Lab News
Oct 2023
Race-Aware Approach
Our paper advocates for moving away from race-based medicine and encourages a race-aware approach. We propose steps like diversifying clinical trial populations, broadening precision medicine, and promoting cultural responsiveness in care to address these issues and promote health equity.
April 2022
There is always something to celebrate at the Boussard Lab. Today we celebrate many successes: qualifying examinations, Masters thesis project, birthdays and home ownership. Congratulations to all!
Recent Publications
Trends in Influenza Vaccination Rates among Medicaid Recipients
Our new study examined influenza vaccination rates among Medicaid enrollees in nine U.S. states from 2016 to 2021, revealing that only 15% of enrollees received the vaccine during this period. Notably, vaccination rates increased during peri-COVID seasons. Analysis demonstrated variations in uptake based on age, gender, health conditions, and urban residency, highlighting the need for targeted outreach services to enhance vaccine equity in this population.
Artificial Intelligence–Enabled Analysis of Statin-Related Topics and Sentiments on Social Media
Read about how we use Bayesian logistic LASSO regression (BLLR) to inform clinical utility, Using real-world data from cancer patients receiving chemotherapy, the BLLR models demonstrated similar predictive performance to the standard model while offering the advantage of uncertainty estimation for risk predictions. The study highlights the importance of quantifying uncertainty in clinical settings and reveals differences in predictive uncertainties among patient subgroups.
Censored Fairness through Awareness
This study addresses bias and discrimination in AI, specifically focusing on fairness issues in real-world deployment. Examining individual unfairness amidst censorship, where class labels may not be guaranteed, the study introduces a new fairness guarantee model independent of the Lipschitz condition. This research aims to quantify and mitigate individual unfairness, providing insights applicable to socially sensitive applications, particularly in scenarios where class labels are uncertain or censored.