Todd MacKenzie, PhD
Professor of Biomedical Data Science
Professor of Medicine
Professor of The Dartmouth Institute
Biostatistician
Todd has used statistics to help clinical and health services researchers from a vast spectrum of disciplines and specialties, resulting in 170 co-authored publications. The majority of these projects involve comparative effectiveness, association studies or prediction modeling. The area of statistics he is most interested in, is survival, aka time-to-event, analysis. Todd has developed statistical methods for estimating functional forms of the hazard ratio, estimating survival in the presence of left truncation and recently, estimating the hazard ratio using instrumental variables ("Using Instrumental Variables to Estimate a Cox's Proportional Hazards Regression Subject to Additive Confounding", MacKenzie et al, 2014 in Health Services and Outcomes Research and Methodology). The latter is a method of deriving unbiased estimates even in the setting of unmeasured confounders which are a threat to any observational study. I am currently interested in deriving an estimate of causal hazard ratio (e.g. compliance adjusted hazard ratio), using the platform of principal stratification for randomized clinical trials.
Professor of Biomedical Data Science
Professor of Medicine
Professor of The Dartmouth Institute
Biostatistician
Todd has used statistics to help clinical and health services researchers from a vast spectrum of disciplines and specialties, resulting in 170 co-authored publications. The majority of these projects involve comparative effectiveness, association studies or prediction modeling. The area of statistics he is most interested in, is survival, aka time-to-event, analysis. Todd has developed statistical methods for estimating functional forms of the hazard ratio, estimating survival in the presence of left truncation and recently, estimating the hazard ratio using instrumental variables ("Using Instrumental Variables to Estimate a Cox's Proportional Hazards Regression Subject to Additive Confounding", MacKenzie et al, 2014 in Health Services and Outcomes Research and Methodology). The latter is a method of deriving unbiased estimates even in the setting of unmeasured confounders which are a threat to any observational study. I am currently interested in deriving an estimate of causal hazard ratio (e.g. compliance adjusted hazard ratio), using the platform of principal stratification for randomized clinical trials.