As principal researcher Anna Zink prepares to start an exciting new chapter as Assistant Professor in the Community Health Department at Tufts University in August, we want to take an opportunity to highlight and celebrate her impactful contributions to the CAAI.
Since joining the team almost 3 years ago, Anna has driven research in machine learning and AI applications within the healthcare sector, focusing largely on promoting health equity and reducing health disparities when implementing care across a diverse population. She consistently considers the unintended consequences associated with the growing adoption of AI in healthcare.
Research Projects
“Ethical is ingrained in everything I do”
Anna has exemplified this through her research contributions at the center, partnering with various healthcare organizations that are committed to responsibly integrating AI into healthcare systems. Her work addresses the risks that AI might reinforce through existing biases and inequalities, if it is not carefully and ethically managed. Anna currently juggles quite a few projects, of which 3 are highlighted below.
The first project is in partnership with the Department of Healthcare Services in California, as they work towards developing a risk tiering algorithm for their Medicaid members, in order to identify members at high risk for adverse social and medical events and underutilization of important health services. Anna worked to address the challenge of building prediction models that include underrepresented populations in medical data, which some colleagues coined “silent missingness”—an issue that stems from incomplete data missing for those patients who don’t frequently engage with healthcare services. The challenge becomes building prediction models to identify the people for whom the data is most likely missing.
Additionally, in collaboration with Providence Health System, Anna works to evaluate the potential benefits and harms of deploying AI-driven triage models in Emergency Departments (ED). Triage refers to the initial assessment process on arrival, where practitioners rapidly determine the severity of illness or injury to prioritize treatment based on urgency. The goal of this project is to assess if and when using AI for triage is necessary while considering the potential ethical implications, benefits, and harms.
Working with Chicago Med, Anna also works to develop governance processes that consider how hospitals should oversee new AI models, evaluate their effectiveness, and establish best practices for their integration. A critical component of this research is adapting these governance frameworks to meet the differing capacities and resources of smaller health organizations, working to ensure that responsible AI integration remains accessible to all.
Over the years, Anna has guided CAAI research considering existing bias in data that is used to build machine learning models in healthcare systems. As she actively works with organizations incorporating the rise of AI adoption into their systems and processes, she helps them consider from an ethical and governance standpoint, what healthcare systems and institutions must prepare for before integrating new models.
Time at the Center
Reflecting on her time at the Center, Anna shared that she will miss her interactions with the center faculty and staff, and the inspiring collaborative environment. "It's not your typical set of research," she explains. "I feel like the research here really pushes boundaries and approaches challenges in interesting and creative ways."
Anna, and her research, have furthered the CAAI’s ongoing commitment to ethically harnessing AI applications while navigating existing biases in data used to build machine learning models, to create meaningful societal impact. We deeply appreciate Anna’s vital contributions to CAAI and look forward to celebrating her ongoing accomplishments at Tufts.
To learn more about CAAI’s research initiatives and our commitment to ethical and equitable AI in healthcare, please see more of our research initiatives here.
