Simulating long-term human weight-loss dynamics in response to calorie restriction.

Area of Science:

• Obesity research
• Mathematical modeling
• Nutritional science

Background:

• Mathematical models for predicting body weight (BW) and composition changes from lifestyle interventions require long-term validation.
• Existing models have not been sufficiently tested over extended periods.

Purpose of the Study:

• To compare the accuracy of mathematical models underlying two popular weight-loss prediction tools: the NIH Body Weight Planner (NIH BWP) and the PBRC Weight Loss Predictor (PBRC WLP).
• To validate these models against data from the 2-year CALERIE study.

Main Methods:

• Mathematical models were initialized with baseline data from the CALERIE study.
• Simulations of changes in body weight (ΔBW), fat mass (ΔFM), and energy expenditure (ΔEE) were performed using time-varying energy intake (ΔEI) data.
• No model parameters were adjusted from their original published values.

Main Results:

• The PBRC WLP model significantly underestimated mean BW loss by 3.8 kg, primarily due to an exaggerated early decrease in energy expenditure (EE) with calorie restriction.
• The NIH BWP model showed a smaller mean ΔBW bias of -0.47 kg.
• Both models demonstrated substantial variability in predicting individual weight changes; measurement uncertainties in ΔEI were a major factor for the NIH BWP model.

Conclusions:

• The NIH BWP model demonstrated superior performance, accurately simulating average weight loss and energy balance dynamics during long-term calorie restriction.
• While accurate on average, the NIH BWP model's predictions for individuals exhibited significant variability, warranting cautious interpretation.
• The study highlights the need for robust validation of weight management models over extended durations.