Analyze a rise in patients over time


  • I work for a hospital and am trying to figure out why there have been an increase in the number of patients admitted to the hospital over the past three months. From March 2014 to May 2014 there has been an increase by about 15 percent in patients. I transformed my data so that I have 92 records (every day from March 1 - May 31) and have a date field called PatDate. I have another field called CensusDays which is the total number of patients in the hospital for each of the 92 days. I then have another 100 fields that I have converted into integers. As an example, we may have five different types of Payers such as Medicaid, Medicare, Self-Pay, Managed Care and Commercial. I created separate fields for each (i.e. PayerMedicaid, PayerMedicare) and have a value for each. So on March 1, the CensusDays may be 300, the PayerMedicaid may be 50 (50 Medicaid patients that day), PayerMedicare and so on. I then have field for all of the zip codes in our city such as Zip30050, Zip30051 which are integers. So every field is an integer except my date field.

    I want to do some data mining on this data to see why there has been a rise in patients but am unsure how to begin. I have the data mining add in installed and have connected to analysis services. I did a correlation matrix and have about 25 fields where there correlation between 25 fields and CensusDays is highest but this isn't take into account time. Ultimately I want to be able to say the reason why we have had a rise in patients is because we have more Medicaid patients from zip code xxxx who are a diagnosis of yyyy or some sort of combination like that. can someone help?

    Monday, July 07, 2014 12:36 PM


  • You might be able to use Analyze Key Influencers tool from the data mining add-ins for this.

    Pick date as the column you want to analyze.

    Report will show you how values of other columns predict date (e.g. if zipcode=12323 is strong influencer for date that is at the end of your period then it means that you have a lot of patients from that zipcode lately).

    Tatyana Yakushev []

    • Marked as answer by ChuWil2 Monday, July 14, 2014 6:13 PM
    Thursday, July 10, 2014 9:21 PM