Premium
shapecoder : a new method for visual quantification of body mass index in young children
Author(s) -
Park B.K. Daniel,
Reed M. P.,
Kaciroti N.,
Love M.,
Miller A. L.,
Appugliese D. P.,
Lumeng J. C.
Publication year - 2018
Publication title -
pediatric obesity
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.226
H-Index - 69
eISSN - 2047-6310
pISSN - 2047-6302
DOI - 10.1111/ijpo.12202
Subject(s) - overweight , body mass index , intraclass correlation , medicine , obesity , correlation , predictive value , reliability (semiconductor) , correlation coefficient , physical therapy , statistics , psychometrics , clinical psychology , mathematics , power (physics) , physics , geometry , quantum mechanics
Summary Background Few tools exist to quantify body mass index visually. Objective To examine the inter‐rater reliability and validity (sensitivity and specificity for overweight/obesity and obesity) of a three‐dimensional visual rating system to quantify body mass index (BMI) in young children. Methods Children ( n = 242, mean age 5.9 years, 50.0% male; 40.5% overweight/ obese) participated in a videotaped protocol and weight and height were measured. Research staff applied a novel three‐dimensional computer‐based figure rating system ( shapecoder ) to the child's videotaped image. Inter‐rater reliability was calculated, as well as correlation with measured body mass index (BMI) and sensitivity, specificity, positive predictive value and negative predictive value for overweight/obesity and obesity. Results Inter‐rater reliability was excellent (intraclass correlation coefficient = 0.98). The correlation of shapecoder ‐generated BMI with measured BMI was 0.89. For overweight/obesity, the sensitivity, specificity, positive predictive value and negative predictive value were 62%, 97%, 94% and 79% respectively. For obesity, these values were 65%, 99%, 97% and 92% respectively. Conclusion shapecoder provides a method to quantify child BMI from video images with high inter‐rater reliability, fair sensitivity and good specificity for overweight/obesity and obesity. The approach offers an improvement over existing two‐dimensional rating scales for BMI.