintain FHIR profiles. It is far more cost-
SNOMED CT and LOINC. Using fit-for-purpose data and effective and safe to invest in the establishment of next-
ontology/terminology standards together can tackle most of generation neutral EHR systems with vendor-neutral data
the difficulties arising from the breadth, depth, complexity, repositories using openEHR and limit the use of FHIR to
variability, changeability, and longevity aspects of health data. support simpler use cases for data exchange.
While openEHR specifications have been purposely
engineered to cover all EHR data domains, including those 4.6. Governance and leadership
that are intended to be exchanged by various systems, Our collective work over the last 20 plus years has
messaging standards such as HL7 v2 and HL7 FHIR highlighted the need for high-level governance leadership to
have been designed to cover only data to be exchanged. maximize collaboration between all relevant stakeholders.
These messaging standards were designed for the sake of Our collective findings over time, supported by this rapid
simplicity for implementation by developers, but many of review, have enabled us to identify the required building

Volume 1 Issue 3 (2024) 20 doi: 10.36922/aih.3056