My philosophy is that Biostatistical Methods need to be accompanied by software, easy to explain to collaborators, and scalable. I have developed multiple methods, including testing for variance components, multilevel and longitudinal PCA, functional and image regression, population value decomposition, movelets, dynamic prediction, and the upstrap. A large component of my research has been dedicated to analyzing and quantifying high resolution signals (from 1 per minute to hundreds per second) and images (structural MRI or CT images). In many cases standard Biostatistical techniques do not scale up well and need to be either modified or invented to address the sheer size and complexity of the new measurements. Following the Hopkins Biostatistics philosophy, I do not follow trends in the Statistical literature, as I find it a lot more rewarding to solve problems and develop methods, when and if necessary. While on this website I separate methods development and data analysis, in fact, the best way is to not separate them and simply do them at the same time. This saves time, energy, and focuses the mind on what is really important.