The future of personalized therapeutics: tailoring interventions to genomic and patient data

With significant advancements in genomic and health data technologies, it has become increasingly clear that there is a large degree of variability between individuals’ responses to conventional therapies, though these are typically delivered identically to populations as a whole. This variability may include differences in the efficacy of the therapeutic, any adverse effects experienced, and even changes in codes for mechanistic pathways implicated by the therapeutic, such as changes in transcriptomic and proteomic profiles. In light of this observed heterogeneity, although patients have traditionally been assigned into broad sub-populations using clinical characteristics and labeled as in need of particular therapies, honing in on specific attributes may allow for a more tailored therapeutic intervention. Such personalization has the potential to both improve patient outcomes by maximizing therapeutic effect while minimizing side effects, as well as reduce costs by limiting unsuccessful interventions and quickly reaching positive health outcomes. Design of personalized therapeutics offers an additional benefit over conventional therapeutic design approaches; if patients are extensively profiled for genomic and clinical characteristics and known goal states alongside available therapeutic options from which to choose, the mapping between profile and best intervention may be directionally learned using machine learning and causal inference techniques.