Situation
As one of the largest manufacturers of prescription drugs in the world, Dutch-based Merck Sharp & Dohme (MSD) relies heavily on input from doctors to help the patients it serves. The end result is better care for patients who are afflicted with illnesses that range from AIDS to osteoporosis to heart failure to migraine headaches to asthma.
MSD uses data mining and text mining technology to help the organization analyze important information it collects, and then create effective programs that best address physician and patient needs.
"Years ago, a consultant showed MSD what could be achieved with data mining. The results spoke for themselves and my job was created," said Werner Kreiter, data mining specialist at MSD.

Challenge
Like any other profession, the medical industry is comprised of a vast array of beliefs and opinions. That's where the challenge comes in for MSD and Kreiter: they must get a firm grasp on what the doctors are saying out in the field and then pass those analyses on to the team so it can create effective marketing campaigns for the drugs MSD manufactures. Not an easy task, considering the wide-ranging target audience.

On one end of the spectrum are the "pioneer" doctors, the physicians who are very open to new insights and research results, and fairly quickly turn scientific findings into practice. At the other end of the scale are the "conservative personality" doctors, the physicians who want to do everything by the book, spend a lot of time researching treatment options, and base their opinions on the thorough study of specialist articles or exchanges with colleagues.

"To be successful, we have to find the right way of approaching all these types of doctors. But before we even consider this, we have to first identify these groups," said Kreiter, who utilizes information from various sources to reach his goal, including internal data and data from external providers.

Solution
Kreiter utilizes text mining and quantitative analyses to get a better understanding of the data collected from surveys conducted at various communications seminars, and then provides that valuable information to MSD's marketing team. Some of the areas measured, for example, include the number of years a doctor has been established, the number of patients a doctor serves, and additional qualifications. The functionality allows MSD to gain further insight into the data with regard to their significance and correlation with other characteristics.

MSD also uses software for profiling. The software allocates doctors to the typologies described above. To track marketing actions, Kreiter uses software to segment doctors based on measures introduced by the marketing department. This, in turn, provides insight into which action catalogue best characterizes the relevant target groups.
The text mining functionality is based on the natural grammatical analysis of text, which isn't dependent solely on keyword search, but analyzes the syntax of language and "understands" the content.

To this end, MSD works with the Gesellschaft für Konsumforschung panel (Association for Consumer Research, GfK), which uses the daily "diary" entries of doctors on the panel to learn which pharmaceutical representatives have visited them, what product communications were conveyed, and whether in the future they will include these products in the range of drugs they prescribe. Text mining analyses of the product conversations noted by the doctors reveals speech patterns that accompany various prescribing behaviors. This enables pharmaceutical manufacturers to optimize the communications skills of their field workers.
"We know, thanks to text mining, which properties of and information about our drugs are particularly well understood in conversations with the doctors, and in which instances the terms used for our marketing campaigns still need to be refined," explained Kreiter.