Estimates Of Body Composition In Normal Weight And Overweight Adults Using The Leanscreen App

Abstract

The LeanScreenTM app (LS) uses photographs taken from the front and side views, touch screen technology and manufacturer algorithms to estimate circumferences of the neck, abdomen, waist, and hips. The LS app then estimates body composition using the Department of Defense (DOD) circumferences regression equation. PURPOSE: This study evaluated the validity and reliability of estimates of percent body fat (%BF) in adult men and women using the LS app compared to DEXA as the criterion method. METHODS: Height and body mass were measured on 63 males and 58 females between 18 and 50 years of age. The participants were categorized as normal weight (n = 61) or overweight (n = 60) based on BMI. A DEXA scan was performed on each subject once. To assess the within-day and between-day reliability of the LS app, an administrator took front- and side-view photographs of each subject twice on the same day and once on a second day. The LS app was used to identify landmarks of the neck, abdomen, waist, and hips on each set of photographs of each subject. The LS app provided an estimate of %BF. Estimates of %BF from the LS app were compared to %BF values obtained from DEXA. RESULTS: A mixed model ANOVA indicated that the LS app significantly (p = 0.0001) underestimated DEXA %BF by an average of 3.6 %BF. The difference in %BF values between the LS app and DEXA were not influenced by gender or BMI category. Regressing the %BF determined from the LS app against DEXA resulted in an R2 = 0.83 and a SEE = 3.4 %BF. The intercept (2.84 %BF) and slope (1.035) of the regression line confirmed the significant bias. Bland-Altman analysis suggest that the bias of the LS app estimates of %BF increased with increasing body fatness. Use of the LS app to estimate %BF resulted in high intrarater (0.99) reliability within and between days. CONCLUSIONS: The results of this study show that the LeanScreenTM app, although highly reliable, significantly underestimated DEXA %BF. We suspect that this is related more to the DOD regression equation rather than the use of photographs and touchscreen technology.