You are on page 1of 9

Dimensional Model (taken Irom http://hosteddocs.ittoolbox.com/KP90806.

pdI):








;er;iew
%oday in medical Iield, a health care institution must spend a lot oI money to maintain the
operating cost. the main goal by using this database is to reduce the operating cost while
maintaining the proper treatment to the patient.
Reduce operating costs at all levels:
O ost oI healthcare ProIessionals
O ost oI lab equipment & consumables
O ost oI pharmaceuticals / medical material
O ost oI a treatment per Diagnosis related grouping (DRG)
O ost per type oI medical intervention (e.g. speciIic medical operation)
On the other hand, an acceptable level oI patient treatment involves:
O vidence based medicine, accurate diagnosis and eIIicient treatment
O On time admittance in the Hospital and healthcare treatment
O %reatment with respect Ior the Patient- analysis oI options
O Reduction oI risks during treatment (e.g. related to the use oI medicine, biomedical
equipment, blood transIusions)
O apture oI medical history oI the patient in order to support evidence based medicine
Moreover, goals oI Healthcare Institution are:
O Reduction oI medical errors and exposure oI the patient to medical hazards (e.g.
inappropriate levels oI radiation)
O $upport medical research with patient & treatment data
O Participate and support a larger Healthcare system, with the exchange oI medical
inIormation on a patient, as well as statistics on population morbidity & mortality.

Model and Description


%he data warehouse would store the dimensions involved in the service: Patient, Date,
Healthcare Unit and Department, exit DRG (diagnosis related grouping) group, Diagnosis (based
on ID), Medical intervention based on selected codiIication, pharmaceutical treatment, medical
material used, and capture all cost related Iacts in the Iact table.
Based on this star schema, a wide range oI analytical tasks can be carried out. Analysis related to
medical treatment oI a speciIic event (indicatively) to produce a medical history report on a
patient, produce stats on morbidity by restricting on speciIic ID codes (e.g. Irequency oI an
ID as a percentage oI total events, ID related to demographic inIo like age or educational
level), analyze the relation oI medical interventions to ID & DRGs, by restricting on speciIic
DRGs, and Analyze mortal events per ID / DRG (this can be coded in a speciIic xitdiagnosis
value)
Analysis related to cost incurred by a speciIic event (indicatively):
O Analysis on the average length oI stay (ALO$) per DRG as well as per Patient
demographics.
O Analysis on the medical material & consumables cost related to a medical intervention
O Analysis on the medical material & consumables cost related to an event oI speciIic DRG
O Analysis on the pharmaceutical cost related to an event oI speciIic DRG
O Delays in payment and collection levels
And Ior department related analysis:
O umber oI treatment events / medical interventions per Department
O comparison oI same specialty Departments
%he star schema could also capture the involvement and perIormance oI medical ProIessionals.
In addition it could capture many more Iacts related to the event like man-hours spent by the
Doctor in charge, by the other proIessionals involved.

%he datamart can be linked to the datamart oI another Hospital or Healthcare system, iI the two
use conIormed dimensions and Iacts. $tandardized codiIications (like ID) aim at achieving this
conIormance, in order to be able to consolidate inIormation at the regional, national and
international level.
Why develop the star schema, instead oI querying directly the Hospital InIormation $ystem
database. For two important reasons:
O %he star schema model is easily understandable by the Business Analyst
O %he analysis on the star schema is computationally more eIIicient (the symmetric
structure`s simplicity allows Ior better query optimization)
Moreover the star schema incorporates enriched inIormation at the dimensions as well as derived
Iacts on the Iact table (cost, duration). %his additional inIormation is produced and appended in
the data mart staging area.
Business intelligence inIrastructures, like the one presented above can Iacilitate the analysis and
continuous improvement in a Healthcare system.

$taff
List oI staII that required Ior the project and the responsibilities :
Manager/Director
%he project manager is responsible Ior day-to-day management oI project tasks and activities,
including resource coordination, status tracking, and communication oI project progress and
issues, working closely with the business project lead.
Project Manager
Project Managers must deliver commitments and must deliver on time. %hey will do this by
culling resources Irom within the data warehouse team and Irom consultancy as necessary and
establishing partnerships with other internal support organizations required to support a data
warehouse iteration.
Chief Architect
%he Manager/Director will need to rely on a hieI Architect (or similar title) position, as one oI
his/her direct reports, to work on complex issues oI architecture, modeling, and tools.
%he person should be able to quickly qualiIy as an authority in data warehousing within the
organization and have mastery oI the data warehousing paradigm, both current and emerging
technologies.
Data $teward
Data $tewardship appointment should be made at the subject area to management-level
personnel in the business areas most impacted by the subject area. For the core subject areas, this
will be the person on the working team by appointment oI the executive sponsor.
Data stewards, unlike other users, have read and write access to their area oI stewardship. Other
users have read-only access to the Data Warehouse areas that they have been approved to read by
the data stewards.

ecuti;e $ponsor
%he Data Warehouse must have high-level and sustainable sponsorship. %he xecutive $ponsor
must be politically viable and be able to garner and retain adequate resources Ior the construction
and maintenance oI the Data Warehouse.

Database Administrator
A key design point Ior a data warehouse group is the placement oI the database administration
Iunction and the division oI roles and responsibilities between the support group and the user
community.
Data Administrator
A key design point Ior a data warehouse group is the placement oI the data administration
Iunction and the division oI roles and responsibilities between the DBAs and the data
administrators. %he Data Warehouse Data Administrator will translate the user requirements into
a logical database design.
Application Programmer - % $pecialist
%he Data Warehouse Application Programmer is responsible Ior applying transIormation rules as
necessary to keep the data clean and consistent and thereIore usable by the user community.
$peciIic tasks include:
O Participation in design sessions chaired by data stewards and/or I% personnel where
decisions are made involving the transIormation Irom source to target
O Programming the data acquisition tool with the rules to be applied to the data
O nsuring the correct application oI the business rules through data query aIter the data is
loaded into the Data Warehouse

AP $pecialist
%he OLAP $pecialist is responsible Ior the management oI the deployment, maintenance, and
user support oI the data access tools.
$stem Administrator
%he Data Warehouse $ystem Administrator will be responsible Ior data destined Ior the Data
Warehouse Irom the operational systems that store the data.

Data Warehouse Infrastructure


For the inIrastructure, the data warehouse will using an Open system servers with on-UniIorm
Memory Architecture. Open system, or UIX, servers are the primary platIorm Ior most
medium-sized or larger data warehouses today. UIX is generally robust enough to support
production applications, and it was adapted Ior parallel processing more than a decade ago. %he
UIX server market is Iairly commoditized.
on-UniIorm Memory Architecture (UMA) is essentially a combination oI $ymmetric
Multiprocessing ($MP) and Massively Parallel Processing (MPP) in an attempt to combine the
shared disk Ilexibility oI $MP with the parallel speed oI MPP. %his architecture is a relatively
recent innovation, and it may prove viable Ior data warehousing in the long run. UMA is
conceptually similar to the idea oI clustering $MP machines, but with tighter connections, more
bandwidth, and greater coordination among nodes.
%he data warehouse will use roles and privileges Ior data warehouse security. %he purpose is to
provide basic level oI database security. %hey are designed to control user access to data and to
limit the kinds oI $"L statements that users can execute. Roles are groupings oI privileges that
you can use to create diIIerent levels oI database access.
%he server will be located in a secure and limited access environment to protect the server Irom
the physical vulnerabilities such as theIt, intentional or accidental destruction, natural Iorce,
technological obsolescence, and hijacking control oI an asset.
%o protect against loss oI inIormation assets, the server will provide with the useIul Iunction
such as backup and restore data. %he server soItware should be under control oI the group that
manages the server and located in machine room under lock and key. the inIormation Ilow will
be encrypted and monitored.

#eferences
http://hosteddocs.ittoolbox.com/KP90806.pdI (accessed $eptember 2011)
Inmon, William (1999). Building the Operational Data $tore, second dition. John Wiley &
$ons.
Kimball (2007). "%he Data Warehouse LiIecycle %oolkit". Wiley.
McKenzie, Diana (1998)."A Little Loss oI Privacy may be Good Ior Patients' Health".
McKnight Associates, Inc (2000). "IIective Data Warehouse Organizational Roles and
Responsibilities".
Oracle(2007). "$ecuring a Data Warehouse".http://download.oracle.com/docs/cd/B2835901/
server.111/b28314/tdpdwsecurity.htm

You might also like