Submit your article of up to 500 words in length, subject to editing for clarity and brevity (please note: I run only original articles that have not appeared on any Web site or in any publication and I can’t use anything that looks like a commercial pitch). I’ll use a phony name for you unless you tell me otherwise. Thanks for sharing!
Goodbye Data Warehouse and Cubes, Hello AQL
By Mark Moffitt
For the last two years, I have been researching systems to replace the data warehouse used for report-writing in our organization. This effort has been driven by the desire to provide better service to other departments that rely heavily on data reporting for day-to-day operations.
The idea is to push data to users so they can perform in-memory analysis and display of large amounts of data, a system that would replace the current process of requesting custom reports and spreadsheets from the information services (IS) department. The current process requires considerable resources in the IS department and requests can take several days if the number of requests for reports in the queue becomes large.
The requirements for a new system are straightforward, but somewhat daunting:
1. Put data into users’ hands so they can perform business intelligence.
2. The cost of the system, including license, hardware, and consulting, must be offset by the direct costs of shutting down existing systems.
At GSMC we operate Meditech Magic and use a data warehouse for analytics and business intelligence. The data warehouse stores about nine years of financial data in about 650 GB. The data in the warehouse is updated nightly. SQL reports have been developed to provide reporting across the organization.
IS at GSMC is bombarded with requests for new reports. These requests come in the form of specialized requests for data that often require modifying an existing SQL query or writing a new query. The process is iterative that starts with gathering requirements for a report, modifying or writing new SQL queries, generating a report and sending it to the customer.
Typical turnaround times are variable and are highly dependent on the number of reports in the queue to be developed. Best case scenario is four hours, typical is two to four days. Often the customer will, upon review of the report, ask to include or exclude specific data. This back-and-forth typically occurs several times until the report meets the customer’s needs.
The IS department at GSMC has several analysts who spend a good part of their time responding to requests for data. It is a never-ending demand.
We researched the use of OLAP (online analytical processing) cubes to provide data to users. The advantages of cubes is well documented and includes the ability to drill down to details and analyze data in ways simply not possible with reports or spreadsheets. The disadvantage to cubes is that data must first be aggregated. If a user needs data not included in the cube, then the cube must be rebuilt. Also, a data warehouse is required. Finally, building and maintaining cubes require personnel with specialized skills.
About seven months ago, I read on HIStalk about a new company named QlikView. I researched the software and it sounded too good to be true. However, I was intrigued that QlikView doubled revenues in 2008, not an especially good year for selling enterprise software as the national economy was in a major recession.
On the surface, QlikView is a business intelligence solution that consists of a data source integration module, analytics engine, and user interface. QlikView is based on AQL and is completely different from other OLAP tools.
Through AQL, QlikView eliminates the need for OLAP cubes and a data warehouse, replacing the cube structure with a Data Cloud. A Data Cloud does not contain any pre-aggregated data but instead builds non-redundant tables and keeps them in memory at all times. Queries are then created on the fly and are run against the Data Cloud’s in-memory data store.
Under AQL, all data is stored only once, and all data associations are stored as pointers, so a Data Cloud database becomes more efficient at retrieving records than do OLAP databases. A Data Cloud database is also much smaller since records are not repeated through aggregation and its structure never has to change. The architecture allows for a flexible end-user experience because it doesn’t require aggregation or pre-canned queries that try to cover every possible analytical scenario a user can create, unlike data cubes that require both. (1)
Data Clouds run in memory and AQL reduces in-memory storage requirements by about 75% as compared to source data. In-memory Data Clouds can be stored as AQL files for archiving. AQL disk files are 90% smaller than source data. Think of an AQL file like an Excel file where data can be added and deleted and the file saved with different names for archiving purposes.
The price point for the software is about $150,000 (one-time fee) for our health system. Hardware costs are about $15,000 for a server with 98 GB of memory. We expect consulting fees to total $150,000 for a SME in hospital financial data with QlikView experience. We worked with RSM McGladrey on a consulting proposal as they have well-qualified personnel in this space.
If you know much about the BI/Analytics space, you may question the low cost of the software and consulting services. This has everything to do with the AQL model. RSM McGladrey quoted a revenue cycle effort at eight weeks and includes:
- Transfer data from existing systems to QlikView
- Data validation
- Census analysis
- AR analysis
- Insurance contract analysis
- Hindsight analysis
- Train IS staff on data extraction
The revenue cycle statement of work is only one component of the $150,000 quote for consulting services from RSM McGladrey for implementing QlikView at our organization.
The total cost for QlikView at GSMC is $315,000. That will be directly offset by shutting down a data warehouse, savings from using QlikView for analytics versus another system where the cost of consulting services had already been quoted and budgeted, and other savings. We expect additional direct benefits from having deep analytic capabilities with our revenue cycle data.
QlikView has a number or healthcare customers. I believe you will be hearing more about the company in healthcare in the years ahead as they achieve market awareness of QlikView software’s capabilities and price point.
We have not yet purchased the package. If we do, I’ll write a follow-up article on our experience.
1 “Qliktech, IBM Provide New View Of OLAP”, Mario Morejon, Technical writer for ChannelWeb, July 18, 2003, http://www.crn.com/software/18839582
Mark Moffitt is CIO at Good Shepherd Medical Center in Longview, TX.
Humpty Dumpty Leaves Wonderland to Visit Health Information Technology
By Jim Kretz
Suppose I told you that “voting” henceforth would mean you would only be shown a ballot, period. No more selecting your preferred candidate.
Now suppose I told you that your consent to disseminating clinical information did not mean your granting permission, but only your acknowledgement that you saw my information policy — take or leave it. This may remind you of Humpty Dumpty’s scornful assertion, “When I use a word it means just what I choose it to mean — neither more nor less.”
The authors of this paper — the American College of Cardiology, the Healthcare Information and Management Systems Society, and the Radiological Society of North America — are not mean-spirited, uninformed, or confused. What could result in their clearly having tumbled into a conceptual rabbit hole?
Jim Kretz is project officer at the Substance Abuse and Mental Health Services Administration of the Department of Health and Human Services. His comments should not be construed to reflect the official position of SAMHSA.
Massachusetts HIT Conference Thoughts
By Bill O’ Toole
I had the pleasure of attending a National Conference hosted by Massachusetts Governor Deval Patrick in Boston last week. The conference was billed as Health Information Technology: Creating Jobs, Reducing Costs and Improving Quality.
Keynote addresses were provided by David Blumenthal, MD, National Coordinator for HIT and Vice Admiral Regina M. Benjamin, MD, MBA, Surgeon General of the United States. Health IT Policies and Standards were addressed by a panel that included John Halamka, MD (CIO, CareGroup and Harvard Medical School), Marc Overhage, MD (CEO, Indiana Health Information Exchange), Paul Tang, MD (CMIO, Palo Alto Medical Foundation), Micky Tripathi, Ph.D (CEO Massachusetts eHealth Collaborative) with Tim O’Reilly (President, O’Reilly Communications) moderating.
Another panel discussion on Health IT, Business Opportunities and Job Creation featured leading Massachusetts vendor executives Girish Kumar Navani (eClinicalWorks), Howard Messing (Meditech), Richard Reese (Iron Mountain), Bradley J. Waugh (NaviNet) moderated by Chris Gabrieli (Bessemer Venture Partners).
I could go on and on, but the list would be too long. I mentioned those above to give readers a sense of magnitude and to perhaps share in this small article the profound comfort I felt that "we" are doing this right. Many other highly qualified participants shared their knowledge on all things HIT- and ARRA-related.
What impressed me most was the overwhelming sense of momentum. The stimulus package and its future incentives have so far done exactly what was intended, serving as the spark that has set this massive project in motion. Remaining at the forefront of it all, though, is the goal of better medical care for all. That theme was never lost and was frequently repeated.
As one who until now has found certain parts of most conferences to be extraneous (ok, boring), I felt obliged to inform the far-flung readership of HIStalk that I was extremely impressed with every minute of this two-day conference. If the energy, knowledge, and sincere interest and enthusiasm expressed by those involved in this conference are carried forward to the project at large, then we are truly in for a remarkable change in our industry.
Congratulations to the Massachusetts Technology Collaborative and its Massachusetts eHealth Institute, the Massachusetts Health Data Consortium, and Governor Patrick for organizing this special event. It should serve as the model and be repeated whenever possible throughout the country.
William O’Toole is the founder of O’Toole Law Group of Duxbury, MA.