Use of Machine Learning to Determine the Information Value of a BMI Screening Program

Abstract

Introduction: Childhood obesity continues to be a significant public health issue in the U.S. and is associated with short-and long-term adverse health outcomes. A number of states have implemented school-based BMI screening programs. However, these programs have been criticized for not being effective in improving students' BMI or reducing childhood obesity. One potential benefit, however, of screening programs is the identification of younger children at risk of obesity as they age.Methods: This study used a unique panel data set from the BMI screening program for public school children in the state of Arkansas collected from 2003 to 2004 through the 2018−2019 academic years and analyzed in 2020. Machine learning algorithms were applied to understand the informational value of BMI screening. Specifically, this study evaluated the importance of BMI information during kindergarten to the accurate prediction of childhood obesity by the 4th grade.Results: Kindergarten BMI z-score is the most important predictor of obesity by the 4th grade and is much more important to prediction than sociodemographic and socioeconomic variables that would otherwise be available to policymakers in the absence of the screening program. Including the kindergarten BMI z-score of students in the model meaningfully increases the accuracy of the prediction.Conclusions: Data from the Arkansas BMI screening program greatly improve the ability to identify children at greatest risk of future obesity to the extent that better prediction can be translated into more effective policy and better health outcomes. This is a heretofore unexamined benefit of school-based BMI screening.

Predicting Childhood Obesity Based on Single and Multiple Well-Child Visit Data Using Machine Learning Classifiers

“…Marcos-Pasero used random forest (RF) and gradient boosting to predict the BMI from 190 multidomain variables (data collected from 221 children aged 6 to 9 years) and determined the relative importance of the predictors [ 25 ]. Zare et al used kindergarten-level BMI information, demographic, socioeconomic information such as family income, poverty level, race, ethnic compositing, housing, parent education, and family structure to predict obesity at the fourth grade and achieved an accuracy of about 87% by using logistic regression and an artificial neural network [ 26 ]. This study also reported that the kindergarten BMI Z-score is more important for BMI prediction than the demographic and socioeconomic variables.…”

“…We choose random forest, artificial neural networks, logistic regression, support vector machine, k -nearest neighbors, and k -means clustering because they are the most commonly used ML algorithms in the decision tree, neural network, probability of occurrence, hyperplane, neighborhood, and unsupervised learning based categories, respectively. These algorithms have been widely used in previous studies for predicting obesity [ 20 , 22 , 23 , 25 , 26 , 31 ]. Although these six ML algorithms have been used in multiple studies, the implementation approaches differ from one study to another.…”

Applications of Artificial Intelligence to Obesity Research: Scoping Review of Methodologies

“…Therefore, of the 3090 articles identified initially through the keyword search, 46 (1.49%) were included in the review . [28,32,33,37,42,46,48,[50][51][52][53]57,58,60,62,63]; 6 (13%) in China [39,40,45,56,64,65]; 3 (7%) each in the United Kingdom [27,68,69] and Korea [35,43,49]; 2 (4%) each in Italy [36,71], Turkey [41,70], Finland [44,59], Germany [54,55], and India [36,71]; and 1 (2%) each in Saudi Arabia [26], Iran [67], Serbia [66], Portugal [61], Spain [47], Singapore [38], Australia [34], and Indonesia [29]. Of the 46 studies, 32 (70%) adopted a cross-sectional study design [26,…”

“…Of the 46 studies, 23 (50%) focused on adults, 14 (30%) on children and adolescents, 1 (2%) on people of all ages, and the remaining 8 (17%) did not report the age range of participants. [26][27][28][29][33][34][35][36][37]39,41,[44][45][46][47][49][50][51][53][54][55][56][58][59][60][61][62][63][64][65][66][67][68]70], 8 (17%) analyzed digital image data [30,31,38,40,42,43,57,71], and 4 (9%) analyzed text data [32,48,52,69]. Obesity-related measures used across the studies included anthropometrics (eg, body weight, BMI, BFP, WC, and WHR) and biomarkers.…”

“…First, the studies found that ML or DL models were generally effective in detecting clinically meaningful patterns of obesity or relationships between covariates and weight outcomes; for example, ML and DL models were found useful in classifying obesity severity [30,47,52], identifying anthropometric [34] and genetic characteristics of obesity [56], and predicting obesity onset in children [28,53,63]. ML algorithms (eg, random forest [RF] and conditional RF) revealed meaningful relationships between school and neighborhood environments and overweight and obesity [45,58,62].…”