This work is a collection of three manuscripts that use novel methodological approaches in causal inference and dimensionality reduction on different classes of Alzheimer’s disease data. The overarching aim of this dissertation is to investigate complementary perspectives that may improve the health of the population affected by Alzheimer’s disease and related demen.as, which is deemed to grow in number and public health relevance globally in the near future. The first manuscript investigates the opportunity for drug repurposing of acetylcholinesterase inhibitors, a medication normally prescribed to Alzheimer’s patients, in the treatment of certain cardiovascular disease. The second manuscript evaluates the effect of dual x-ray absorptiometry bone density scans on the likelihood of subsequent hip fracture in the Alzheimer’s disease population, where osteoporosis is a highly incident comorbidity. The third manuscript presents an algorithm for dimensionality reduction and an application of set theory on pairwise classification problems to identify significant predictors of Alzheimer’s disease progression. The novelty of this work lies in the use of random treatment date generation, in combination with random sampling with replacement, to estimate the average treatment effect on the treated in the first two manuscripts. In the third manuscript, a novel algorithm is presented, which improves performance over the sparse-group lasso by adding a forward selection step on an external validation set of features. Taken together, this work aims to contribute to the methodological advancement of statistical approaches for coefficient estimation in the context of causal inference, as well as to the empirical identification of elements that can be translated into actionable policies, on one hand, or utilized in clinical settings as part of diagnostic biomarkers, on the other.