Date of Graduation


Document Type


Degree Type



Eberly College of Arts and Sciences



Committee Chair

Casian Pantea

Committee Member

Brian Hendricks

Committee Member

James Broda


In order to maintain normal operations and prevent unnecessary morbidity and mortality during times of disease outbreak, institutions find a need to conduct frequent and widespread testing of their constituents, often under significantly limited testing resource constraints. Faced with the challenge of how best to allo- cate these limited resources to maximum effect, institutions are increasingly turning to group (or “pooled”) testing, which involves testing strategically-chosen groups of patient samples rather than individual samples, producing significant testing resource savings under certain regimes of disease prevalence. While group test- ing can be conducted without any a priori knowledge of individual disease risk probabilities, incorporating such knowledge—a process called informative group testing—to assign testing groups has the potential to further enhance testing efficiency. Here, we focus on one particular informative group testing procedure which groups samples into two-dimensional arrays for disease status identification (so-called informative ar- ray testing). While others have reported algorithms to optimize the construction of a testing array given its constituent samples, we focus instead on algorithms to assign array groups from a population with heteroge- nous disease risks. We propose two new array assignment strategies—concentrated risk and dispersed risk array assignment—and compare their performance to random array assignment by simulating informative array testing on a heterogenous-risk population. Overall, our results suggest that informative array testing is surprisingly agnostic to array assignment strategy, with one potential exception at high disease prevalence. Furthermore, our consideration of two distinct performance metrics reveals nuance in choosing an optimal informative array testing strategy with regards to both test savings and case identification efficiency.