TITLE: Principles of Design for Computer Experiments and Non-Communicable Disease in Pakistan

SPEAKER: Dr. Ben Haaland

ABSTRACT:

This talk will will consist of two parts as follows.

Simulations are used to study real systems for which real experimentation is not feasible. In the usual situation where runs of the simulation take a non-negligible amount of time and
computing, experimental design is extremely important in gaining high quality information from simulations. Principles of experimental design will be proposed and justified which ensure
high numeric accuracy, high nominal accuracy, and accurate parameter estimates for deterministic simulations. Specific situations addressed include non-stationarity, effect sparsity, non-trivial functional ANOVAs, and simplified functional representations, as well as the worst case scenario where the unknown function or mean function is highly non-linear with no low-dimensional representation.

Non-communicable disease (NCD), such as cardiovascular disease, cancer, and respiratory disease, have become the major cause of mortality in Pakistan. Estimates for 2010 and projections for 2025 of cause-specific mortality are presented along with estimates and
projections of deaths attributable to several important risk factors, such as tobacco use and hypertension, and deaths avertable through feasible interventions. We project that Pakistan can reduce NCD deaths by 20% in 2025 by implementing largely cost-neutral interventions.