Meta-Analytic Decision Tree Learning🔗

Article 4 is entitled “Predicting Heterogeneous Treatment Effects of an Internet based Depression Intervention for Patients with Chronic Back Pain: Secondary Analysis of Two Randomized Controlled Trials”. In this study, we employed multilevel model based recursive partitioning (“decision tree learning”) across two randomized trials. This algorithm was used to derive an easily interpretable decision tree to predict the benefits of a digital depression intervention in chronic pain patients. The performance of the developed tree was externally validated in a third trial examining a similar digital intervention. Results of the tree architecture indicate that the digital intervention may be particularly helpful in patients with moderate depressive symptoms (PHQ 9 scores of 10 15) who experience low pain related self efficacy. Conversely, patients with high pain related self efficacy in this substratum benefit the least.