D.3.01. Health Record Structure: Design and Open Specification v1

Holistic Health Records (HHRs) are structured health records that may include several types of information that are relevant to a patient’s health status, such as laboratory medical data; clinical data; lifestyle data collected by the patient or related people; social care data; physiological and environment data collected by medical devices and sensors. Currently, many data models have been specified for representing the aforementioned data (section 3 State of the Art), but there is no single model that covers in an integrated way all the needs of CrowdHEALTH use cases. This deliverable presents a first version of such an integrated HHR model.

In order to have a strong foundation, the model is mainly based on the new emerging FHIR standard. FHIR is considered easier to implement than previous standards, covers very well the clinical aspects of human health and has a good extension mechanism that allows adding information models for aspects not yet covered by the model.

The HHR model has been obtained by first producing a separate conceptual data model for each use case, then clarifying their semantics with a preliminary mapping to the FHIR standard (annex B1 to B5, which comply to the template reported in annex B) and finally merging the separate models in a unique HHR conceptual model. A coherent mapping to FHIR has therefore been defined with respect to the merged conceptual model, in order to guarantee that different teams adopt the same FHIR representation (as FHIR allows different representations for the same information). The conceptual HHR model is specified in UML, with the adoption of specific constraints and stereotypes, while the mapping is expressed using structured tables and the FHIRPath language (section 4.3 UML conceptual model).

Although based on FHIR, the HHR model is designed at a higher conceptual level, making explicit several concepts that are implicit in the FHIR standard, other than extending it by adding missing concepts. By maintaining a double view, the HHR model aims on one hand to guarantee the interoperability and the possibility to implement it on top of existing FHIR libraries, and on the other hand it is also intended to be usable independently from FHIR (and its future evolutions) and applicable also for different purposes than the exchange of health data. For example, it can be more suitable than FHIR as data schema for Object Oriented local APIs.

The current HHR model aims to represent the information enabling the execution of the first cycle of use case demonstrations, expected for the first year of the project. In particular, the current release of the HHR model fully covers CareAcross and SLOfit datasets and partially covers HULAFE and BIOASSIST datasets. During the second year, the HHR model will be extended to satisfy new requirements, to complete the coverage of HULAFE and BIOASSIST datasets and to include the DFKI and KI datasets.