Thanks to rapidly falling disk costs, companies can now afford to store trillions of bytes of valuable information. However, data professionals know that storing the data is just the foundation and the real challenge is making sense from these huge data stores and mining them for statistically significant events. This is an area where Oracle excels, and Oracle customers can now store unprecedented amounts of information and use Oracle tools to mine gold from these vast arrays of data.
Scientists know that valid research requires scanning through gigabytes of data to locate significant correlations, often using heuristic models to find important relationships between data items. Powerful tools such as Oracle Discoverer and Oracle Data Mining (ODM) improve the quality of health care, and these tools are helping Oracle scientists to save lives.
Inside Oracle medical informatics
Nowhere is this life-saving research as prevalent as it is in the area of medical informatics, where highly skilled doctors are leveraging patient treatment information to perform advanced analytics such as predictive modeling, measuring the efficacy of life-saving drugs and detecting dangers that have a direct impact on the health and safety of millions of patients.
Oracle medical informaticists require the most scientific training of all Oracle professionals, and it is a database role that requires knowledge of both medicine and advanced statistics. Today's clinical databases store detailed information about patient diagnoses, lab test results and details from patient treatments, a virtual gold mine of information for medical researchers. Utilizing data mining techniques with medical treatment data is a virtually unexplored frontier.
According to Mike Ault, a recognized expert in Oracle data warehousing, Oracle Discoverer and data mining tools are a godsend to medical researchers:
Oracle Discoverer and ODM give doctors the tools that they need to find drug contraindications and to measure the efficacy of treatment regimens. I've seen informaticists develop medical decision support systems that allow doctors to drill-down and show details that are impossible with ordinary query tools. If you want to see how something works in the real-world, these are the tools that will get you there.
While scientific research and clinical trials are essential in discovering safe and effective drugs, the recent drug recalls by the U.S. Food and Drug Administration highlight some of the unavoidable risks that are associated with clinical trials. Empirical evidence suggests that the only way to fully understand a system of complex drug interactions is to sample and analyze real-world data.
Whether scientists are researching the behavior of drugs, studying the demand for products or learning about the behavior of software, the application of heuristic techniques has answered the question of "how things work." Medical informaticists specialize in using Oracle Discoverer, Oracle OLAP and Oracle Data Warehouse Builder, and many are also experts in multivariate statistics using statistical programs such as SAS and SPSS. The new release of Oracle 10g has incorporated similar statistical functions allowing fast and powerful analytical capabilities for hypothesis testing.
Using powerful Oracle tools such as Oracle Discoverer and Oracle Data Mining, medical researchers are discovering trends and patterns that will improve the health care for millions of people around the globe.
We are also starting to see medical experts combining their medical qualifications with Oracle database certification, as with Dr. Timothy Wu, MD OCP, an Oracle scientist working for the Walter Reed Medical Center in Washington, DC.
Dr. Wu says that Oracle tools give medical doctors a tool to help them quickly make sense of vast clinical databases. "Understanding the complex relationships that occur among patient symptoms, diagnoses and behavior is one of the most fascinating areas I have studied," says Dr. Wu. "I use Oracle's advanced analytical features to drill from the overall summary at the hospital level to the patient-level detail."
Mining gold from Oracle
Scientists can now search for important facts within huge clinical databases, a task that would be virtually impossible without tools such as Oracle Discoverer. The concepts behind data mining have been around for decades, but they are gaining popularity because cheap disks now make it economically feasible to have instant access to vast databases. In the area of Oracle Business Intelligence, data mining can save companies millions of dollars each year by targeting advertising to consumers who are the most likely to want their products, but these techniques are also being used to improve the quality of health care. For example, knowing that patients with specific demographics have a predilection for certain kinds of diseases can lead to preventative measures to control certain kinds of cancer.
The Oracle Data Mining (ODM) product is one of the most sophisticated tools for understanding "how things work" in the real world. Using advanced data mining techniques, ODM scans trillions of bytes of data seeking out statistically significant correlations and presents heuristics, all based on probabilities. One of the most powerful aspects of ODM is that the data stays in the database and allows users to access the results with reporting tools, display the results in a portal or conduct further studies of the data using OLAP.
This approach has been so successful that many Oracle Data Warehouse customers report a payback period of less than a year, with direct immediate savings from the insights revealed by the Oracle tools. It's also the stuff of legend and urban myths, such as the beer and diapers legend that is used by college professors to explain the huge benefits of data mining technology:
A number of convenience store clerks, the story goes, noticed that men often bought beer at the same time they bought diapers. The store mined its receipts and proved the clerks' observations correct. So, the store began stocking diapers next to the beer coolers, and sales skyrocketed. The story is a myth, but it shows how data mining seeks to understand the relationship between different actions.
As George Santayana said, "Those who forget the past are condemned to repeat it," and this is the guiding principle for making forecasts with Oracle data.
When reviewing trends in historical data, it is often advantageous to discard outliers, or data points that lie outside expected parameters. In health care, the outliers are often the most interesting cases and bear further investigation. The Oracle data mining development team has released a new algorithm in 10g release 2 that can find these "rare cases," which could correspond to unusual symptoms that may foretell the outbreak of disease.
"Health care is the virtually untouched area where higher analytical tools can be used to find solutions that will benefit not only ourselves, but our children and generations that follow," says Jacek Myczkowski, Oracle's Vice President of Data Mining Technologies and Life Sciences, who spoke at DCI's Business Intelligence and Data Warehousing Conference recently. Dr. Myczkowski represented Oracle in the DM Review's World Class Solution Awards, where Oracle partnered with Walter Reed as one of three finalists in the Business Intelligence category. He went on to say that health care is one of the biggest businesses in the world today, and that money invested in learning how to improve patient care will have greater impact on society as a whole than in marketing, retail or sales data.
Data mining helps tune Oracle databases
Tools such as Oracle Discoverer and Oracle Data Mining are now being applied to dozens of types of applications, even including Oracle's own metadata. The Oracle 10g Automatic Workload Repository (AWR) captures information about Oracle's performance over time, and this wealth of performance information is revolutionizing Oracle tuning.
Don Burleson, author of the book "Oracle tuning: The definitive reference," notes that Oracle professionals are applying the same principles of medical informaticists and that it's almost impossible to reproduce performance issues in a lab environment:
"A busy Oracle database changes so rapidly that trend-based analysis is the only solution for tuning transient Oracle performance problems. The Oracle AWR keeps details on how tables are accessed, what indexes are used and a wealth of other information that is critical to any successful running effort."
Oracle Data Mining has also been shown to help medical informaticists select attributes for an OLAP cube, as was demonstrated at the recent Oracle Open World. When the analyst is faced with selecting from a vast array of characteristics about health relating to some measure of outcome, such as patient age, gender, weight, blood pressure, blood fat level and so on which impact acquiring diabetes, for example, ODM tools can be used to find the most important attributes and narrow the selections. This feature reduces the dimensionality of the OLAP cube, making it far easier to understand the relationships among the various demographic attributes, and enhances performance of the query tool.
Conclusion
This short article is just an introduction to how Oracle software is saving lives, and I'm both thrilled and honored to have a job where I can apply my skills to help improve the lives of others. It would be very difficult without Oracle tools, and I'm very proud that Oracle has invested in making these powerful tools that enable us to help others.
Whereas data mining is being used to reduce churn, when for example mobile phone users switch to alternative wireless businesses, or to increase the return on investment for companies, applying business intelligence in the life sciences to improve the quality of health care has tremendous benefits for all of us.
About the Author
Dr. Carolyn Hamm is a recognized expert in Oracle data warehouse technologies, advanced analytics and Oracle data mining. Earning her Ph.D. in Experimental Psychology, Dr. Hamm has spent the past eight years developing Web-enabled data systems for population health, accessed by research, clinical and administrative staff. Dr. Hamm is the author of the forthcoming book "Oracle Data Mining: Mining gold from your warehouse."
