Harvard Catalyst Profiles

Diversity is one of the most basic properties of large, complex systems, and therefore one of the first things investigators measure, but is often mismeasured. Examples include clonality, clonotypes per kiloread, and various other ad hoc measures, joining richness, Shannon entropy, the Gini index, and other more familiar measures. There is a framework for measuring diversity correctly, where "correctly" means "having reasonable properties" (e.g. the diversity of the whole being at least as large as the diversity of the parts). Anything else risks erroneous conclusions, which (in the form of misleading biomarkers) can neagtively impact clinical care. But how common are such errors? This short project will chart this landscape quantitatively.

Laboratory testing is the single highest-volume medical activity and thus a major determinant of healthcare quality. However, the appropriateness of laboratory utilization across medicine remains understudied. A landmark 2013 study, now cited over 400 times---first author, an HMS IV---showed high inappropriate overutilization and, perhaps more surprisingly, even higher inappropriate underutilization. The question is, what has changed in the near-decade since. We seek to conduct a systematic review and meta-analysis to followup the previous study. The student will lead. If the student/s has/have computational skills and an interest in informatics, this may be supplemented with a review of our own hospitals millions of laboratory results.