Predictive Model for Baseline Serum Estradiol Concentration of Female Laboratory Mice;Bias, Accountability, and Public Trust: Navigating the Complexities of Predictive Policing in Modern Law Enforcement

“Male [animal] bias is still present in 8 out of 10 surveyed fields [of study], including neurosciences, physiology, pharmacology, endocrinology…” (Plevkova, J. et al., 2020). My technical research team developed an agent-based model that empowers researchers to estimate hormone levels more efficiently and cost-effectively in female lab mice. Even with the success of this project, what I did not understand was why female animal subjects are neglected from research studies. After diving deeper, I realized that sex bias in research is prevalent and extremely dangerous to the field of medicine. My later STS research focused on exploring the ethical and operational complexities of predictive crime mapping and law enforcement practices, and their impact on perpetuating historical bias.
My technical report focused on my experience as an undergraduate researcher in the University of Virginia Medical Research facilities. Graduate researchers approached me to design and build a model that would allow them to efficiently estimate the estradiol (E2) hormone in female mice. I learned that the graduate students conducting E2 research on reproductive muscle tissue using female mice, and that it was difficult to pinpoint which stage of the estrous (hormone) cycle each lab mouse was in. This meant they had to spend unnecessary amounts of time collecting cells and drawing blood serum to determine the estrous cycle stage, which require expensive testing kits to analyze. My team and I built an algorithm in Python language that receives cell counts and estimates the hormone concentration at any given cycle stage. This model enables endocrine researchers to save time and resources on baseline data, focusing instead on developing therapies for reproductive medicine.
The deliverable of my STS research was a report that synthesizes the current field of predictive algorithms, and the complex relationship between private equity and law enforcement practices. My analysis highlighted the potential benefit for predictive mapping software to empower policing agencies in proactively stop crime. Powerful advancements in the field of predictive AI can solve common resource limitations and increase the efficiency of smaller police agencies. My analysis concluded that historical bias in crime data is detrimental to the success of predictive mapping within law enforcement. Private firms lack incentives to share proprietary technology with law enforcement, highlighting a misalignment between private interests and public safety priorities. The research highlights the potential of predictive policing to enhance resource allocation and crime prevention while emphasizing the need for transparency, accountability, and community engagement to address inherent biases and ensure fair public safety outcomes.
These projects were executed completely independent of each other, and I initially struggled to connect the two. After considering my STS research on predictive policing and my technical project developing an agent-based model to estimate hormone concentration, I realized that these projects shared exploration of bias within data, society, and each individual person. When working towards a successful hormone model, my primary aim was to streamline data collection for endocrine research. However, as the project progressed, I began to recognize the pervasive sex bias in pre-clinical research animals that continued into clinical research. This realization revealed to me that systemic bias towards female animal subjects prioritized male-centric data and fundamentally neglected the unique profiles of female subjects. Similarly, my STS research on predictive policing, grounded in historical data, reinforced existing biases in the criminal justice system, particularly against marginalized communities. Both projects highlight how bias is not just a flaw in datasets, but a reflection of societal structures shaped by individual belief. As Jonathan Haidt notes in The Righteous Mind, “Intuitions come first, strategic reasoning second.” Unconscious biases influence all decision-making, and it is my ethical responsibility as an engineer to bridge these biases with technological innovation.

School of Engineering and Applied Science
Bachelor of Science in Biomedical Engineering
Technical Advisors: Jacob Dunn, Megan Haase
STS Advisor: Kathryn Neeley
Technical Team Members: Mikayla Jackson, Khatiana Perez, Samiyah Syeda, Ramya Tangirala