Transcript System Design

MAHI Research
Database
Project Directions
June 7, 2001
Overview of Client Requirements
• Quaterly reports
(STS, ACC)
• Data to/from external
organizations
Management
Biostatisticians
IS
MAHI
Research
Database
Physicians
• General patient data
• Simple queries
• Fast & reliable queries
• Add new data points
Patients
• Efficient data storage
• Internal security
• Extensibility
• Maintenance
Concept Diagram of MAHI Architecture
(designed by: Kelly Kerns)
Management Database
(contains metadata)
M
Generates user interface
using DHTML, ASP
Patient
Processing, normalization
using ATL, COM, C++
Primary customer:
Physicians
MAHI DB Interface
(HTML?)
SAS
Ad-Hoc
Query
HL7
SPSS
ODBC
T
Q
R
A
ODBC
ASCII
ASCII
HL7
Change Request
/Record Editor
ACC
Cardiff
(optical recognition
software)
DTS
L
Legacy database (no
quarantine/data
cleansing)
STS
Currently manually
implemented using SQL
scripts
Digital Sender
Digital form
scanner/transmitter
(model HP 9100C)
DB Explorer
user interface
Interim Solution
Current Implementation
Status:
Interim solution in place
Angioplasty and Cardiac
Cath data available
Simple queries available
via DB Explorer
Limitations:
Missing Surgical data
(legacy database)
Unable to perform
quarterly reports
No single repository to
store common data
No simple analytic tools
built-in
Why are we using a database?
To expose legacy data
To record/store/maintain/access patient and treatment
information
To allow extraction and investigative analysis of above
information
To meet clinical organization IS standards
To collect and store data from external entities
Storage and Processing
MAHI’s storage and processing needs are datacentric (as opposed to document-centric),
related to dynamic procedures, and patientoriented.
Storage and Retrieval technologies
Database (relational, OO, hierarchical) and middleware
(built-in or 3rd party)
Data warehousing
XML server
- Native XML database to manage large numbers of XML
documents
XML-enabled web server
- Web server that can build XML documents from data in a
database
Database features of XML
Storage (XML documents)
Schemas (DTDs, XML Schema)
Query language (XQuery)
Programming interface (SAX, DOM)
Why XML as storage format? (contd.)
System needs to store data with repeating fields

Example:
Caregiver has several patients (some repeat patients)
Location can be used for multiple encounters
Patient may be administered multiple materials/medication (some
prescriptions are repeated)

XML provides mechanism to quickly parse and extract data with
repeating structure, or to generate new documents.
System needs to store data with arbitrary length

Example:
Text of medical notes

XML can store nested elements of arbitrary depth and length.
Why XML as storage format? (contd.)
System needs to exchange information with other
organizations

Example:
External Stroke database, external Women’s Health Clinic
database, external HL7 compliant database; currently this
formatting has to be done manually because data formats are
different.

XML can be used to make formatting easier (similar data items
will have similar markup tags; more human readable), and
possibly automate some of the work by using DTDs/Schemas as
reference.
Why XML as storage format? (contd.)
System needs to store standard queries/analytical
results for display on multiple end-user client apps

Example:
Currently, Kelly and biostatisticians have to make specialized query
every time it is requested by a user.

Standard queries can be stored as SQL procedures, and query
results can be returned in a single XML format document (with its
associated DTD) to be transformed into views for multiple endusers (e.g. via XML-enabled web server).
Why XML as storage format? (contd.)
System needs to store metadata

Example:
Medical groups access to information, originating source of data,
etc.


Metadata can be easily incorporated into XML documents as
attributes in the element tags.
Example:
<Caregiver id=“1234” medgroup=“MHI”>
<name> … </name>
<ssn> … </ssn>
<type> … </type>
</Caregiver>
Limitations of XML as storage format
Efficient storage?
Security?
Transactions and data integrity?
Multi-user access?
Recovery and fault-tolerance?
Queries across multiple documents?
Others?
An XML extension to existing architecture
Current
architecture
Input data
Q
R
I
Adapter
Adaptable front-end
apps for Query/Report
Extender
Neutral
storage facility
Q/R
XQ
XR
XML
H
Converter
Legacy
storage
Legacy
(e.g. Surgical
database)
Report Harvester (e.g.
ACC, STS)
Key tasks (initial thoughts)
Develop storage facility for XML documents
Design schema/DTD for data to be converted and stored
Develop retrieval (via query) facility for searching,
indexing, extraction, and analysis of documents
Perform data conversion of legacy database
Integrate into existing architecture
Modified Project Tasks
Re-visit the tables in the Repository data store

Re-engineer
Validate structure of Repository data model
Develop a X-Quarantine using XML (from Legacy & New
Data)
Develop a process from X-Quarantine to Repository
Develop a process from Repository to X-Repository
Query tool (for Bio-Statistician) from X-Repository
Other Issues
Ensure X-Repository information is ‘secure’ (i.e. biostatisticians can ‘trust’.)