DOES PERSONALIZED TREATMENT BENEFIT EVERYONE? PREDICTIVE ANALYSIS OF OPTIONS FROM CLINICAL TRIALS

Friday, January 8, 2016
Foyer, G/F (Jockey Club School of Public Health and Primary Care Building at Prince of Wales Hospital)

Georgiy Bobashev, Ph.D., RTI International, Center fot Data Science, Durham, NC and Barry Eggleston, MS, RTI International, Durham, NC
Purpose: To identify who will benefit from potentially more expensive personalized approach to treatment vs. random treatment or no treatment

Method(s): A few years ago we have developed and validated methodology (mobForest) to predict patient’s outcome if the patient is assigned to an alternative treatment. 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 (including placebo). We tested the statistical significance of the difference between the best and the worst. We have also tested the difference in outcomes between the best and the second best treatment as part of the sensitivity analysis. Our predictive methodology is based on model ensemble which is difficult to interpret in clinical settings. However we have identified a set of manageable and interpretable influential predictors.

Result(s): We have shown that for some patients the potential difference in the outcomes between the most effective and the least effective treatment might not be significant, while others could substantially benefit from personalized best choice of treatment. For the patients in alcohol treatment study the proportion of people significantly benefiting from personalized treatment is <30%. We also illustrate the results through innovative graphics.

Conclusion(s): Our methodology allows one to forecast patient’s outcomes from alternative treatments and evaluate the proportion of individuals who would benefit from personalized treatment and identify characteristics of such population. This approach adds value to the evaluation of personalized treatment options, especially when some treatments are more expensive than the others. This methodology has been applied to a new study of heroin teratment with naltrexone in Russia where substitutional therapy is illegal.