Predictive Modeling of Weight Loss Curves

• Activity Number: 172
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Shailaja Suryawanshi* and Kaifeng Lu+
• Companies: Merck & Co., Inc. and Merck & Co., Inc.
• Keywords: Non-linear models ; Polynomial model ; Spline-fitting

Abstract

Obesity is a significant worldwide health problem associated with increased risk of cardio-metabolic problems. Lifestyle and pharmacological interventions often have limited success in preventing the regain after these interventions or after weight loss nadir is reached. Patients typically regain up to 40% of their lost weight in the year following treatment. In this presentation, change or percent change in body weight from baseline over time, controlled for factors such as baseline body weight, will be modeled using non-linear models. Specifically polynomial or spline models will be assessed, recognizing that these could be specific to the time course of treatment, for example a flatter curve after weight loss nadir is reached. The parametric models will be built from a set of studies and evaluated using an independent set of studies.