Over the course of six months, five student teams from the University of Washington School of Public Health, mentored by faculty from the Department of Biostatistics, partnered with regional health organizations to tackle significant real-world problems.
The students, part of the Master of Science in Biostatistics Capstone Program, gained valuable professional experience and were able to provide important data insights to help organizations move toward solutions.
One team worked with the Seattle & King County public housing authorities to analyze why people leave public housing and the impact various reasons have on long-term housing outcomes, specifically, the probability of becoming homeless within a year.
Student insights from the Seattle & King Country project
“Through this project, I’ve learned the importance of having collaboration on a study," said Taylor Keating, a member of team.
"You need experts in the field who can communicate the scientific question and address questions and problems as they arise because problems most definitely will come up.”
Collaborating with Keating, the team was also comprised of Zichen Liu, Niki Petrakos, and Hantong Hu.
Liu agrees with Taylor and said the project gave her additional insight into how biostatisticians and epidemiologists complement and support those who work directly with populations in need, and how the depth of knowledge and context provided by population experts are essential to informing analysis decisions.
Niki Petrakos, another team member added, “I was surprised by how much time and effort was dedicated to the planning stage of the project. I didn't expect I'd need to read so many papers and spend as much time as I did deciding between various statistical methods. This experience has taught me the importance of a detailed project timeline, how to tailor oral presentations for different audiences, and how to work as an efficient and communicative team member,” said Petrakos.
2022 capstone projects
- HUD Hears Study (Sponsor: Public Health Seattle and King County) - The team analyzed the different reasons people leave public housing and the impact these reasons have on long-term housing outcomes, specifically, the probability of becoming homeless within a year.
- Evaluation of Epic Predictive Analytics Models Using UW Medicine HER Data (Sponsor: UW Medicine) - The team worked to understand appointment no-show behavior, evaluated and generated predictions, and assessed how the projections worked across different groups of people.
- Applications of Nowcasting Methods to Notifiable Disease Surveillance (Sponsor: Washington Department of Health) - The team looked to associate COVID-19 hospitalizations with date of disease of transmission. Conducted an analysis to understand the nature of existing data and delays between positive test and hospitalization, develop nowcasting models to mitigate the impact of delays, evaluated models, and worked with WA DOH to establish how the model will be used moving forward.
- Prediction Models of Inpatient Admissions (Sponsor: Community Health Plan of Washington) - The team developed a predictive tool to help identify individuals who are likely to be admitted to the hospital in the near future so that care management teams and health insurance agencies can provide timely support to high-risk plan members and reduce the likelihood of adverse health consequences.
- Medication Use and Pain Score Trajectories After Pediatric Tonsillectomy (Sponsor: Seattle Children’s Hospital) - The team evaluated a phone app used to track post-operative pain and medication use. Analyzed data to assess usage (how many people used it and for how long), characterize pain score trajectories, and explore usage rates and next-day impact of oxycodone.