DEVELOPING BODY SURFACE AREA MODELS FOR HUMAN VARIABILITY

Open Access
Author:
Sen, Shweta
Area of Honors:
Mechanical Engineering
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
  • Matthew Parkinson, Thesis Supervisor
  • Sean Brennan, Honors Advisor
Keywords:
  • Design for Human Variability
  • Body Surface Area
  • Human Modeling
  • Population Modeling
Abstract:
The objective of this work is to develop a model that can accurately predict body surface area (BSA) across a wide range of anthropometry, particularly at height-weight extremes. Body surface area is a biometric used to study a variety of physiological measures, such as metabolic rate, cardiac output, oxygen consumption, and drug dosage in chemotherapy. Current mathematical models used to predict BSA are limiting because they do not capture the effects of human variability as a result of having small, homogeneous sample populations. While the Dubois-Dubois model (developed in 1916) is most commonly used in clinical practice, it is anthropometrically outdated and often underestimates BSA for obese (BMI = 30-35) and extremely obese (BMI > 35) individuals. Given that more than 30% of the U.S. adult population has a BMI greater than 30, the Dubois-Dubois model does not accurately estimate BSA for at least one third of the target population. To improve the accuracy and reliability of BSA prediction, this work pools data from two national anthropometric databases (NHANES and ANSUR) to synthesize a population of 2,000 men and 2,000 women that mimics the anthropometric variation observed in the U.S. adult population. Using principal component analysis and template meshing, 3D tessellated body models are generated and used to compute BSA. Regression analysis is applied to develop two gender-specific models that accurately predict BSA as a function of stature and weight for the U.S. adult population: BSA_Male(m^2) = 0.01415296 + 0.011098 × Mass (kg) + 0.0050889 × Stature (cm) BSA_Female(m^2) = 0.015968879 + 0.0125053 × Mass (kg) + 0.0047726 × Stature (cm) Results indicate that the model presented in this work accommodates a larger proportion of the U.S. adult population in comparison to the Dubois-Dubois (clinical standard) model. The model is shown to predict BSA (within ±3% error) for adults with a BMI between 20 and 45. Application of this work in the clinical setting can improve the fidelity of BSA-based drug dosing regimens commonly used in chemotherapy and burn treatment.