POWER POINT

Knowledge Discovery in Large Data Sets (KDD)

Importance of Data What is data mining? Imagine all of the data that one hospital generates in a given day from: Patient care Diagnostic coding Inventory Work flow patterns Wait times Human resources Any other opportunity for data collection

Importance of Data Mining All of this data is stored someplace whether on internal servers, external data 'warehouses', or in data 'cemeteries' Data mining is the technique used to look back on all the data on a specific subject to identify trends, patterns, outliers, etc. Provides valuable information for current issue

Importance of Data Mining Knowledge discovery in datasets (KDD) is capable of finding relationships between elements not previously considered Example: running an open query on all admissions to a unit within the past 5 years. KDD portends to identify relationships, bringing previously unknown correlations to light. Looks for the 'needle in the haystack'ExampleExample Credit card companies keep detailed records of their customers, including spending habits. Millions of customers with billions of transactions KDD monitors spending activities and red flags any unusual expenditures

DISCUSSION QUESTION:

One little known and newly emerging technique to generate new knowledge is data mining in large data sets. The PP lecture describes the technique and how it can be used. Think about the clinical setting, whether it is where you currently work or have had a clinical experience for school.

Describe how you would set up a data mining topic. Identify two factors that you would like to see if a correlation exists (read the PP lecture for an example). Remember, the two factors would have to be capturable data meaning that the data has to be available in the electronic charting system or other computer database. Provide a rationale as to why you selected the two variables and what you expect the results to look like.

Finally, explain how the results would be used to improve performance or quality on the unit.