Analysis of personalized treatment options from a large clinical trial

The goal of personalized medicine is to identify treatment with the highest expected success given individual factors. For many diseases that involve individual behavior, such as substance use, finding a small set of strong predictors, might not be feasible due to a large number of factors influencing the outcome. Recently we have developed methodology that considers an ensemble of models and allow one to explore “what if” scenarios. Thus, for an individual with a specific combination of factors our approach provides smooth and stable prediction of potential treatment outcomes under several alternative treatments. We have applied our and other competing methods to identify best personalized treatments of alcoholism to a dataset from the largest clinical trial of several alcohol treatment approaches called COMBINE. We have identified best individual treatments and assessed the effectiveness of the personalized treatment when applied to each patient. We have also assessed the consequences of applying the least effective treatment. We show that for some patients the potential difference in the outcomes between the most effective and the least effective treatment might not have a significant difference, while others could substantially benefit from personalized best choice of treatment.