Application of Bayesian Sequential Analyses to Single Case Data: A Proof-of-Principle Study

Timely identification of whether a treatment is having the expected effect on symptoms could help improve outcomes and inform treatment personalization. Application of Bayesian sequential analysis to single case data may offer a useful method for rapidly identifying treatment success or failure at an individual level, tailored to an individual patient. This paper provides proof-of-principle for this concept, applied to simple internet-delivered interventions in an ABC randomized single case design. Seven participants with elevated symptoms of depression seeking treatment at an outpatient psychotherapy centre completed a two-week baseline period. They then completed two weeks of first one and then another internet-delivered cognitive training intervention, with the order of intervention randomized between participants. Participants rated depression symptoms on a daily basis throughout the study. Each participant’s baseline data was used to simulate hypothetical continuations after treatment initiation, and this in turn was used to determine individualised Bayes factor thresholds and other decision parameters for determining treatment success or failure. The study illustrates the potential application of sequential Bayesian analyses at a single case level to inform clinical decision-making, as well as highlighting potential challenges in implementing this approach.