Positive responses among subjects treated with an active drug may be due to the chemical component of the drug, a placebo effect, spontaneous remission or a combination of these factors. Outcome profiles over time for depressed subjects treated with antidepressants and placebos are estimated and classified. Clustering techniques are used to determine representative profiles for various outcome profiles (placebo responder, true-drug responder, etc). The profiles for placebo treated subjects are used to help validate the classification process. In addition, the methodology will be evaluated using data from discontinuation trials where all subjects received the active drug in the first phase and in the second phase, subjects either continue on the drug or they are switched to a placebo.
Available at: http://works.bepress.com/thaddeus_tarpey/24/