For a study, researchers sought to find a way to identify children’s height outliers in electronic health records (EHR) of children. So they made two groups of children, 2 to 8 years old, to train and test a model that could predict heights from things like age, gender, race, and weight. The first group had 1,376 children, and the second group had 318 patients. Investigators looked at each child’s height and determined if they were an outlier or not. In the training and testing cohorts, the model’s R2 explained 82.2% and 75.3 % of the range in height values. As measured by the area under the receiver operating characteristic curve, the discriminatory ability to identify height outliers in the testing group was very good, 0.841. The outlier sensitivity is 0.713, the specificity is 0.793, the positive predictive value is 0.615, and the negative predictive value is 0.856 based on a somewhat aggressive threshold of 0.075. Investigators have developed a new method for detecting the outliers in height measurements from pediatric EHRs. This largely automated method can be applied to ensure the accuracy of height measurements and indices of body proportionality, such as body mass index.