Healthcare and Data Mining - Technology Information

Health Management Technology, August, 2000 by Anne Milley

Using data for clinical, customer service and financial results.

To most healthcare professionals, statistics is a dirty word, one that evokes visions of penny-pinching bureaucrats interested only in saving money and balancing the books. But this is the old way of thinking. Recent developments in statistical analysis technology mean that today, number crunchers can be the physician's best friends. In particular, data mining software promises to become a major component of tomorrow's medical kit and a leading ally in the fight against illness and disease.

Data mining--the process of data selection, exploration, and building models using vast data stores to uncover previously unknown patterns--is already utilized in business by thousands of major institutions. Most of these companies use the software for customer management, which includes managing customer loyalty plans and cross-marketing campaigns.

These same techniques are contributing to huge advances in healthcare. The power and versatility of modern data manipulation techniques can improve treatment procedures, increase patient satisfaction and wipe out over-prescription, bad practice and fraud.

Finding Causes and Cures

We need only apply data mining techniques to the analysis of whole treatments for particular ailments to see its vast potential for clinical improvement. For example, an HMO could set up a regional test that collectively examines all local cases of allergies and asthma, examining the relationships between the two to find the most effective treatment regime. The data collated and compared could include lists of all available drugs--for example, antihistamines and more holistic treatments such as taking a spoonful of honey every day--with the placebo effect easily monitored within the study.

Because pollution and airborne microorganisms contribute to such ailments, the healthcare provider might also examine specific local causes and associate these with different treatment applications. By comparing and contrasting causes, symptoms and courses of treatment, data mining can deliver a precise, accurate analysis of which courses of action prove effective.

As the Centers for Disease Control and Prevention has discovered, sometimes the cause of disease persistence or resurgence after it is nearly eradicated is attributable to people who stop taking medication when their symptoms cease instead of finishing the dosage. With a comprehensive monitoring and analysis program, this data can produce a graphic representation of the progress of a disease. Researchers can then spot disease patterns in different areas and take action to address them.

Carnegie Melon University's professor William F. Eddy, Ph.D., used spatial-temporal techniques to produce a graphic representation of the course of mumps from 1968 to 1988 by state. He found that outbreaks of the disease were largely seasonal and temperature-related. The worst outbreaks occurred when cold weather forced people indoors and into prolonged close contact with each other. While this only confirmed what many knew already, by the end of the study it was possible to watch the graphic representation of the spread of mumps expand and contract as those 20 years of seasons passed.

Other data mining applications include: 1) associating the various side effects of treatments; 2) collating common symptoms to aid diagnosis; 3) determining the most effective drug compounds for treating sub-populations that respond differently than the mainstream populace to certain drugs; 4) determining proactive steps that can reduce the risk of affliction. With the ability to conduct robust, accurate surveys, the applications are virtually unlimited so long as you have data available to feed into the software.

Making Better Service Possible

Analysis depends on having good, reliable data upon which to base your investigations. Data can be gleaned from patients in a manner that also makes them believe they are being offered a better service. HMOs often ask patients to fill out short, easy forms. Many people are happy to do this if they realize it's for their own benefit, so they answer questions such as how long they had to wait to see a doctor, how effective their treatment was, and how happy they were with the service provided.

Analyzing this information will help set reasonable expectations about waiting times and reveal possible ways to improve service. It will also provide substantial knowledge about what patients want from their healthcare providers. This, in turn, will let health professionals tailor their services to patient needs and take a more proactive role--for example, approaching patients who have expressed an interest in new treatments whenever they become available.

Westport, CT-based Rx Remedy, a marketing, research and communications firm, and publisher of Remedy magazine, looks at healthcare from the patient's perspective. Through sophisticated market research, the company tracks utilization of healthcare services and drugs, matching these behaviors to detailed consumer group profiles.