Using Data Mining techniques to establish and predict trends is a black art with as many myths and legends as any book by Terry Pratchett. Like any magic it’s only credible if it works and it clearly does, because certain communities are becoming increasingly worried about the unconstrained use of data mining in the marketing world. A good example is in the pharmaceutical industry where there is now legislation in New Hampshire to ban the sale or use of of prescriber or patient data for use in marketing, because of the success of these tools in targeting doctors.
Of course like any magic or piece of technology it is the use these techniques the decides whether they are malign or benign. In healthcare these techniques can be used to spot trends and relationships in the causes of disease. This could be used in several ways , for example:
- To inform us of the consequences of lifestyle choices such as over eating smoking etc. and few of us would have any problems with that research e.g. the recent work on the link between obesity and cancer
- If that data set, by some ‘accident’, was made available to health insurance companies, and was used to bombard the healthy amongst us with offers of lower health insurance, or was used as a basis to set premiums on life assurance policies, we would be much more concerned.
So how does all this work in SQL Server and the answer is …
check my next post.