First of all you have to pilot it in some unit or clinics for test run, then address identified issues such as technical mistakes, formulas, validation rules, etc. redeploy again for another test and acceptance before implementation.  

Hi Farideh,

I believe you are asking about the deployment methodology. If so, please see the following two links:

http://stackoverflow.com/questions/22309236/options-for-deploying-r-models-in-production

http://www.revolutionanalytics.com/sites/default/files/zementis-streamlining-analytics-deployment-whitepaper.pdf?mkt_tok=3RkMMJWWfF9wsRonuK/AZKXonjHpfsX66uwrT/rn28M3109ad%2BrmPBy63IAXD4cwVfubBwsIToJk1gNUD/Scc45D9fhNHlW7Vjw%3D

please read "Abdulfattah Suliman Mashat and Mohammed Abdel Razek, “Dominant Student Modeling for Web Personalization” 2011 International Conference on Internet Computing, July 18-21, 2011, Las Vegas, USA, 2011."