Haoyu Yang

I am a Postdoctoral Research Fellow in the Department of Biostatistics at the Harvard T.H. Chan School of Public Health, working with Professor Xihong Lin.

My work starts from one observation: every dataset contains more information than a standard analysis extracts.

Some of it sits in missing measurements — the 90% of a biobank whose proteome was never assayed, though its genotypes and clinical records are complete. Some sits in heterogenous populations - the target population is too small, and the available source populations are too heterogeneous to combine. Some is discarded by the analysis itself, when half a million hypothesis tests are evaluated one at a time, and everything the ensemble knows is thrown away. And some information is ignored when participants are enrolled, by a design that leaves balance to chance.

I build methods that convert this potential information into scientific discovery. The difficulty is that each of these sources is untrustworthy in its own specific way — a generative model can be misspecified, another population can carry different biology, a composite null distribution is hard to characterize — which is why the interesting part of the problem is statistical rather than computational.

Before Harvard, I earned my Ph.D. in Statistics at Renmin University of China.