Research
Each project below has its own page that explains it from the ground up, in the style of my notes: the question, the ideas you need to follow it, and what was found. Nothing past first-year calculus is assumed, and every technical term is defined where it first appears.
CICADAS-II: Dosing Anti-Seizure Drugs in the ICU
S. D. V. Stephens*, T. Westover*, S. F. Zafar, M. B. Westover. “CICADAS-II: Simulation-Based Causal Inference and Adaptive Dosing from Sparse Seizure Counts and Bolus Anti-Seizure Medication in the ICU.” Under major revision at the Journal of Neural Engineering (2026). *Co-first authors.
Manuscript under review. Begun at Beth Israel Deaconess Medical Center and Harvard Medical School; continued at Stanford and Mass General Brigham.
How to tell, from the records an intensive care unit already keeps, whether a way of dosing anti-seizure drugs saves lives, when the sickest patients are the ones who get the most drug.
Monotone Progressions in Rearrangements of the Integers (Erdős Problems 195 and 196)
S. D. V. Stephens. “Monotone Arithmetic Progressions in the Gray Code and in Negabinary Enumerations of the Integers.” Manuscript in preparation (2026).
Work in progress, done for fun. The results below come from a draft that has not been refereed or independently checked.
Erdős asked how long an evenly spaced run of numbers must appear in order inside any rearrangement of the integers. Two simple orderings, one built from binary digits and one from base −2 digits, turn out to avoid runs of five, giving a new proof of the best known bound with a construction that fits on one line.