Chen R, et al. - Researchers examined how data volume and diversity and training conditions influence recurrent neural network methods compared with traditional machine learning methods. The relative performance of machine learning models trained to detect a future diagnosis of heart failure in primary care patients was assessed. Fron longitudinal electronic health record data, they obtained data on 4,370 incident heart failure cases and 30,132 group-matched controls. Analysis revealed superiority of recurrent neural network model performance under a variety of conditions that included (1) when data were less diverse (eg, a single data domain like medication or vital signs) given the same training size; (2) as data quantity extended; (3) as density raised; (4) as the observation window length extended; and (5) as the prediction window length reduced. This supports the efficacy of recurrent neural networks in predicting a future diagnosis of heart failure given sufficient training set size. Continued improvement in model performance was observed in direct relation to training set size.