A Clinical Data Warehouse Architecture based on the Electronic

Healthcare Record Infrastructure

Fabrizio Pecoraro

, Daniela Luzi

and Fabrizio L. Ricci

Institute for Research on Population and Social Policies, National Research Council, Via Palestro 32, Rome, Italy

Institute for Systems Analysis and Computer Science, National Research Council, Viale Manzoni 30, Rome, Italy

Keywords: Electronic Healthcare Record (EHR), Secondary Uses, Clinical Indicators, Data Warehouse, Dimensional

Model.

Abstract: The development of clinical data warehouses is becoming increasingly important in the healthcare domain

to support organizations in the improvement of decision-making, business processes as well as the

communication between clinicians, patients and the administration. However, data and process integration is

a big challenge considering the heterogeneous and distributed nature of healthcare information systems.

This paper proposes a data warehouse architecture based on the Italian Electronic Health Record (EHR)

technological infrastructure. It describes the main advantages in the application of EHR systems for

secondary purposes and reports the data warehouse design framework outlining its architecture as well as a

dimensional model based on a dashboard defined to manage the intervention of patients with diabetes. The

adoption of EHR systems enhances interoperability given that these systems share standardized clinical data

among multiple parties involved in different healthcare settings.

1 INTRODUCTION

The widespread diffusion of Information

Technologies in different healthcare settings has led

to a production of a massive amount of both clinical

and administrative data. Despite all the efforts, these

information are often stored in standalone

heterogeneous information systems developed for

specific specialties (radiology, admissions, general

ledger, scheduling, pharmacy and patient records)

that do not interchange data with each other

(Wickramasinghe and Schaffer, 2006). Therefore,

researchers doing data analysis still face

interoperability and technical challenges in the

support of administrative and clinical processes for

purposes different than those they were gathered for

(e.g. management information, quality assessment

and research) (Kush et al., 2008, Taylor, 2008). In

addition, data analysis can be a complex task

considering that medical and sensitive information

(Koh and Tan, 2011) are usually restricted in access

due to ethical, legal and privacy issues. This makes

it also necessary to collect and integrate these

information before data analysis can be performed

and to adopt specific techniques for data

anonymisation and processing.

An important initial step toward the integration of

data provided by heterogeneous multiple systems is

the development of enterprise clinical data

warehouses that are becoming increasingly

important in the healthcare domain (Botsis et al.,

2010, Kamal et al., 2005). This approach is adopted

by different parties (e.g. hospital, GP, specialists)

(Zhou et al., 2010, Sahama and Croll, 2007) and at

different organizational level (i.e. local, regional and

national authorities) (De Mul et at., 2012, Stow et

al., 2006) in order to improve decision-making,

business processes as well as communication

between clinicians, patients and the administration.

However, the implementation of data warehouses

implies the solution of issues related to missing,

corrupted, inconsistent or non-standardized data

collected in different formats and data sources. In

particular, the lack of a standard vocabularies is a

serious barrier for the integration and analysis of

data (Gillespie, 2000).

Electronic Health Record (EHR) systems

represent a core source of information, managed and

processed by multiple parties involved in healthcare

settings. Their main purposes are to support

physicians and other professionals in the delivery of

care management services giving direct benefits to

287

Pecoraro F., Luzi D. and L. Ricci F..

A Clinical Data Warehouse Architecture based on the Electronic Healthcare Record Infrastructure.

DOI: 10.5220/0004764502870294

In Proceedings of the International Conference on Health Informatics (HEALTHINF-2014), pages 287-294

ISBN: 978-989-758-010-9

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

citizens in terms of speed and appropriateness of

healthcare service delivery (primary uses). However,

different recent experiences have recognized the

successful use of EHRs for secondary purposes

(Safran et al., 2007), such as clinical research

(Hussain et al., 2012), epidemiological studies

(Diomidous et al., 2009), ambulatory clinical care

(Jensen et al., 2012), pharmacovigilance (Wang et

al., 2009), comorbidity detection (Roque et al.,

2011) or to alert providers of potential clinical risks

(Lurio et al., 2010). However, these secondary uses

of EHR are generally limited to a single institution

(e.g. Hospital) or a single provider (e.g. General

Practitioners) and/or on a specific target population

(e.g. diabetics, investigational patients). Conversely,

in our approach EHR is considered as a large-scale

information infrastructure that integrates

heterogeneous information systems managed by

different organizations in a distributed environment.

In our vision this EHR infrastructure can be the basis

to develop a data warehouse as a business

intelligence tool in a clinical governance framework.

In this perspective EHR systems can be used to

evaluate the effectiveness and appropriateness of

healthcare services from the structural,

organizational, financial and professional points of

view. This improves the transparency of economic

and clinical activities as well as the availability of

real time information to decision makers (Mettler

and Rohner, 2009). Moreover, the use of EHRs

increases healthcare data quality and facilitates the

interoperability between different systems and

organizations. In particular, standardized data model

as well as international widespread vocabularies and

nomenclatures ensure a reliable data collection and a

consistent data comparison even when collected

from heterogeneous information systems. This

approach can also provide additional values in

different healthcare-related sectors including

education, clinical research, public health, security

and policy support (Committee on Data Standards

for Patient Safety, 2003).

This paper proposes a clinical data warehouse

architecture based on the EHR infrastructure

developed in Italy. To demonstrate the feasibility of

the data warehouse architecture a set of indicators

related to diabetes pathology is proposed in

particular to implement a chronic disease

management intervention. Afterwards, starting from

these indicators a business process is presented

highlighting the design methodology of a

dimensional model.

2 DATA WAREHOUSE DESIGN

2.1 EHR Infrastructure and

Conceptual Model

In Italy the development of a local EHR system is

delegated to each regional administration. To ensure

the interoperability between the different local

solutions the Department for the digitization of

Public Administration and Innovation Technology in

collaboration with the Department of Information

and Communication Technologies of the National

Research Council (CNR) have carried out the InFSE

project (EHR technological infrastructure) (Ciampi

et al., 2012) that defines a set of technological

requirements with the aim of developing an

interoperable EHR national infrastructure. InFSE

provides a set of infrastructural components that

notify clinical events to the involved local EHR

systems through the adoption of a publish-subscribe

pattern. Moreover, it archives clinical documents as

generated by authorized users during a clinical event

guaranteeing their persistency, security and

reliability. To ensure semantic interoperability

among local systems clinical documents stored in

the relevant repository are structured using HL7

CDA (Clinical Document Architecture) Release 2

standard (Dolin et al., 2006). This standard allows to

structure the content of both header and body of a

document using the XML standard based on HL7

Reference Information Model (RIM) (Schadow et

al., 2006), coupled with terminology. Moreover, to

facilitate documents retrieval and localization InFSE

components manage a set of descriptive metadata,

such as document type, patient ID, document author,

organization responsible for the document security,

date of creation and update of the document,

location of the document (URI), etc. However, to

simplify the relationship between documents

produced in different clinical events as well as to

facilitate their sharing between different actors it is

necessary to introduce a higher level of document

aggregation and classification schema, providing a

set of concepts that represents both content and

context of healthcare services. These concepts were

defined in the CONTsys standard (EN 13940, 2007)

to describe different aspects of clinical and

organizational processes such as health issue,

contact and episode of care. They enable the

information management of the healthcare delivery

process to an individual subject of care as well as its

continuity, taking into consideration data handling,

decision processes, quality control and resource

management. These concepts and their relationships

HEALTHINF2014-InternationalConferenceonHealthInformatics

288

have been the basis of the EHR conceptual model as

shown in Figure 1 mapped in the HL7 RIM.

Figure 1: Portion of the EHR domain model based on the

CONTsys concepts.

The RIM is not just the starting point to represent

a clinical document backbone but it is also used as a

standard for defining the structure of a message to be

exchanged between heterogeneous information

systems to achieve semantic interoperability (Oemig

and Blobel, 2005). The main class of this model is

the Contact that describes an encounter between the

Patient and a Healthcare Provider (e.g. GP). Each

Contact is associated with one or more Episode Of

Care (e.g. weight gain), each one related to one or

more Health Issue (e.g. diabetes) suffered by the

Patient. During a Contact a set of Clinical

Documents are produced by the Provider. An

extended model of the proposed diagram has been

used to define the conceptual model of LuMiR, a

region-wide EHR system (Contenti et al., 2008).

Based on the conceptual model shown in Figure 1 it

is possible to define relevant indicators that can be

used for secondary purposes such as the comparison

of different phenomena over a period of time and

between different areas of the same region.

2.2 Data Warehouse Architecture

From the architectural point of view, the data

warehouse has been designed as a three-tier

architecture considering the information collected

and the relevant infrastructure of the EHR described

in the previous paragraphs. As highlighted in Figure

2, data are extracted from the legacy operational

systems (EHR data sources) and subsequently

cleaned-up and integrated in a data staging area

represented by an Operational Data Store (ODS).

Moreover, ETL (Extract, Transform, Load) tools

feed the data warehouse (and data marts) with

already integrated data based on a shared message

model that do not require transformation. This

makes it easier to extract the information from the

data source layer and load them in the data

warehouse for statistical purposes.

Figure 2: Data warehouse architecture proposed.

The source layer is represented by the set of

legacy systems and repositories (e.g. GP’s electronic

healthcare record, laboratory information systems,

radiology information systems) that manage

healthcare and administrative information related to

citizens. Identification of data sources represents a

critical process to be carried out for the success of a

data warehouse project. It is important to obtain

information about selected data sources identifying

the role played by each information system in the

development of the data warehouse and data marts.

Given that data collected in source information

systems are usually stored in different formats and

on a variety of platforms, it is necessary to define a

common data model that integrates data handled by

multiple sources into a single database. To achieve

this aim some authors have proposed to use a

module called wrapper (Roth and Schwarz, 1997),

that is responsible for data gathering from different

sources, data cleansing, format conversions as well

as data integration. Data managed by each wrapper

can therefore be loaded in the data warehouse and

thus in the relevant data mart. However, a change in

a data warehouse schema makes the revision of each

wrapper not straightforward. Therefore, it is

necessary to include in the architecture an

intermediate stage between the data sources and the

data warehouse tiers. This middleware system,

called ODS can also be used as a database for

operational processing, as proposed by Inmon

AClinicalDataWarehouseArchitecturebasedontheElectronicHealthcareRecordInfrastructure

289

(Inmon, 1999). It contains detailed and integrated

data with specific constraints including referential

integrity that ensures data accessibility by relevant

units. In the proposed architecture the ODS is

represented by two inter-related systems: a) the EHR

repository that contains individual’s structured

clinical documents together with the EHR registry

that manages metadata for indexing clinical

documents stored in relevant repositories; and b) the

Virtual Healthcare Record (VHR) that manages data

extracted from the documents contained in EHR

repositories, parsing the HL7 CDA document. The

integration of these systems is an appropriate

approach to define the ODS considering both

conceptual and operational point of views. Referring

to the first one, a common data model has already

been defined to ensure semantic interoperability

between heterogeneous legacy systems and the

EHR/VHR system based on the conceptual model

shown in Figure 1. This model comprises message

information models to exchange data about events,

health issues, episode of care, etc. as well as to

structure and share clinical documents produced in

each event. Considering the operational point of

view message exchange between legacy software

and EHR/VHR is based on a standard protocol

already implemented in some regional EHR

infrastructure. In particular, each source system has

installed a specific driver that extracts relevant

information from the proprietary database describing

the event happened (e.g. GPs visit) along with data

and documents provided already in HL7 CDA

format (e.g. drug prescription). The driver creates an

XML standard file that is then provided to a generic

wrapper that transforms it in a standardized HL7

message subsequently loaded in the EHR/VHR

system. In this way the specific driver installed in

each legacy system and the generic wrapper can be

seen as an ETL tool in the data warehouse

architecture. This makes it easier to integrate data

provided by different systems in the EHR/VHR and

thus in the data analytical process. The data

warehouse collects information stored in the

EHR/VHR in an OLAP (On-Line Analytical

Processing) approach that facilitates the integrated

analysis to develop specific dashboards based on

business processes and clinical indicators. Given that

VHR is composed by data extracted from clinical

documents, there is a need to develop a specific tool

that parses relevant information from EHR/VHR and

store them in the data warehouse. At this stage of the

design process an important role is played by the

Hierarchical Event Manager the InFSE module

responsible for a real-time notification of clinical

data and documents generated during each notified

event (Ciampi et al., 2012). This component can be

used not only to integrate different systems in the

EHR infrastructure, but also to feed the data

warehouse as well as the data marts. In particular,

this manager notifies users with events that occurred

and they are interested in, through a federation of

brokers operating on a publish-subscribe paradigm.

Of course other techniques are allowed to extract

data from the EHR and load them in the data

warehouse, in particular in case of frequent

monitoring of services. It is important to emphasize

that confidential information exchanged between the

ODS and the data warehouse must be anonymized to

preserve patient’s privacy. At this architectural level

clinical information managed by the EHR/VHR can

be also integrated with other data such as social,

demographic, economic, etc.

Finally, the analysis layer concerns tools and

techniques for data analysis, such as data mining,

reporting and OLAP tools. For instance, they can be

used to define a set of clinical indicators, as

highlighted in the following.

2.3 Dashboard

In order to identify a set of reliable indicators, in this

study the attention has been focused on the diabetes

mellitus type 2 pathology. This disease leads to

several extremely debilitating complications that can

impair the function of vital organs such as heart (e.g.

myocardial infarction), kidney (e.g. renal failure),

blood vessels (e.g. cardiovascular diseases, stroke)

and eyes (e.g. retinopathies). These health issues are

monitored in complex processes with frequent

contacts between patient and healthcare facilities

and professionals. This makes it necessary to timely

collect data and to foresee the verification of the

healthcare assistance processes in a regional and

national benchmarking framework. In Italy, a

comprehensive strategy for implementing a chronic

disease management intervention for people with

diabetes is defined in the IGEA project (Maggini,

2009). Within this project a set of clinical indicators

has been defined to measure the process, structure

and outcomes of patient care in order to assess a

particular clinical situation and to indicate whether

the healthcare delivered is appropriate. They are

used by different bodies at different levels to

identify areas of concerns, which might require

further review or development. Clinical indicators

are usually included in a benchmarked framework

comparing the outcome of each local agency over

different period of time with other agencies or with

HEALTHINF2014-InternationalConferenceonHealthInformatics

290

the literature results published, for instance, for other

populations or pathologies. In this paper the

attention is focused on this type of indicators that are

further divided into two categories: 1) process: to

assess whether a program is properly implemented

by the provider; 2) outcome: to measure the

program’s level of success in improving service

accessibility, utilization or quality. They capture the

effect of care processes on the health and wellbeing

of patients and populations. Outcome indicators are

intermediate if they reflect changes in biological

status or in life-style or final indicators if they assess

the incidence of new episodes, events or

comorbidity within a specific period of time.

Examples of clinical indicators are reported in Table

1 highlighting: 1) the type of indicator, 2) the source

class (based on the Figure 1 conceptual model), and

3) an example of the indicator based on the IGEA

project dashboard (Maggini, 2009). Moreover, for

each example the attribute to be taken into account

to extract the relevant information is reported.

Process indicators are assessed considering the

proportion of patients that have been treated in a

specified number of visits/controls/measurements

over the total number of diabetic patients involved in

the program. From the EHR/VHR point of view

these indicators can be measured considering on the

one hand the contact type (e.g. GP-patient encounter

to assess, for instance, the proportion of patients

who carried out at least two GP visits in one year)

and on the other hand a specific therapy used to treat

a patient’s health problem (e.g. proportion of

patients treated with antihypertensive drug).

Intermediate outcome indicators are assessed

considering the proportion of patients that have

values for a clinical parameter (e.g. micro

albuminuria, glycated hemoglobin, central arterial

pressure, weight) within a relevant threshold over

the total number of diabetic patients involved in the

program with at least one measure of that parameter.

This type of indicators is measured similarly to the

process indicators thus considering a specific

healthcare service (e.g. laboratory test) provided to

measure a metabolic parameter (e.g. glycated

hemoglobin). Another type of intermediate

indicators is related to the patient change in life style

(e.g. number of patients that has reduced the number

of cigarettes smoked during a day) and is assessed

considering the patient summary document

established and updated by the GP to snapshot, at a

specific time point, the health status of a patient with

a given pathology. Final outcome indicators are

measured considering the proportion of patients that

have suffered an event during a specific period of

time over the total number of patients who did not

suffer from the same event at the beginning of the

period.

Table 1: Examples of clinical indicators.

Indicator Concept / Class Example

Process indicators

Services

accessibility

HCService

Proportion of patients with at least two GP visits in a year

 HCService.Type: GP visit

Treatment of

medical condition

HCService

Proportion of patients treated with antihypertensive drug

 HCService.Type: Pharmacotherapy

 Pharmacotherapy.Type: Antihypertensive drug

Intermediate outcome indicators

Services outcome

monitoring

HCService

Laboratory test

Proportion of patients whit hemoglobin < 7%

 HCService.Type: LaboratoryTest

 LaboratoryTest.Measure: Hemoglobin

 Measure.Value: < 7%

Lifestyle behavior

changes

Document

Patient summary

Proportion of patients who reduced the number of cigarettes smoked

 Document.Type: PatientSummary

 PatientSummary.Entry: CigarettesSmoked

 Entry.Value (to be compared over time)

Final outcome indicators

Services

accessibility

Episode of Care

HCService

Hospitalization

Proportion of patients hospitalized by episode category

 HCService.Type: Hospitalization

 EpisodeOfCare.Type

Episode of care

observation

Episode of Care

Incidence of proliferative diabetic retinopathy

 EpisodeOfCare.Type: Diabetic retinopathy

AClinicalDataWarehouseArchitecturebasedontheElectronicHealthcareRecordInfrastructure

291

Figure 3: EHR high-level Entity Relationship diagram.

From the EHR/VHR point of view final outcome

indicators are mainly assessed considering two EHR

concepts: 1) the episode of care (e.g. incidence of

proliferative diabetic retinopathy) and 2) the

healthcare service (e.g. proportion of patients who

accessed the emergency room) that is usually

calculated considering the related disease (e.g.

episode of care classified using the ICD

classification).

2.4 Business Process Modelling

According to the dimensional modelling framework

the first step of the data warehouse design was to

identify the business processes to be modelled. In

this paper the attention is focused on the treatment of

medical condition to define process indicators, such

as the percentage of patients treated with

antihypertensive drug. This business process has

been used as a starting point to identify the main

tables and attributes to be used to design the high

level Entity Relationship (ER) diagram reported in

Figure 3. This diagram is based on the conceptual

model proposed in Figure 1 where essential concepts

are further identified, such as Patient who interacts

with the healthcare Provider that can be either an

Organization (e.g. the diabetologic centre) or a

Physician (e.g. the GP). Each Physician can also

belong to an Organization (e.g. the radiologist who

works in an hospital). Moreover, the HCService

describes the different activities performed during

each Contact. A HCService can be for instance a

Laboratory test or a Pharmacotherapy. Given a

portion of the ER schema needed to properly

describe the investigated business process the

attribute tree has been defined and afterward pruned

and grafted in order to remove the unnecessary

levels of detail such as, irrelevant attributes or

relationships (Figure 4). Starting from the attribute

tree a dimensional fact model was defined (Figure 5)

with the following characteristics: a) Fact table

(TreatmentIndex) is based on the atomic data

extracted from the structured clinical Document and

stored in the VHR that describe drugs delivered to a

patient to treat a specific episode of care; b) four

dimensions have been detected: Drug (type of

pharmacological product delivered by the

pharmacists to the patient, e.g. antihypertensive),

Patient, Date (when the drug has been delivered)

and District (where the drug has been provided to

the patient, i.e. the pharmacy district). To determine

costs attributable to a particular drug or treatment

and related to a specific episode of care the model

depicted in Figure 5 includes an additional

dimension (EpisodeOfCare) and two measures: the

quantity of product sold and the price for each unit

of product. These information are provided by the

Italian Medicines Agency (AIFA) that is the national

HEALTHINF2014-InternationalConferenceonHealthInformatics

292

authority responsible for drugs regulation in Italy.

Figure 4: Attribute tree after pruning and grafting.

Figure 5: Dimensional model based on service

accessibility and outcome monitoring business processes.

Figure 6: Diabetes dashboard displaying clinical indicators

based on the IGEA project (see paragraph 2.2).

A diagram picturing a hypothetical dashboard for

diabetes patients is reported in Figure 6. It provides

a snapshot of a set of clinical indicators to monitor

the effectiveness of healthcare service delivery in a

benchmarking framework.

3 CONCLUSIONS

In this paper we demonstrated the feasibility of

secondary uses of EHR to develop an enterprise data

warehouse architecture in a clinical governance

framework. To our knowledge this is a novel

approach that intends to exploit the entire EHR

infrastructure to develop a Business Intelligence tool

that supports the evaluation of healthcare activities

under different points of view. As shown in the case

study reported in this paper this approach can be

used for instance to benchmark local/national

healthcare services as well as to monitor the

effectiveness of applied guidelines. The EHR

represents the core of the proposed architecture as it

entails different advantages. First of all, it ensures

data quality using standardized data and documents

already integrated in a health infrastructure. In

particular, HL7 CDA used to exchange structured

clinical documents makes it easier to retrieve and

process atomic data that are usually coded using

international nomenclatures and dictionaries such as

ICPC (International Classification of Primary Care),

SNOMED (Systematized Nomenclature of Medicine

Clinical Terms) and ICD (International

Classification of Diseases). This also supports

semantic interoperability between different source

systems. Data are provided using a publish-subscribe

paradigm that guarantees health data to be promptly

exchanged at the moment an event is published, with

an automatic detection of relevant information

directly from source systems. This ensures a timely

and continuously updated information flow as well

as data integrity and consistency based on a sample

size that covers the entire target population.

Moreover, this approach makes it possible to

gradually integrate other applications within the data

warehouse infrastructure such as geographical or

socio-demographic information systems. EHR

represents the operational data store that guarantees

the separation between the transactional and the

analytical processing. The feasibility study described

in this paper will be implemented in a regional EHR

environment.

REFERENCES

Botsis T., Hartvigsen G., Chen F., Weng C., 2010.

Secondary use of EHR: data quality issues and

informatics opportunities. In AMIA Summits

Translational Science Proceedings pp. 1–5.

Ciampi M., De Pietro G., Esposito C., Sicuranza M.,

Donzelli P., 2012. On Federating Health Information

AClinicalDataWarehouseArchitecturebasedontheElectronicHealthcareRecordInfrastructure

293

Systems. In International Conference on Green and

Ubiquitous Technology (GUT), pp. 139-143.

Committee on Data Standards for Patient Safety, 2003.

Key Capabilities of an Electronic Health Record

System. Institute of Medicine Report 5.

Contenti M., Mercurio G., Ricci F.L., Serbanati L.D.,

2008. LuMiR: A Region-wide Virtual Healthcare

Record. In Proceeding of the 9th International HL7

Interoperability Conference. Crete, Greece. pp. 80-3.

De Mul, M., Alons, P., Van der Velde, P., Konings, I.,

Bakker, J., Hazelzet, J., 2012. Development of a

clinical data warehouse from an intensive care clinical

information system. Computer methods and programs

in biomedicine 105 (1) pp. 22-30.

Diomidous M., Zimeras S., Mantas J., 2009. Spatial

Electronic Health Record for the epidemiological

clinical data. Travel Health Informatics and Telehealth

1, pp. 66-72.

Dolin R. H., Alschuler L., Boyer S., Beebe C., Behlen F.

M., Biron P. V., Shvo A. S., 2006. HL7 clinical

document architecture, release 2. Journal of the

American Medical Informatics Association 13 (1) pp.

30-9.

EN 13940-1:2007 Health informatics - System of concepts

to support continuity of care - Part 1: Basic concepts.

CEN 2007. British Standards Institution.

Gillespie G., 2000. There’s gold in them thar’ databases.

Health Data Management 8 (11) pp. 40-52.

Hussain S., Ouagne D., Sadou E., Dart T., Jaulent M. C.,

De Vloed B., Daniel C., 2012. EHR4CR: A semantic

web based interoperability approach for reusing

electronic healthcare records in protocol feasibility

studies. In Proceedings of the 5th International

Workshop on Semantic Web Applications and Tools

for Life Sciences. Paris, France.

Inmon W., 1999. Building the Operational Data Store.

John Wiley & Sons. New York. 2

edition.

Jensen P. B., Jensen L. J., Brunak S., 2012. Mining

electronic health records: towards better research

applications and clinical care. Nature Reviews

Genetics 13 (6) pp. 395-405.

Kamal J., Pasuparthi K., Rogers P., Buskirk J., Mekhjian

H., 2005. Using an information warehouse to screen

patients for clinical trials: a prototype. In Proceeding

of American Medical Informatics Association

Symposium. Washington DC, USA. pp. 1004.

Koh H. C., Tan G., 2011. Data mining applications in

healthcare. Journal of Healthcare Information

Management 19 (2) p. 65.

Kush R. D., Helton E., Rockhold F. W., Hardison C. D.,

2008. Electronic health records, medical research, and

the Tower of Babel. New England Journal of Medicine

358 pp. 1738–40.

Lurio J., Morrison F. P., Pichardo M., Berg R., Buck M.

D., Wu W., Calman N., 2010. Using electronic health

record alerts to provide public health situational

awareness to clinicians. Journal of the American

Medical Informatics Association 17 (2) pp. 217-9.

Maggini M., 2009. IGEA-A chronic disease management

project for people with diabetes. Annali Istituto

Superiore di sanità 45 pp. 349-352.

Mettler T., Rohner P., 2009. Supplier Relationship

Management: A Case Study in the Context of Health

Care. Journal of Theoretical and Applied Electronic

Commerce Research 4 pp. 58-71.

Oemig F., and Blobel B., 2005. Does HL7 Go towards an

architecture standard? Studies in Health Technology

and Informatics 116 pp. 761-6.

Roque F. S., Jensen P. B., Schmock H., Dalgaard M.,

Andreatta M., Hansen T., Søeby K., Bredkjær S., Juul

A., Werge T., Jensen L. J., Brunak S., 2011. Using

Electronic Patient Records to Discover Disease

Correlations and Stratify Patient Cohorts. PLoS

Computational Biology 7 (8), pp. 1-10.

Roth M. T., Schwarz P., 1997. Don’t scrap it, wrap it! a

wrapper architecture for legacy data sources. In

Proceeding of the Conference on Very Large Data

Bases (VLDB). Athens, Greece. pp. 266–275.

Safran C., Bloomrosen M., Hammond W. E., Labkoff S.,

Markel-Fox S., Tang P. C., Detmer D. E., 2007.

Toward a national framework for the secondary use of

health data: an American Medical Informatics

Association white paper. Journal of the American

Medical Informatics Association 14 (1) pp. 1-9.

Sahama T. R., Croll P. R., 2007. A data warehouse

architecture for clinical data warehousing. In

Proceedings of the fifth Australasian symposium on

ACSW frontiers 68. Ballarat, Victoria, Australia. pp.

227-32.

Schadow G., Mead C. N., Walker D. M., 2006. The HL7

reference information model under scrutiny. Studies

Health Technology Informatics 124 pp. 151–6.

Stow P. J., Hart G. K., Higlett T., George C., Herkes R.,

McWilliam D., Bellomo R., 2006. Development and

implementation of a high-quality clinical database: the

Australian and New Zealand Intensive Care Society

Adult Patient Database. Journal of Critical Care 21

(2) pp. 133-41.

Taylor P., 2008. When consent gets in the way. Nature

456 pp. 32–3.

Wang X., Hripcsak G., Markatou M., Friedman C., 2009.

Active Computerized Pharmacovigilance using

Natural Language Processing, Statistics, and

Electronic Health Records: a Feasibility Study.

Journal of the American Medical Informatics

Association 16 (3) pp. 328-37.

Wickramasinghe N., Schaffer J.L., 2006. Creating

knowledge-driven healthcare processes with the

intelligence continuum. International Journal of

Electronic Healthcare 2 (2) pp. 164-74.

Zhou X., Chen S., Liu B., Zhang R., Wang Y., Li P., Guo

Y., Zhang H., Gao Z., Yan X., 2010. Development of

traditional Chinese medicine clinical data warehouse

for medical knowledge discovery and decision

support. Artificial Intelligence in Medicine 48 (2) pp.

139-52.

HEALTHINF2014-InternationalConferenceonHealthInformatics

294