Page History
...
The script uses a set of 10 heuristic filters that attempt to eliminate patients who are sick or abnormal or who might no longer be receiving care at your hospital. It returns only the aggregate counts of the number of patients who pass each heuristic filter and gives the demographic breakdowns of the patients identified as healthy and normal. When run at Partners Healthcare in Boston, only about 1.5% of approximately 2 million patients were identified as healthy and normal.
Is the number of healthy normal patients at Partners Healthcare normal, or does this vary by the type of hospital and patient population? I need YOUR help to answer this question. Below is a draft manuscript describing the filters and the results at Partners Healthcare. Run the script in your i2b2 CRC database and share your results, and you will be added as a co-author. (Make sure you have Institutional Review Board approval to run this query.)
The first 6 filters are part of a separate project to estimate the fraction of patients who have at least one chronic disease diagnosis code. Note that in identifying healthy patients, the patients with chronic diseases are first eliminated.
This script should be run in a standard implementation of the i2b2 CRC cell (version 1.3 or higher) in Microsoft SQL Server 2005 or later. It may require about 10 GB of storage and 1 hour processing per one million patients, though this can vary greatly depending on the particular data in the i2b2 instance and the database hardware.
...
