About:
I’m a data scientist and 3rd year medical student at the University of New Mexico.
I recently left a PhD program in
Biomedical Data Science
with a MS from Stanford University.
As an undergraduate, I completed an interdisciplinary BA in Mathematics,
Statistics, and Physical, Natural & Social Sciences with minors in Navajo
Language & Linguistics and Chicana & Chicano Studies in 2015. I was a
Visiting Student
in the Dept. of Economics at Harvard 2015-2017. In 2018, I was a
fellow
at the Washington University School of Medicine Mallinckrodt Institute of
Radiology.
Research:
Variation in human sex physiology is essential to consider in all biomedical
research, but there are few tractable ways to measure or represent an
individuals underlying sex physiology in data. Most researchers in human health
use assigned sex at birth (ASAB), which is a legal record of the gross
morphology of a neonatal genitalia, and is assigned to every baby regardless of
differences in sexual development (e.g. chromosomal anomalies, ambiguous
genitalia, sex organ agenesis). ASAB also doesn’t reflex the differences
in sex physiology between childhood, adolescence, adulthood, and advanced age.
Moreover, scientists and physicians should expect gene variants, environmental
exposures, acquired disorders, and medical interventions to alter sex
physiology as they do for any biological process.
I study machine learning methods to construct a representation of human sex
physiology that can be used in place of ASAB in clinical settings. My research
evaluates how clinical biomarkers can be used to construct a continuous
representation of human sex physiology to better inform medical decision making.
In particular, I research auto-encoder algorithms in multi-task settings,
probabilistic graph models, variational inference, and meta-learning networks.
Previously, I research Computer Vision methods to isolate and featurize
autofluorescent signals in megapixel resolution, multiplexed immunofluorescence
experiments, multi-class segmentation of intracranial tumors, characterization
of provider-patient emails with classical natural language processing methods,
and disparity in provision of health services and insurance.
Publications: