Radebe L, van der Kaay DCM, Wasserman JD, et al. - The most common endocrine malignancy is papillary thyroid carcinoma. Because of the benign characteristic of most nodules, it is challenging for the clinician to recognize those most likely to harbor malignancy while limiting exposure to surgical risks among those with benign nodules. Researchers herein used random Forests (augmented to select features based on our clinical measure of interest), along with interpretable rule sets, on demographic, ultrasound and biopsy data of thyroid nodules from children <18 years at a tertiary pediatric hospital. Non-benign cytology and malignant histology were better predicted using the models than historical outcomes. They identified this study to be the first attempt at generating an interpretable machine learning based clinical tool to aid clinicians. While a model adequate to replace existing practice can not be generated in the current retrospective study, largely due to the limitation in cohort size with sufficient data points, the better accuracy and AUROC for recognizing biopsy and surgical candidates is encouraging. This serves as a proof-of-principle that systematic data ascertainment and mathematical modeling may eventually allow the generation of a robust tool to help guide clinical decision making and to avert unnecessary interventions.