Suicide is the 10th leading cause of death in the United States, and its incidence is rising, placing unprecedented emotional and financial burdens on its victims' families (Brathwaite et al., 2016). In the afore-cited study, by Brigham Young University, researchers recruited 135 participants through Amazon's Mechanical Turk (www.mturk.com) and asked them to complete three clinically validated questionnaires for assessing suicide risk. In addition, the researchers analyzed the tweets of each participant using a language analysis tool that extracts features from the text. The features largely measure the frequency of certain word categories including "family," "anger," and "sadness." They used Python and the scikit-learn library to then build models that classified each of these "training set" participants as suicidal or not suicidal. To test their model on previously unseen data, they used leave-one-out cross-validation, a procedure in which only one participant was used as test data, and this was repeated 135 times. Their decision tree model obtained an accuracy of 91.9% and had a sensitivity of 53% and a specificity of 97%. Certainly, this is encouraging for identifying suicidal patients before they commit suicidal acts, although it remains to be seen how privacy would be preserved in such a suicide-monitoring system.